Authors: Gary Liang, Adam Kabbara, Cindy Liu, Ronaldo Luo, Kina Kim, Michael Guerzhoy
Abstract: We show that human players' gameplay in the game of Wordle is influenced by the semantics, orthography, and morphology of the player's previous guesses. We demonstrate this influence by comparing actual human players' guesses to near-optimal guesses, showing that human players' guesses are biased to be similar to previous guesses semantically, orthographically, and morphologically.
Authors: Miles Q. Li, Benjamin C. M. Fung, Shih-Chia Huang
Abstract: Training large language models is a computationally intensive process that often requires substantial resources to achieve state-of-the-art results. Incremental layer-wise training has been proposed as a potential strategy to optimize the training process by progressively introducing layers, with the expectation that this approach would lead to faster convergence and more efficient use of computational resources. In this paper, we investigate the effectiveness of incremental training for LLMs, dividing the training process into multiple stages where layers are added progressively. Our experimental results indicate that while the incremental approach initially demonstrates some computational efficiency, it ultimately requires greater overall computational costs to reach comparable performance to traditional full-scale training. Although the incremental training process can eventually close the performance gap with the baseline, it does so only after significantly extended continual training. These findings suggest that incremental layer-wise training may not be a viable alternative for training large language models, highlighting its limitations and providing valuable insights into the inefficiencies of this approach.
Authors: Alexander Spangher, Kung-Hsiang Huang, Hyundong Cho, Jonathan May
Abstract: As events progress, news articles often update with new information: if we are not cautious, we risk propagating outdated facts. In this work, we hypothesize that linguistic features indicate factual fluidity, and that we can predict which facts in a news article will update using solely the text of a news article (i.e. not external resources like search engines). We test this hypothesis, first, by isolating fact-updates in large news revisions corpora. News articles may update for many reasons (e.g. factual, stylistic, narrative). We introduce the NewsEdits 2.0 taxonomy, an edit-intentions schema that separates fact updates from stylistic and narrative updates in news writing. We annotate over 9,200 pairs of sentence revisions and train high-scoring ensemble models to apply this schema. Then, taking a large dataset of silver-labeled pairs, we show that we can predict when facts will update in older article drafts with high precision. Finally, to demonstrate the usefulness of these findings, we construct a language model question asking (LLM-QA) abstention task. We wish the LLM to abstain from answering questions when information is likely to become outdated. Using our predictions, we show, LLM absention reaches near oracle levels of accuracy.
Authors: Jianyou Wang, Weili Cao, Longtian Bao, Youze Zheng, Gil Pasternak, Kaicheng Wang, Xiaoyue Wang, Ramamohan Paturi, Leon Bergen
Abstract: Systems that answer questions by reviewing the scientific literature are becoming increasingly feasible. To draw reliable conclusions, these systems should take into account the quality of available evidence, placing more weight on studies that use a valid methodology. We present a benchmark for measuring the methodological strength of biomedical papers, drawing on the risk-of-bias framework used for systematic reviews. The four benchmark tasks, drawn from more than 500 papers, cover the analysis of research study methodology, followed by evaluation of risk of bias in these studies. The benchmark contains 2000 expert-generated bias annotations, and a human-validated pipeline for fine-grained alignment with research paper content. We evaluate a range of large language models on the benchmark, and find that these models fall significantly short of expert-level performance. By providing a standardized tool for measuring judgments of study quality, the benchmark can help to guide systems that perform large-scale aggregation of scientific data. The dataset is available at https://github.com/RoBBR-Benchmark/RoBBR.
Authors: Ananjan Nandi, Christopher D. Manning, Shikhar Murty
Abstract: While compositional accounts of human language understanding are based on a hierarchical tree-like process, neural models like transformers lack a direct inductive bias for such tree structures. Introducing syntactic inductive biases could unlock more robust and data-efficient learning in transformer language models (LMs), but existing methods for incorporating such structure greatly restrict models, either limiting their expressivity or increasing inference complexity. This work instead aims to softly inject syntactic inductive biases into given transformer circuits, through a structured regularizer. We introduce TREEREG, an auxiliary loss function that converts bracketing decisions from silver parses into a set of differentiable orthogonality constraints on vector hidden states. TREEREG integrates seamlessly with the standard LM objective, requiring no architectural changes. LMs pre-trained with TreeReg on natural language corpora such as WikiText-103 achieve up to 10% lower perplexities on out-of-distribution data and up to 9.5 point improvements in syntactic generalization, requiring less than half the training data to outperform standard LMs. TreeReg still provides gains for pre-trained LLMs: Continued pre-training of Sheared Llama with TreeReg results in improved syntactic generalization, and fine-tuning on MultiNLI with TreeReg mitigates degradation of performance on adversarial NLI benchmarks by 41.2 points.
Authors: Junan Li, Yunxiang Li, Yuren Wang, Xixin Wu, Helen Meng
Abstract: Alzheimer's disease (AD) has become one of the most significant health challenges in an aging society. The use of spoken language-based AD detection methods has gained prevalence due to their scalability due to their scalability. Based on the Cookie Theft picture description task, we devised an explainable and effective feature set that leverages the visual capabilities of a large language model (LLM) and the Term Frequency-Inverse Document Frequency (TF-IDF) model. Our experimental results show that the newly proposed features consistently outperform traditional linguistic features across two different classifiers with high dimension efficiency. Our new features can be well explained and interpreted step by step which enhance the interpretability of automatic AD screening.
Authors: Meher Bhardwaj, Hrishikesh Ethari, Dennis Singh Moirangthem
Abstract: The SQL-to-text generation task traditionally uses template base, Seq2Seq, tree-to-sequence, and graph-to-sequence models. Recent models take advantage of pre-trained generative language models for this task in the Seq2Seq framework. However, treating SQL as a sequence of inputs to the pre-trained models is not optimal. In this work, we put forward a new SQL intermediate representation called EzSQL to align SQL with the natural language text sequence. EzSQL simplifies the SQL queries and brings them closer to natural language text by modifying operators and keywords, which can usually be described in natural language. EzSQL also removes the need for set operators. Our proposed SQL-to-text generation model uses EzSQL as the input to a pre-trained generative language model for generating the text descriptions. We demonstrate that our model is an effective state-of-the-art method to generate text narrations from SQL queries on the WikiSQL and Spider datasets. We also show that by generating pretraining data using our SQL-to-text generation model, we can enhance the performance of Text-to-SQL parsers.
Authors: Lui Yoshida
Abstract: This study investigates the impact of example selection on the performance of au-tomated essay scoring (AES) using few-shot prompting with GPT models. We evaluate the effects of the choice and order of examples in few-shot prompting on several versions of GPT-3.5 and GPT-4 models. Our experiments involve 119 prompts with different examples, and we calculate the quadratic weighted kappa (QWK) to measure the agreement between GPT and human rater scores. Regres-sion analysis is used to quantitatively assess biases introduced by example selec-tion. The results show that the impact of example selection on QWK varies across models, with GPT-3.5 being more influenced by examples than GPT-4. We also find evidence of majority label bias, which is a tendency to favor the majority la-bel among the examples, and recency bias, which is a tendency to favor the label of the most recent example, in GPT-generated essay scores and QWK, with these biases being more pronounced in GPT-3.5. Notably, careful example selection enables GPT-3.5 models to outperform some GPT-4 models. However, among the GPT models, the June 2023 version of GPT-4, which is not the latest model, exhibits the highest stability and performance. Our findings provide insights into the importance of example selection in few-shot prompting for AES, especially in GPT-3.5 models, and highlight the need for individual performance evaluations of each model, even for minor versions.
Authors: Rao Fu, Ziyang Luo, Hongzhan Lin, Zhen Ye, Jing Ma
Abstract: Recent advancements in large multimodal models (LMMs) have showcased impressive code generation capabilities, primarily evaluated through image-to-code benchmarks. However, these benchmarks are limited to specific visual programming scenarios where the logic reasoning and the multimodal understanding capacities are split apart. To fill this gap, we propose ScratchEval, a novel benchmark designed to evaluate the visual programming reasoning ability of LMMs. ScratchEval is based on Scratch, a block-based visual programming language widely used in children's programming education. By integrating visual elements and embedded programming logic, ScratchEval requires the model to process both visual information and code structure, thereby comprehensively evaluating its programming intent understanding ability. Our evaluation approach goes beyond the traditional image-to-code mapping and focuses on unified logical thinking and problem-solving abilities, providing a more comprehensive and challenging framework for evaluating the visual programming ability of LMMs. ScratchEval not only fills the gap in existing evaluation methods, but also provides new insights for the future development of LMMs in the field of visual programming. Our benchmark can be accessed at https://github.com/HKBUNLP/ScratchEval .
Authors: Jinghui Liu, Anthony Nguyen
Abstract: Clinical language models are important for many applications in healthcare, but their development depends on access to extensive clinical text for pretraining. However, obtaining clinical notes from electronic health records (EHRs) at scale is challenging due to patient privacy concerns. In this study, we rephrase existing clinical notes using LLMs to generate synthetic pretraining corpora, drawing inspiration from previous work on rephrasing web data. We examine four popular small-sized LLMs (<10B) to create synthetic clinical text to pretrain both decoder-based and encoder-based language models. The method yields better results in language modeling and downstream tasks than previous synthesis approaches without referencing real clinical text. We find that augmenting original clinical notes with synthetic corpora from different LLMs improves performances even at a small token budget, showing the potential of this method to support pretraining at the institutional level or be scaled to synthesize large-scale clinical corpora.
Authors: Mansi Rana, Kadri Hacioglu, Sindhuja Gopalan, Maragathamani Boothalingam
Abstract: Zero-shot slot filling is a well-established subtask of Natural Language Understanding (NLU). However, most existing methods primarily focus on single-turn text data, overlooking the unique complexities of conversational dialogue. Conversational data is highly dynamic, often involving abrupt topic shifts, interruptions, and implicit references that make it difficult to directly apply zero-shot slot filling techniques, even with the remarkable capabilities of large language models (LLMs). This paper addresses these challenges by proposing strategies for automatic data annotation with slot induction and black-box knowledge distillation (KD) from a teacher LLM to a smaller model, outperforming vanilla LLMs on internal datasets by 26% absolute increase in F1 score. Additionally, we introduce an efficient system architecture for call center product settings that surpasses off-the-shelf extractive models by 34% relative F1 score, enabling near real-time inference on dialogue streams with higher accuracy, while preserving low latency.
Authors: Jianjian Li, Shengwei Liang, Yong Liao, Hongping Deng, Haiyang Yu
Abstract: Cross-lingual semantic textual relatedness task is an important research task that addresses challenges in cross-lingual communication and text understanding. It helps establish semantic connections between different languages, crucial for downstream tasks like machine translation, multilingual information retrieval, and cross-lingual text understanding.Based on extensive comparative experiments, we choose the XLM-R-base as our base model and use pre-trained sentence representations based on whitening to reduce anisotropy.Additionally, for the given training data, we design a delicate data filtering method to alleviate the curse of multilingualism. With our approach, we achieve a 2nd score in Spanish, a 3rd in Indonesian, and multiple entries in the top ten results in the competition's track C. We further do a comprehensive analysis to inspire future research aimed at improving performance on cross-lingual tasks.
Authors: Ludovic Tanguy (CLLE), C\'eline Poudat (CLLE), Lydia-Mai Ho-Dac (CLLE)
Abstract: This study proposes a qualitative analysis of self replies in Wikipedia talk pages, more precisely when the first two messages of a discussion are written by the same user. This specific pattern occurs in more than 10% of threads with two messages or more and can be explained by a number of reasons. After a first examination of the lexical specificities of second messages, we propose a seven categories typology and use it to annotate two reference samples (English and French) of 100 threads each. Finally, we analyse and compare the performance of human annotators (who reach a reasonable global efficiency) and instruction-tuned LLMs (which encounter important difficulties with several categories).
Authors: Susmita Das, Arpita Dutta, Kingshuk Roy, Abir Mondal, Arnab Mukhopadhyay
Abstract: The expanding influence of social media platforms over the past decade has impacted the way people communicate. The level of obscurity provided by social media and easy accessibility of the internet has facilitated the spread of hate speech. The terms and expressions related to hate speech gets updated with changing times which poses an obstacle to policy-makers and researchers in case of hate speech identification. With growing number of individuals using their native languages to communicate with each other, hate speech in these low-resource languages are also growing. Although, there is awareness about the English-related approaches, much attention have not been provided to these low-resource languages due to lack of datasets and online available data. This article provides a detailed survey of hate speech detection in low-resource languages around the world with details of available datasets, features utilized and techniques used. This survey further discusses the prevailing surveys, overlapping concepts related to hate speech, research challenges and opportunities.
Authors: Ben Ganon, Alon Zolfi, Omer Hofman, Inderjeet Singh, Hisashi Kojima, Yuval Elovici, Asaf Shabtai
Abstract: In recent years, conversational large language models (LLMs) have shown tremendous success in tasks such as casual conversation, question answering, and personalized dialogue, making significant advancements in domains like virtual assistance, social interaction, and online customer engagement. However, they often generate responses that are not aligned with human values (e.g., ethical standards, safety, or social norms), leading to potentially unsafe or inappropriate outputs. While several techniques have been proposed to address this problem, they come with a cost, requiring computationally expensive training or dramatically increasing the inference time. In this paper, we present DIESEL, a lightweight inference guidance technique that can be seamlessly integrated into any autoregressive LLM to semantically filter undesired concepts from the response. DIESEL can function either as a standalone safeguard or as an additional layer of defense, enhancing response safety by reranking the LLM's proposed tokens based on their similarity to predefined negative concepts in the latent space. This approach provides an efficient and effective solution for maintaining alignment with human values. Our evaluation demonstrates DIESEL's effectiveness on state-of-the-art conversational models (e.g., Llama 3), even in challenging jailbreaking scenarios that test the limits of response safety. We further show that DIESEL can be generalized to use cases other than safety, providing a versatile solution for general-purpose response filtering with minimal computational overhead.
Authors: Yutong Zhang, Lixing Chen, Shenghong Li, Nan Cao, Yang Shi, Jiaxin Ding, Zhe Qu, Pan Zhou, Yang Bai
Abstract: Large language models (LLMs) have demonstrated exceptional performance across a wide variety of domains. Nonetheless, generalist LLMs continue to fall short in reasoning tasks necessitating specialized knowledge. Prior investigations into specialized LLMs focused on domain-specific training, which entails substantial efforts in domain data acquisition and model parameter fine-tuning. To address these challenges, this paper proposes the Way-to-Specialist (WTS) framework, which synergizes retrieval-augmented generation with knowledge graphs (KGs) to enhance the specialized capability of LLMs in the absence of specialized training. In distinction to existing paradigms that merely utilize external knowledge from general KGs or static domain KGs to prompt LLM for enhanced domain-specific reasoning, WTS proposes an innovative "LLM$\circlearrowright$KG" paradigm, which achieves bidirectional enhancement between specialized LLM and domain knowledge graph (DKG). The proposed paradigm encompasses two closely coupled components: the DKG-Augmented LLM and the LLM-Assisted DKG Evolution. The former retrieves question-relevant domain knowledge from DKG and uses it to prompt LLM to enhance the reasoning capability for domain-specific tasks; the latter leverages LLM to generate new domain knowledge from processed tasks and use it to evolve DKG. WTS closes the loop between DKG-Augmented LLM and LLM-Assisted DKG Evolution, enabling continuous improvement in the domain specialization as it progressively answers and learns from domain-specific questions. We validate the performance of WTS on 6 datasets spanning 5 domains. The experimental results show that WTS surpasses the previous SOTA in 4 specialized domains and achieves a maximum performance improvement of 11.3%.
Authors: Sanjay Suryanarayanan, Haiyue Song, Mohammed Safi Ur Rahman Khan, Anoop Kunchukuttan, Mitesh M. Khapra, Raj Dabre
Abstract: Mining parallel document pairs poses a significant challenge because existing sentence embedding models often have limited context windows, preventing them from effectively capturing document-level information. Another overlooked issue is the lack of concrete evaluation benchmarks comprising high-quality parallel document pairs for assessing document-level mining approaches, particularly for Indic languages. In this study, we introduce Pralekha, a large-scale benchmark for document-level alignment evaluation. Pralekha includes over 2 million documents, with a 1:2 ratio of unaligned to aligned pairs, covering 11 Indic languages and English. Using Pralekha, we evaluate various document-level mining approaches across three dimensions: the embedding models, the granularity levels, and the alignment algorithm. To address the challenge of aligning documents using sentence and chunk-level alignments, we propose a novel scoring method, Document Alignment Coefficient (DAC). DAC demonstrates substantial improvements over baseline pooling approaches, particularly in noisy scenarios, achieving average gains of 20-30% in precision and 15-20% in F1 score. These results highlight DAC's effectiveness in parallel document mining for Indic languages.
Authors: Eduard Manziuk, Oleksander Barmak, Pavlo Radiuk, Vladislav Kuznetsov, Iurii Krak, Sergiy Yakovlev
Abstract: This paper proposes a novel approach to semantic ontology alignment using contextual descriptors. A formalization was developed that enables the integration of essential and contextual descriptors to create a comprehensive knowledge model. The hierarchical structure of the semantic approach and the mathematical apparatus for analyzing potential conflicts between concepts, particularly in the example of "Transparency" and "Privacy" in the context of artificial intelligence, are demonstrated. Experimental studies showed a significant improvement in ontology alignment metrics after the implementation of contextual descriptors, especially in the areas of privacy, responsibility, and freedom & autonomy. The application of contextual descriptors achieved an average overall improvement of approximately 4.36%. The results indicate the effectiveness of the proposed approach for more accurately reflecting the complexity of knowledge and its contextual dependence.
Authors: Anirudh Phukan, Divyansh, Harshit Kumar Morj, Vaishnavi, Apoorv Saxena, Koustava Goswami
Abstract: The rapid development of Large Multimodal Models (LMMs) has significantly advanced multimodal understanding by harnessing the language abilities of Large Language Models (LLMs) and integrating modality-specific encoders. However, LMMs are plagued by hallucinations that limit their reliability and adoption. While traditional methods to detect and mitigate these hallucinations often involve costly training or rely heavily on external models, recent approaches utilizing internal model features present a promising alternative. In this paper, we critically assess the limitations of the state-of-the-art training-free technique, the logit lens, in handling generalized visual hallucinations. We introduce a refined method that leverages contextual token embeddings from middle layers of LMMs. This approach significantly improves hallucination detection and grounding across diverse categories, including actions and OCR, while also excelling in tasks requiring contextual understanding, such as spatial relations and attribute comparison. Our novel grounding technique yields highly precise bounding boxes, facilitating a transition from Zero-Shot Object Segmentation to Grounded Visual Question Answering. Our contributions pave the way for more reliable and interpretable multimodal models.
Authors: Joy Mahapatra, Utpal Garain
Abstract: Large Language Models (LLMs) have shown exceptional performance across various Data-to-Text Generation (DTG) tasks. However, generating factually consistent text in DTG remains challenging for LLMs. Despite this, in-depth evaluations of LLM factual consistency for DTG remain missing in the current literature. This paper addresses this gap by providing an extensive evaluation of factual consistency in LLMs for DTG. Our evaluation covers five widely used DTG datasets (E2E, ViGGo, WikiTableText, DART, and WebNLG) and five prominent LLM families (T5, BART, OPT, BLOOM, and Llama 2). To ensure a thorough evaluation of factual consistency, we use four state-of-the-art automatic metrics and include essential human assessments. Our extensive evaluations reveals three key findings regarding factual consistency in LLMs for DTG. First, Llama 2 often excels in generating factually consistent text, although smaller models like T5 and BART can achieve strong factual consistency on larger, lexically less-diverse datasets. Second, the average rate of change (AROC) indicates that increasing model size (number of model trainable parameters) generally enhances factual consistency of LLMs in DTG. Third, we observe that source-reference divergence (i.e., when the reference text diverges semantically from the source) typically reduces the factual consistency of LLMs in DTG.
Authors: Marion Thaler, Abdullatif K\"oksal, Alina Leidinger, Anna Korhonen, Hinrich Sch\"utze
Abstract: As LLMs are increasingly integrated into user-facing applications, addressing biases that perpetuate societal inequalities is crucial. While much work has gone into measuring or mitigating biases in these models, fewer studies have investigated their origins. Therefore, this study examines the correlation between gender-occupation bias in pre-training data and their manifestation in LLMs, focusing on the Dolma dataset and the OLMo model. Using zero-shot prompting and token co-occurrence analyses, we explore how biases in training data influence model outputs. Our findings reveal that biases present in pre-training data are amplified in model outputs. The study also examines the effects of prompt types, hyperparameters, and instruction-tuning on bias expression, finding instruction-tuning partially alleviating representational bias while still maintaining overall stereotypical gender associations, whereas hyperparameters and prompting variation have a lesser effect on bias expression. Our research traces bias throughout the LLM development pipeline and underscores the importance of mitigating bias at the pretraining stage.
Authors: Jinu Nyachhyon, Mridul Sharma, Prajwal Thapa, Bal Krishna Bal
Abstract: The Nepali language has distinct linguistic features, especially its complex script (Devanagari script), morphology, and various dialects, which pose a unique challenge for natural language processing (NLP) evaluation. While the Nepali Language Understanding Evaluation (Nep-gLUE) benchmark provides a foundation for evaluating models, it remains limited in scope, covering four tasks. This restricts their utility for comprehensive assessments of NLP models. To address this limitation, we introduce eight new datasets, creating a new benchmark, the Nepali Language Understanding Evaluation (NLUE) benchmark, which covers a total of 12 tasks for evaluating the performance of models across a diverse set of Natural Language Understanding (NLU) tasks. The added tasks include single-sentence classification, similarity and paraphrase tasks, and Natural Language Inference (NLI) tasks. On evaluating the models using added tasks, we observe that the existing models fall short in handling complex NLU tasks effectively. This expanded benchmark sets a new standard for evaluating, comparing, and advancing models, contributing significantly to the broader goal of advancing NLP research for low-resource languages.
Authors: Cl\'emence Sebe, Sarah Cohen-Boulakia, Olivier Ferret, Aur\'elie N\'ev\'eol
Abstract: Bioinformatics workflows are essential for complex biological data analyses and are often described in scientific articles with source code in public repositories. Extracting detailed workflow information from articles can improve accessibility and reusability but is hindered by limited annotated corpora. To address this, we framed the problem as a low-resource extraction task and tested four strategies: 1) creating a tailored annotated corpus, 2) few-shot named-entity recognition (NER) with an autoregressive language model, 3) NER using masked language models with existing and new corpora, and 4) integrating workflow knowledge into NER models. Using BioToFlow, a new corpus of 52 articles annotated with 16 entities, a SciBERT-based NER model achieved a 70.4 F-measure, comparable to inter-annotator agreement. While knowledge integration improved performance for specific entities, it was less effective across the entire information schema. Our results demonstrate that high-performance information extraction for bioinformatics workflows is achievable.
Authors: Hui Dai, Dan Pechi, Xinyi Yang, Garvit Banga, Raghav Mantri
Abstract: The Needle-in-a-haystack (NIAH) test is a general task used to assess language models' (LMs') abilities to recall particular information from long input context. This framework however does not provide a means of analyzing what factors, beyond context length, contribute to LMs' abilities or inabilities to separate and recall needles from their haystacks. To provide a systematic means of assessing what features contribute to LMs' NIAH capabilities, we developed a synthetic benchmark called DENIAHL (Data-oriented Evaluation of NIAH for LLM's). Our work expands on previous NIAH studies by ablating NIAH features beyond typical context length including data type, size, and patterns. We find stark differences between GPT-3.5 and LLaMA 2-7B's performance on DENIAHL, and drops in recall performance when features like item size are increased, and to some degree when data type is changed from numbers to letters. This has implications for increasingly large context models, demonstrating factors beyond item-number impact NIAH capabilities.
Authors: Tian Yu, Shaolei Zhang, Yang Feng
Abstract: Iterative retrieval refers to the process in which the model continuously queries the retriever during generation to enhance the relevance of the retrieved knowledge, thereby improving the performance of Retrieval-Augmented Generation (RAG). Existing work typically employs few-shot prompting or manually constructed rules to implement iterative retrieval. This introduces additional inference overhead and overlooks the remarkable reasoning capabilities of Large Language Models (LLMs). In this paper, we introduce Auto-RAG, an autonomous iterative retrieval model centered on the LLM's powerful decision-making capabilities. Auto-RAG engages in multi-turn dialogues with the retriever, systematically planning retrievals and refining queries to acquire valuable knowledge. This process continues until sufficient external information is gathered, at which point the results are presented to the user. To this end, we develop a method for autonomously synthesizing reasoning-based decision-making instructions in iterative retrieval and fine-tuned the latest open-source LLMs. The experimental results indicate that Auto-RAG is capable of autonomous iterative interaction with the retriever, effectively leveraging the remarkable reasoning and decision-making abilities of LLMs, which lead to outstanding performance across six benchmarks. Further analysis reveals that Auto-RAG can autonomously adjust the number of iterations based on the difficulty of the questions and the utility of the retrieved knowledge, without requiring any human intervention. Moreover, Auto-RAG expresses the iterative retrieval process in natural language, enhancing interpretability while providing users with a more intuitive experience\footnote{Code is available at \url{https://github.com/ictnlp/Auto-RAG}.
Authors: Yujin Han, Lei Xu, Sirui Chen, Difan Zou, Chaochao Lu
Abstract: Large language models (LLMs) have shown remarkable capability in natural language tasks, yet debate persists on whether they truly comprehend deep structure (i.e., core semantics) or merely rely on surface structure (e.g., presentation format). Prior studies observe that LLMs' performance declines when intervening on surface structure, arguing their success relies on surface structure recognition. However, surface structure sensitivity does not prevent deep structure comprehension. Rigorously evaluating LLMs' capability requires analyzing both, yet deep structure is often overlooked. To this end, we assess LLMs' comprehension ability using causal mediation analysis, aiming to fully discover the capability of using both deep and surface structures. Specifically, we formulate the comprehension of deep structure as direct causal effect (DCE) and that of surface structure as indirect causal effect (ICE), respectively. To address the non-estimability of original DCE and ICE -- stemming from the infeasibility of isolating mutual influences of deep and surface structures, we develop the corresponding quantifiable surrogates, including approximated DCE (ADCE) and approximated ICE (AICE). We further apply the ADCE to evaluate a series of mainstream LLMs, showing that most of them exhibit deep structure comprehension ability, which grows along with the prediction accuracy. Comparing ADCE and AICE demonstrates closed-source LLMs rely more on deep structure, while open-source LLMs are more surface-sensitive, which decreases with model scale. Theoretically, ADCE is a bidirectional evaluation, which measures both the sufficiency and necessity of deep structure changes in causing output variations, thus offering a more comprehensive assessment than accuracy, a common evaluation in LLMs. Our work provides new insights into LLMs' deep structure comprehension and offers novel methods for LLMs evaluation.
Authors: Yanxi Chen, Xuchen Pan, Yaliang Li, Bolin Ding, Jingren Zhou
Abstract: We propose a general two-stage algorithm that enjoys a provable scaling law for the test-time compute of large language models (LLMs). Given an input problem, the proposed algorithm first generates $N$ candidate solutions, and then chooses the best one via a multiple-round knockout tournament where each pair of candidates are compared for $K$ times and only the winners move on to the next round. In a minimalistic implementation, both stages can be executed with a black-box LLM alone and nothing else (e.g., no external verifier or reward model), and a total of $N \times (K + 1)$ highly parallelizable LLM calls are needed for solving an input problem. Assuming that a generated candidate solution is correct with probability $p_{\text{gen}} > 0$ and a comparison between a pair of correct and incorrect solutions identifies the right winner with probability $p_{\text{comp}} > 0.5$ (i.e., better than a random guess), we prove theoretically that the failure probability of the proposed algorithm decays to zero exponentially with respect to $N$ and $K$: $$\mathbb{P}(\text{final output is incorrect}) \le (1 - p_{\text{gen}})^N + \lceil \log_2 N \rceil e^{-2 K (p_{\text{comp}} - 0.5)^2}.$$ Our empirical results with the challenging MMLU-Pro benchmark validate the technical assumptions, as well as the efficacy of the proposed algorithm and the gains from scaling up its test-time compute.
Authors: Abhinav Joshi, Areeb Ahmad, Ashutosh Modi
Abstract: Large Language Models (LLMs) have shown state-of-the-art performance in a variety of tasks, including arithmetic and reasoning; however, to gauge the intellectual capabilities of LLMs, causal reasoning has become a reliable proxy for validating a general understanding of the mechanics and intricacies of the world similar to humans. Previous works in natural language processing (NLP) have either focused on open-ended causal reasoning via causal commonsense reasoning (CCR) or framed a symbolic representation-based question answering for theoretically backed-up analysis via a causal inference engine. The former adds an advantage of real-world grounding but lacks theoretically backed-up analysis/validation, whereas the latter is far from real-world grounding. In this work, we bridge this gap by proposing the COLD (Causal reasOning in cLosed Daily activities) framework, which is built upon human understanding of daily real-world activities to reason about the causal nature of events. We show that the proposed framework facilitates the creation of enormous causal queries (~ 9 million) and comes close to the mini-turing test, simulating causal reasoning to evaluate the understanding of a daily real-world task. We evaluate multiple LLMs on the created causal queries and find that causal reasoning is challenging even for activities trivial to humans. We further explore (the causal reasoning abilities of LLMs) using the backdoor criterion to determine the causal strength between events.
Authors: Dihong Gong, Pu Lu, Zelong Wang, Meng Zhou, Xiuqiang He
Abstract: Large Language Models (LLMs) offer a promising basis for creating agents that can tackle complex tasks through iterative environmental interaction. Existing methods either require these agents to mimic expert-provided trajectories or rely on definitive environmental feedback for reinforcement learning which limits their application to specific scenarios like gaming or code generation. This paper introduces a novel training method for LLM-based agents using weakly supervised signals from a critic LLM, bypassing the need for expert trajectories or definitive feedback. Our agents are trained in iterative manner, where they initially generate trajectories through environmental interaction. Subsequently, a critic LLM selects a subset of good trajectories, which are then used to update the agents, enabling them to generate improved trajectories in the next iteration. Extensive tests on the API-bank dataset show consistent improvement in our agents' capabilities and comparable performance to GPT-4, despite using open-source models with much fewer parameters.
Authors: Kaustubh Ponkshe, Raghav Singhal, Eduard Gorbunov, Alexey Tumanov, Samuel Horvath, Praneeth Vepakomma
Abstract: Low-rank adapters have become a standard approach for efficiently fine-tuning large language models (LLMs), but they often fall short of achieving the performance of full fine-tuning. We propose a method, LoRA Silver Bullet or LoRA-SB, that approximates full fine-tuning within low-rank subspaces using a carefully designed initialization strategy. We theoretically demonstrate that the architecture of LoRA-XS, which inserts a trainable (r x r) matrix between B and A while keeping other matrices fixed, provides the precise conditions needed for this approximation. We leverage its constrained update space to achieve optimal scaling for high-rank gradient updates while removing the need for hyperparameter tuning. We prove that our initialization offers an optimal low-rank approximation of the initial gradient and preserves update directions throughout training. Extensive experiments across mathematical reasoning, commonsense reasoning, and language understanding tasks demonstrate that our approach exceeds the performance of standard LoRA while using 27-90x fewer parameters, and comprehensively outperforms LoRA-XS. Our findings establish that it is possible to simulate full fine-tuning in low-rank subspaces, and achieve significant efficiency gains without sacrificing performance. Our code is publicly available at https://github.com/RaghavSinghal10/lora-sb.
Authors: Georg Niess, Roman Kern
Abstract: The rapid advancement of large language models (LLMs) has made it increasingly difficult to distinguish between text written by humans and machines. While watermarks already exist for LLMs, they often lack flexibility, and struggle with attacks such as paraphrasing. To address these issues, we propose a multi-feature method for generating watermarks that combines multiple distinct watermark features into an ensemble watermark. Concretely, we combine acrostica and sensorimotor norms with the established red-green watermark to achieve a 98% detection rate. After a paraphrasing attack the performance remains high with 95% detection rate. The red-green feature alone as baseline achieves a detection rate of 49%. The evaluation of all feature combinations reveals that the ensemble of all three consistently has the highest detection rate across several LLMs and watermark strength settings. Due to the flexibility of combining features in the ensemble, various requirements and trade-offs can be addressed. Additionally, for all ensemble configurations the same detection function can be used without adaptations. This method is particularly of interest to facilitate accountability and prevent societal harm.
Authors: Mingyu Xu, Wei Cheng, Bingning Wang, Weipeng Chen
Abstract: The current large language models are mainly based on decode-only structure transformers, which have great in-context learning (ICL) capabilities. It is generally believed that the important foundation of its ICL capability is the induction heads mechanism, which requires at least two layers attention. In order to more efficiently implement the ability of the model's induction, we revisit the induction heads mechanism and proposed a KV shifting attention. We theoretically prove that the KV shifting attention reducing the model's requirements for the depth and width of the induction heads mechanism. Our experimental results demonstrate that KV shifting attention is beneficial to learning induction heads and language modeling, which lead to better performance or faster convergence from toy models to the pre-training models with more than 10 B parameters.
Authors: Soham Poddar, Biswajit Paul, Moumita Basu, Saptarshi Ghosh
Abstract: A lot of claims are made in social media posts, which may contain misinformation or fake news. Hence, it is crucial to identify claims as a first step towards claim verification. Given the huge number of social media posts, the task of identifying claims needs to be automated. This competition deals with the task of 'Claim Span Identification' in which, given a text, parts / spans that correspond to claims are to be identified. This task is more challenging than the traditional binary classification of text into claim or not-claim, and requires state-of-the-art methods in Pattern Recognition, Natural Language Processing and Machine Learning. For this competition, we used a newly developed dataset called HECSI containing about 8K posts in English and about 8K posts in Hindi with claim-spans marked by human annotators. This paper gives an overview of the competition, and the solutions developed by the participating teams.
Authors: Junyong Kang, Donghyun Son, Hwanjun Song, Buru Chang
Abstract: In-context learning refers to the emerging ability of large language models (LLMs) to perform a target task without additional training, utilizing demonstrations of the task. Recent studies aim to enhance in-context learning performance by selecting more useful demonstrations. However, they overlook the presence of inevitable noisy labels in task demonstrations that arise during the labeling process in the real-world. In this paper, we propose a new task, in-context learning with noisy labels, which aims to solve real-world problems for in-context learning where labels in task demonstrations would be corrupted. Moreover, we propose a new method and baseline methods for the new task, inspired by studies in learning with noisy labels. Through experiments, we demonstrate that our proposed method can serve as a safeguard against performance degradation in in-context learning caused by noisy labels.
Authors: Anthony G Cohn, Robert E Blackwell
Abstract: Qualitative Spatial Reasoning is a well explored area of Knowledge Representation and Reasoning and has multiple applications ranging from Geographical Information Systems to Robotics and Computer Vision. Recently, many claims have been made for the reasoning capabilities of Large Language Models (LLMs). Here, we investigate the extent to which a set of representative LLMs can perform classical qualitative spatial reasoning tasks on the mereotopological Region Connection Calculus, RCC-8. We conduct three pairs of experiments (reconstruction of composition tables, alignment to human composition preferences, conceptual neighbourhood reconstruction) using state-of-the-art LLMs; in each pair one experiment uses eponymous relations and one, anonymous relations (to test the extent to which the LLM relies on knowledge about the relation names obtained during training). All instances are repeated 30 times to measure the stochasticity of the LLMs.
Authors: Taja Kuzman, Nikola Ljube\v{s}i\'c
Abstract: With the ever-increasing number of news stories available online, classifying them by topic, regardless of the language they are written in, has become crucial for enhancing readers' access to relevant content. To address this challenge, we propose a teacher-student framework based on large language models (LLMs) for developing multilingual news classification models of reasonable size with no need for manual data annotation. The framework employs a Generative Pretrained Transformer (GPT) model as the teacher model to develop an IPTC Media Topic training dataset through automatic annotation of news articles in Slovenian, Croatian, Greek, and Catalan. The teacher model exhibits a high zero-shot performance on all four languages. Its agreement with human annotators is comparable to that between the human annotators themselves. To mitigate the computational limitations associated with the requirement of processing millions of texts daily, smaller BERT-like student models are fine-tuned on the GPT-annotated dataset. These student models achieve high performance comparable to the teacher model. Furthermore, we explore the impact of the training data size on the performance of the student models and investigate their monolingual, multilingual and zero-shot cross-lingual capabilities. The findings indicate that student models can achieve high performance with a relatively small number of training instances, and demonstrate strong zero-shot cross-lingual abilities. Finally, we publish the best-performing news topic classifier, enabling multilingual classification with the top-level categories of the IPTC Media Topic schema.
Authors: Alessandro Scir\`e, Andrei Stefan Bejgu, Simone Tedeschi, Karim Ghonim, Federico Martelli, Roberto Navigli
Abstract: After the introduction of Large Language Models (LLMs), there have been substantial improvements in the performance of Natural Language Generation (NLG) tasks, including Text Summarization and Machine Translation. However, LLMs still produce outputs containing hallucinations, that is, content not grounded in factual information. Therefore, developing methods to assess the factuality of LLMs has become urgent. Indeed, resources for factuality evaluation have recently emerged. Although challenging, these resources face one or more of the following limitations: (i) they are tailored to a specific task or domain; (ii) they are limited in size, thereby preventing the training of new factuality evaluators; (iii) they are designed for simpler verification tasks, such as claim verification. To address these issues, we introduce LLM-Oasis, to the best of our knowledge the largest resource for training end-to-end factuality evaluators. LLM-Oasis is constructed by extracting claims from Wikipedia, falsifying a subset of these claims, and generating pairs of factual and unfactual texts. We then rely on human annotators to both validate the quality of our dataset and to create a gold standard test set for benchmarking factuality evaluation systems. Our experiments demonstrate that LLM-Oasis presents a significant challenge for state-of-the-art LLMs, with GPT-4o achieving up to 60% accuracy in our proposed end-to-end factuality evaluation task, highlighting its potential to drive future research in the field.
Authors: Wanyue Zhang, Ziyong Li, Wen Yang, Chunlin Leng, Yinan Bai, Qianlong Du, Chengqing Zong, Jiajun Zhang
Abstract: During the development of large language models (LLMs), pre-training data play a critical role in shaping LLMs' capabilities. In recent years several large-scale and high-quality pre-training datasets have been released to accelerate the research of LLMs, including ChineseWebText1.0, C4, Pile, WanJuan, MAPCC and others. However, as LLMs continue to evolve, focus has increasingly shifted to domain-specific capabilities and safety concerns, making those previous coarse-grained texts insufficient for meeting training requirements. Furthermore, fine-grained information, such as quality, domain and toxicity, is becoming increasingly important in building powerful and reliable LLMs for various scenarios. To address these challenges, in this paper we propose a new tool-chain called MDFG-tool for constructing large-scale and high-quality Chinese datasets with multi-dimensional and fine-grained information. First, we employ manually crafted rules to discard explicit noisy texts from raw contents. Second, the quality evaluation model, domain classifier, and toxicity evaluation model are well-designed to assess the remaining cleaned data respectively. Finally, we integrate these three types of fine-grained information for each text. With this approach, we release the largest, high-quality and fine-grained Chinese text ChineseWebText2.0, which consists of 3.8TB and each text is associated with a quality score, domain labels, a toxicity label and a toxicity score, facilitating the LLM researchers to select data based on various types of fine-grained information. The data, codes and the tool-chain are available on this website https://github.com/CASIA-LM/ChineseWebText-2.0
Authors: John Francis, Saba Esnaashari, Anton Poletaev, Sukankana Chakraborty, Youmna Hashem, Jonathan Bright
Abstract: Large language models (LLMs) have demonstrated remarkable capabilities in text analysis tasks, yet their evaluation on complex, real-world applications remains challenging. We define a set of tasks, Multi-Insight Multi-Document Extraction (MIMDE) tasks, which involves extracting an optimal set of insights from a document corpus and mapping these insights back to their source documents. This task is fundamental to many practical applications, from analyzing survey responses to processing medical records, where identifying and tracing key insights across documents is crucial. We develop an evaluation framework for MIMDE and introduce a novel set of complementary human and synthetic datasets to examine the potential of synthetic data for LLM evaluation. After establishing optimal metrics for comparing extracted insights, we benchmark 20 state-of-the-art LLMs on both datasets. Our analysis reveals a strong correlation (0.71) between the ability of LLMs to extracts insights on our two datasets but synthetic data fails to capture the complexity of document-level analysis. These findings offer crucial guidance for the use of synthetic data in evaluating text analysis systems, highlighting both its potential and limitations.
Authors: David Duki\'c, Marin Petri\v{c}evi\'c, Sven \'Curkovi\'c, Jan \v{S}najder
Abstract: TakeLab Retriever is an AI-driven search engine designed to discover, collect, and semantically analyze news articles from Croatian news outlets. It offers a unique perspective on the history and current landscape of Croatian online news media, making it an essential tool for researchers seeking to uncover trends, patterns, and correlations that general-purpose search engines cannot provide. TakeLab retriever utilizes cutting-edge natural language processing (NLP) methods, enabling users to sift through articles using named entities, phrases, and topics through the web application. This technical report is divided into two parts: the first explains how TakeLab Retriever is utilized, while the second provides a detailed account of its design. In the second part, we also address the software engineering challenges involved and propose solutions for developing a microservice-based semantic search engine capable of handling over ten million news articles published over the past two decades.
Authors: Syed Mohammed Mostaque Billah, Ateya Ahmed Subarna, Sudipta Nandi Sarna, Ahmad Shawkat Wasit, Anika Fariha, Asif Sushmit, Arig Yousuf Sadeque
Abstract: Around seven million individuals in India, Bangladesh, Bhutan, and Nepal speak Santali, positioning it as nearly the third most commonly used Austroasiatic language. Despite its prominence among the Austroasiatic language family's Munda subfamily, Santali lacks global recognition. Currently, no translation models exist for the Santali language. Our paper aims to include Santali to the NPL spectrum. We aim to examine the feasibility of building Santali translation models based on available Santali corpora. The paper successfully addressed the low-resource problem and, with promising results, examined the possibility of creating a functional Santali machine translation model in a low-resource setup. Our study shows that Santali-English parallel corpus performs better when in transformers like mt5 as opposed to untrained transformers, proving that transfer learning can be a viable technique that works with Santali language. Besides the mT5 transformer, Santali-English performs better than Santali-Bangla parallel corpus as the mT5 has been trained in way more English data than Bangla data. Lastly, our study shows that with data augmentation, our model performs better.
Authors: Reyhaneh Hashempour, Barbara Plank, Aline Villavicencio, Renato Cordeiro de Amorim
Abstract: This work presents a set of experiments conducted to predict the gender of Twitter users based on language-independent features extracted from the text of the users' tweets. The experiments were performed on a version of TwiSty dataset including tweets written by the users of six different languages: Portuguese, French, Dutch, English, German, and Italian. Logistic regression (LR), and feed-forward neural networks (FFNN) with back-propagation were used to build models in two different settings: Inter-Lingual (IL) and Cross-Lingual (CL). In the IL setting, the training and testing were performed on the same language whereas in the CL, Italian and German datasets were set aside and only used as test sets and the rest were combined to compose training and development sets. In the IL, the highest accuracy score belongs to LR whereas in the CL, FFNN with three hidden layers yields the highest score. The results show that neural network based models underperform traditional models when the size of the training set is small; however, they beat traditional models by a non-trivial margin, when they are fed with large enough data. Finally, the feature analysis confirms that men and women have different writing styles independent of their language.
Authors: Angelika Romanou, Negar Foroutan, Anna Sotnikova, Zeming Chen, Sree Harsha Nelaturu, Shivalika Singh, Rishabh Maheshwary, Micol Altomare, Mohamed A. Haggag, Snegha A, Alfonso Amayuelas, Azril Hafizi Amirudin, Viraat Aryabumi, Danylo Boiko, Michael Chang, Jenny Chim, Gal Cohen, Aditya Kumar Dalmia, Abraham Diress, Sharad Duwal, Daniil Dzenhaliou, Daniel Fernando Erazo Florez, Fabian Farestam, Joseph Marvin Imperial, Shayekh Bin Islam, Perttu Isotalo, Maral Jabbarishiviari, B\"orje F. Karlsson, Eldar Khalilov, Christopher Klamm, Fajri Koto, Dominik Krzemi\'nski, Gabriel Adriano de Melo, Syrielle Montariol, Yiyang Nan, Joel Niklaus, Jekaterina Novikova, Johan Samir Obando Ceron, Debjit Paul, Esther Ploeger, Jebish Purbey, Swati Rajwal, Selvan Sunitha Ravi, Sara Rydell, Roshan Santhosh, Drishti Sharma, Marjana Prifti Skenduli, Arshia Soltani Moakhar, Bardia Soltani Moakhar, Ran Tamir, Ayush Kumar Tarun, Azmine Toushik Wasi, Thenuka Ovin Weerasinghe, Serhan Yilmaz, Mike Zhang, Imanol Schlag, Marzieh Fadaee, Sara Hooker, Antoine Bosselut
Abstract: The performance differential of large language models (LLM) between languages hinders their effective deployment in many regions, inhibiting the potential economic and societal value of generative AI tools in many communities. However, the development of functional LLMs in many languages (\ie, multilingual LLMs) is bottlenecked by the lack of high-quality evaluation resources in languages other than English. Moreover, current practices in multilingual benchmark construction often translate English resources, ignoring the regional and cultural knowledge of the environments in which multilingual systems would be used. In this work, we construct an evaluation suite of 197,243 QA pairs from local exam sources to measure the capabilities of multilingual LLMs in a variety of regional contexts. Our novel resource, INCLUDE, is a comprehensive knowledge- and reasoning-centric benchmark across 44 written languages that evaluates multilingual LLMs for performance in the actual language environments where they would be deployed.
Authors: Fangze Fu, Wei Ai, Fan Yang, Yuntao Shou, Tao Meng, Keqin Li
Abstract: Multimodal Emotion Recognition in Conversations (MERC) aims to classify utterance emotions using textual, auditory, and visual modal features. Most existing MERC methods assume each utterance has complete modalities, overlooking the common issue of incomplete modalities in real-world scenarios. Recently, graph neural networks (GNNs) have achieved notable results in Incomplete Multimodal Emotion Recognition in Conversations (IMERC). However, traditional GNNs focus on binary relationships between nodes, limiting their ability to capture more complex, higher-order information. Moreover, repeated message passing can cause over-smoothing, reducing their capacity to preserve essential high-frequency details. To address these issues, we propose a Spectral Domain Reconstruction Graph Neural Network (SDR-GNN) for incomplete multimodal learning in conversational emotion recognition. SDR-GNN constructs an utterance semantic interaction graph using a sliding window based on both speaker and context relationships to model emotional dependencies. To capture higher-order and high-frequency information, SDR-GNN utilizes weighted relationship aggregation, ensuring consistent semantic feature extraction across utterances. Additionally, it performs multi-frequency aggregation in the spectral domain, enabling efficient recovery of incomplete modalities by extracting both high- and low-frequency information. Finally, multi-head attention is applied to fuse and optimize features for emotion recognition. Extensive experiments on various real-world datasets demonstrate that our approach is effective in incomplete multimodal learning and outperforms current state-of-the-art methods.
Authors: Dimosthenis Antypas, Indira Sen, Carla Perez-Almendros, Jose Camacho-Collados, Francesco Barbieri
Abstract: The detection of sensitive content in large datasets is crucial for ensuring that shared and analysed data is free from harmful material. However, current moderation tools, such as external APIs, suffer from limitations in customisation, accuracy across diverse sensitive categories, and privacy concerns. Additionally, existing datasets and open-source models focus predominantly on toxic language, leaving gaps in detecting other sensitive categories such as substance abuse or self-harm. In this paper, we put forward a unified dataset tailored for social media content moderation across six sensitive categories: conflictual language, profanity, sexually explicit material, drug-related content, self-harm, and spam. By collecting and annotating data with consistent retrieval strategies and guidelines, we address the shortcomings of previous focalised research. Our analysis demonstrates that fine-tuning large language models (LLMs) on this novel dataset yields significant improvements in detection performance compared to open off-the-shelf models such as LLaMA, and even proprietary OpenAI models, which underperform by 10-15% overall. This limitation is even more pronounced on popular moderation APIs, which cannot be easily tailored to specific sensitive content categories, among others.
Authors: Mohammed Q. Shormani
Abstract: This study provides a comprehensive analysis of artificial intelligence (AI) contribution to translation industry (ACTI) research, synthesizing it over forty-one years from 1980-2024. 13220 articles were retrieved from three sources, namely WoS, Scopus, and Lens. We provided two types of analysis, viz., scientometric and thematic, focusing on cluster, subject categories, keywords, burstness, centrality and research centers as for the former. For the latter, we thematically review 18 articles, selected purposefully from the articles involved, centering on purpose, approach, findings, and contribution to ACTI future directions. The findings reveal that in the past AI contribution to translation industry was not rigorous, resulting in rule-based machine translation and statistical machine translation whose output was not satisfactory. However, the more AI develops, the more machine translation develops, incorporating Neural Networking Algorithms and (Deep) Language Learning Models like ChatGPT whose translation output has developed considerably. However, much rigorous research is still needed to overcome several problems encountering translation industry, specifically concerning low-source languages, multi-dialectical and free word order languages, and cultural and religious registers.
Authors: Mohammed Q. Shormani
Abstract: There is a strong correlation between linguistics and artificial intelligence (AI), best manifested by deep learning language models. This study provides a thorough scientometric analysis of this correlation, synthesizing the intellectual production during 51 years, from 1974 to 2024. It involves 5750 Web of Science-indexed articles published in 2124 journals, which are written by 20835 authors belonging to 13773 research centers in 794 countries. Two powerful software, viz., CiteSpace and VOSviewer, were used to generate mapping visualizations of the intellectual landscape, trending issues and (re)emerging hotspots. The results indicate that in the 1980s and 1990s, linguistics and AI research was not robust, characterized by unstable publication over time. It has, however, witnessed a remarkable increase of publication since then, reaching 1478 articles in 2023, and 546 articles in January-March timespan in 2024, involving emerging issues and hotspots, addressing new horizons, new topics, and launching new applications and powerful deep learning language models including ChatGPT.
Authors: Justin Chih-Yao Chen, Zifeng Wang, Hamid Palangi, Rujun Han, Sayna Ebrahimi, Long Le, Vincent Perot, Swaroop Mishra, Mohit Bansal, Chen-Yu Lee, Tomas Pfister
Abstract: Reverse thinking plays a crucial role in human reasoning. Humans can reason not only from a problem to a solution but also in reverse, i.e., start from the solution and reason towards the problem. This often enhances overall reasoning performance as it enables consistency checks between their forward and backward thinking. To enable Large Language Models (LLMs) to perform reverse thinking, we introduce Reverse-Enhanced Thinking (RevThink), a framework composed of data augmentation and learning objectives. In RevThink, we augment the dataset by collecting structured forward-backward reasoning from a teacher model, consisting of: (1) the original question, (2) forward reasoning, (3) backward question, and (4) backward reasoning. We then employ three objectives to train a smaller student model in a multi-task learning fashion: (a) generate forward reasoning from a question, (b) generate a backward question from a question, and (c) generate backward reasoning from the backward question. Experiments across 12 datasets covering commonsense, math, and logical reasoning show an average 13.53% improvement over the student model's zero-shot performance and a 6.84% improvement over the strongest knowledge distillation baselines. Moreover, our method demonstrates sample efficiency -- using only 10% of the correct forward reasoning from the training data, it outperforms a standard fine-tuning method trained on 10x more forward reasoning. RevThink also exhibits strong generalization to out-of-distribution held-out datasets.
Authors: Leonardo Almeida, Pedro Rodrigues, Diogo Magalh\~aes, Armando J. Pinho, Diogo Pratas
Abstract: This paper introduces AIDetx, a novel method for detecting machine-generated text using data compression techniques. Traditional approaches, such as deep learning classifiers, often suffer from high computational costs and limited interpretability. To address these limitations, we propose a compression-based classification framework that leverages finite-context models (FCMs). AIDetx constructs distinct compression models for human-written and AI-generated text, classifying new inputs based on which model achieves a higher compression ratio. We evaluated AIDetx on two benchmark datasets, achieving F1 scores exceeding 97% and 99%, respectively, highlighting its high accuracy. Compared to current methods, such as large language models (LLMs), AIDetx offers a more interpretable and computationally efficient solution, significantly reducing both training time and hardware requirements (e.g., no GPUs needed). The full implementation is publicly available at https://github.com/AIDetx/AIDetx.
Authors: Daixuan Cheng, Shaohan Huang, Ziyu Zhu, Xintong Zhang, Wayne Xin Zhao, Zhongzhi Luan, Bo Dai, Zhenliang Zhang
Abstract: Recent years have witnessed the rapid development of general multimodal large language models (MLLMs). However, adapting general MLLMs to specific domains, such as scientific fields and industrial applications, remains less explored. This paper systematically investigates domain adaptation of MLLMs through post-training, focusing on data synthesis, training pipelines, and task evaluation. (1) Data Synthesis: Using open-source models, we develop a visual instruction synthesizer that effectively generates diverse visual instruction tasks from domain-specific image-caption pairs. Our synthetic tasks surpass those generated by manual rules, GPT-4, and GPT-4V in enhancing the domain-specific performance of MLLMs. (2) Training Pipeline: While the two-stage training--initially on image-caption pairs followed by visual instruction tasks--is commonly adopted for developing general MLLMs, we apply a single-stage training pipeline to enhance task diversity for domain-specific post-training. (3) Task Evaluation: We conduct experiments in two domains, biomedicine and food, by post-training MLLMs of different sources and scales (e.g., Qwen2-VL-2B, LLaVA-v1.6-8B, Llama-3.2-11B), and then evaluating MLLM performance on various domain-specific tasks. To support further research in MLLM domain adaptation, we will open-source our implementations.
Authors: Zicheng Lin, Tian Liang, Jiahao Xu, Xing Wang, Ruilin Luo, Chufan Shi, Siheng Li, Yujiu Yang, Zhaopeng Tu
Abstract: Large Language Models (LLMs) have exhibited remarkable performance on reasoning tasks. They utilize autoregressive token generation to construct reasoning trajectories, enabling the development of a coherent chain of thought. In this work, we explore the impact of individual tokens on the final outcomes of reasoning tasks. We identify the existence of ``critical tokens'' that lead to incorrect reasoning trajectories in LLMs. Specifically, we find that LLMs tend to produce positive outcomes when forced to decode other tokens instead of critical tokens. Motivated by this observation, we propose a novel approach - cDPO - designed to automatically recognize and conduct token-level rewards for the critical tokens during the alignment process. Specifically, we develop a contrastive estimation approach to automatically identify critical tokens. It is achieved by comparing the generation likelihood of positive and negative models. To achieve this, we separately fine-tune the positive and negative models on various reasoning trajectories, consequently, they are capable of identifying identify critical tokens within incorrect trajectories that contribute to erroneous outcomes. Moreover, to further align the model with the critical token information during the alignment process, we extend the conventional DPO algorithms to token-level DPO and utilize the differential likelihood from the aforementioned positive and negative model as important weight for token-level DPO learning.Experimental results on GSM8K and MATH500 benchmarks with two-widely used models Llama-3 (8B and 70B) and deepseek-math (7B) demonstrate the effectiveness of the propsoed approach cDPO.
Authors: Nirmal Joshua Kapu, Raghav Karan
Abstract: This article surveys convolution-based models including convolutional neural networks (CNNs), Conformers, ResNets, and CRNNs-as speech signal processing models and provide their statistical backgrounds and speech recognition, speaker identification, emotion recognition, and speech enhancement applications. Through comparative training cost assessment, model size, accuracy and speed assessment, we compare the strengths and weaknesses of each model, identify potential errors and propose avenues for further research, emphasizing the central role it plays in advancing applications of speech technologies.
Authors: Cheng-Fu Yang, Da Yin, Wenbo Hu, Nanyun Peng, Bolei Zhou, Kai-Wei Chang
Abstract: Humans recognize objects after observing only a few examples, a remarkable capability enabled by their inherent language understanding of the real-world environment. Developing verbalized and interpretable representation can significantly improve model generalization in low-data settings. In this work, we propose Verbalized Representation Learning (VRL), a novel approach for automatically extracting human-interpretable features for object recognition using few-shot data. Our method uniquely captures inter-class differences and intra-class commonalities in the form of natural language by employing a Vision-Language Model (VLM) to identify key discriminative features between different classes and shared characteristics within the same class. These verbalized features are then mapped to numeric vectors through the VLM. The resulting feature vectors can be further utilized to train and infer with downstream classifiers. Experimental results show that, at the same model scale, VRL achieves a 24% absolute improvement over prior state-of-the-art methods while using 95% less data and a smaller mode. Furthermore, compared to human-labeled attributes, the features learned by VRL exhibit a 20% absolute gain when used for downstream classification tasks. Code is available at: https://github.com/joeyy5588/VRL/tree/main.
Authors: Ted Kwartler, Nataliia Bagan, Ivan Banny, Alan Aqrawi, Arian Abbasi
Abstract: The Single-Turn Crescendo Attack (STCA), first introduced in Aqrawi and Abbasi [2024], is an innovative method designed to bypass the ethical safeguards of text-to-text AI models, compelling them to generate harmful content. This technique leverages a strategic escalation of context within a single prompt, combined with trust-building mechanisms, to subtly deceive the model into producing unintended outputs. Extending the application of STCA to text-to-image models, we demonstrate its efficacy by compromising the guardrails of a widely-used model, DALL-E 3, achieving outputs comparable to outputs from the uncensored model Flux Schnell, which served as a baseline control. This study provides a framework for researchers to rigorously evaluate the robustness of guardrails in text-to-image models and benchmark their resilience against adversarial attacks.
Authors: Mohamed Aghzal, Xiang Yue, Erion Plaku, Ziyu Yao
Abstract: Despite their promise to perform complex reasoning, large language models (LLMs) have been shown to have limited effectiveness in end-to-end planning. This has inspired an intriguing question: if these models cannot plan well, can they still contribute to the planning framework as a helpful plan evaluator? In this work, we generalize this question to consider LLMs augmented with visual understanding, i.e., Vision-Language Models (VLMs). We introduce PathEval, a novel benchmark evaluating VLMs as plan evaluators in complex path-planning scenarios. Succeeding in the benchmark requires a VLM to be able to abstract traits of optimal paths from the scenario description, demonstrate precise low-level perception on each path, and integrate this information to decide the better path. Our analysis of state-of-the-art VLMs reveals that these models face significant challenges on the benchmark. We observe that the VLMs can precisely abstract given scenarios to identify the desired traits and exhibit mixed performance in integrating the provided information. Yet, their vision component presents a critical bottleneck, with models struggling to perceive low-level details about a path. Our experimental results show that this issue cannot be trivially addressed via end-to-end fine-tuning; rather, task-specific discriminative adaptation of these vision encoders is needed for these VLMs to become effective path evaluators.
Authors: Feng Xiong, Runxi Cheng, Wang Chen, Zhanqiu Zhang, Yiwen Guo, Chun Yuan, Ruifeng Xu
Abstract: Model merging has gained increasing attention as an efficient and effective technique for integrating task-specific weights from various tasks into a unified multi-task model without retraining or additional data. As a representative approach, Task Arithmetic (TA) has demonstrated that combining task vectors through arithmetic operations facilitates efficient capability transfer between different tasks. In this framework, task vectors are obtained by subtracting the parameter values of a pre-trained model from those of individually fine-tuned models initialized from it. Despite the notable effectiveness of TA, interference among task vectors can adversely affect the performance of the merged model. In this paper, we relax the constraints of Task Arithmetic Property and propose Task Consistency Property, which can be regarded as being free from task interference. Through theoretical derivation, we show that such a property can be approximately achieved by seeking orthogonal task vectors. Guiding by this insight, we propose Adaptive Weight Disentanglement (AWD), which decomposes traditional task vectors into a redundant vector and several disentangled task vectors. The primary optimization objective of AWD is to achieve orthogonality among the disentangled task vectors, thereby closely approximating the desired solution. Notably, these disentangled task vectors can be seamlessly integrated into existing merging methodologies. Experimental results demonstrate that our AWD consistently and significantly improves upon previous merging approaches, achieving state-of-the-art results. Our code is available at \href{https://github.com/FarisXiong/AWD.git}{https://github.com/FarisXiong/AWD.git}.
URLs: https://github.com/FarisXiong/AWD.git, https://github.com/FarisXiong/AWD.git
Authors: Weiqiu You, Youngja Park
Abstract: Understanding the attack patterns associated with a cyberattack is crucial for comprehending the attacker's behaviors and implementing the right mitigation measures. However, majority of the information regarding new attacks is typically presented in unstructured text, posing significant challenges for security analysts in collecting necessary information. In this paper, we present a sentence classification system that can identify the attack techniques described in natural language sentences from cyber threat intelligence (CTI) reports. We propose a new method for utilizing auxiliary data with the same labels to improve classification for the low-resource cyberattack classification task. The system first trains the model using the augmented training data and then trains more using only the primary data. We validate our model using the TRAM data1 and the MITRE ATT&CK framework. Experiments show that our method enhances Macro-F1 by 5 to 9 percentage points and keeps Micro-F1 scores competitive when compared to the baseline performance on the TRAM dataset.
Authors: Haomin Zhuang, Yihua Zhang, Kehan Guo, Jinghan Jia, Gaowen Liu, Sijia Liu, Xiangliang Zhang
Abstract: Recent advancements in large language model (LLM) unlearning have shown remarkable success in removing unwanted data-model influences while preserving the model's utility for legitimate knowledge. However, despite these strides, sparse Mixture-of-Experts (MoE) LLMs--a key subset of the LLM family--have received little attention and remain largely unexplored in the context of unlearning. As MoE LLMs are celebrated for their exceptional performance and highly efficient inference processes, we ask: How can unlearning be performed effectively and efficiently on MoE LLMs? And will traditional unlearning methods be applicable to MoE architectures? Our pilot study shows that the dynamic routing nature of MoE LLMs introduces unique challenges, leading to substantial utility drops when existing unlearning methods are applied. Specifically, unlearning disrupts the router's expert selection, causing significant selection shift from the most unlearning target-related experts to irrelevant ones. As a result, more experts than necessary are affected, leading to excessive forgetting and loss of control over which knowledge is erased. To address this, we propose a novel single-expert unlearning framework, referred to as UOE, for MoE LLMs. Through expert attribution, unlearning is concentrated on the most actively engaged expert for the specified knowledge. Concurrently, an anchor loss is applied to the router to stabilize the active state of this targeted expert, ensuring focused and controlled unlearning that preserves model utility. The proposed UOE framework is also compatible with various unlearning algorithms. Extensive experiments demonstrate that UOE enhances both forget quality up to 5% and model utility by 35% on MoE LLMs across various benchmarks, LLM architectures, while only unlearning 0.06% of the model parameters.
Authors: Peter Kulits, Michael J. Black, Silvia Zuffi
Abstract: The idea of 3D reconstruction as scene understanding is foundational in computer vision. Reconstructing 3D scenes from 2D visual observations requires strong priors to disambiguate structure. Much work has been focused on the anthropocentric, which, characterized by smooth surfaces, coherent normals, and regular edges, allows for the integration of strong geometric inductive biases. Here, we consider a more challenging problem where such assumptions do not hold: the reconstruction of natural scenes containing trees, bushes, boulders, and animals. While numerous works have attempted to tackle the problem of reconstructing animals in the wild, they have focused solely on the animal, neglecting environmental context. This limits their usefulness for analysis tasks, as animals exist inherently within the 3D world, and information is lost when environmental factors are disregarded. We propose a method to reconstruct natural scenes from single images. We base our approach on recent advances leveraging the strong world priors ingrained in Large Language Models and train an autoregressive model to decode a CLIP embedding into a structured compositional scene representation, encompassing both animals and the wild (RAW). To enable this, we propose a synthetic dataset comprising one million images and thousands of assets. Our approach, having been trained solely on synthetic data, generalizes to the task of reconstructing animals and their environments in real-world images. We will release our dataset and code to encourage future research at https://raw.is.tue.mpg.de/
Authors: Hazem Darwish, Abdalrahman Al Malah, Khloud Al Jallad, Nada Ghneim
Abstract: Brain-Computer-Interface (BCI) aims to support communication-impaired patients by translating neural signals into speech. A notable research topic in BCI involves Electroencephalography (EEG) signals that measure the electrical activity in the brain. While significant advancements have been made in BCI EEG research, a major limitation still exists: the scarcity of publicly available EEG datasets for non-English languages, such as Arabic. To address this gap, we introduce in this paper ArEEG_Words dataset, a novel EEG dataset recorded from 22 participants with mean age of 22 years (5 female, 17 male) using a 14-channel Emotiv Epoc X device. The participants were asked to be free from any effects on their nervous system, such as coffee, alcohol, cigarettes, and so 8 hours before recording. They were asked to stay calm in a clam room during imagining one of the 16 Arabic Words for 10 seconds. The words include 16 commonly used words such as up, down, left, and right. A total of 352 EEG recordings were collected, then each recording was divided into multiple 250ms signals, resulting in a total of 15,360 EEG signals. To the best of our knowledge, ArEEG_Words data is the first of its kind in Arabic EEG domain. Moreover, it is publicly available for researchers as we hope that will fill the gap in Arabic EEG research.
Authors: Adam Karvonen, Can Rager, Samuel Marks, Neel Nanda
Abstract: Sparse Autoencoders (SAEs) are an interpretability technique aimed at decomposing neural network activations into interpretable units. However, a major bottleneck for SAE development has been the lack of high-quality performance metrics, with prior work largely relying on unsupervised proxies. In this work, we introduce a family of evaluations based on SHIFT, a downstream task from Marks et al. (Sparse Feature Circuits, 2024) in which spurious cues are removed from a classifier by ablating SAE features judged to be task-irrelevant by a human annotator. We adapt SHIFT into an automated metric of SAE quality; this involves replacing the human annotator with an LLM. Additionally, we introduce the Targeted Probe Perturbation (TPP) metric that quantifies an SAE's ability to disentangle similar concepts, effectively scaling SHIFT to a wider range of datasets. We apply both SHIFT and TPP to multiple open-source models, demonstrating that these metrics effectively differentiate between various SAE training hyperparameters and architectures.
Authors: Vishnou Vinayagame, Gregory Senay, Luis Mart\'i
Abstract: Mathematical reasoning capabilities are increasing with tool-augmented language agents, but methods often rely either on closed-source or large models, external data, or extensive prompt engineering. This work introduces MATATA, a novel cost-effective method to train LLM agents for tabular data problems through reasoning, planning, and tool use. With a progressive self-improvement paradigm and an iterative weak supervision, it empowers 3.8B/8B Small Language Models (SLMs), particularly suited for local hosting and sensitive business contexts where data privacy is crucial. By employing a flexible and reusable tools across different datasets, it achieves robust performance with effective scalability across shared tasks. Experiments show that MATATA reaches state-of-the-art performances on FinQA and TAT-QA among reasoning frameworks based on open-source models. Moreover, MATATA models compete with GPT-4 based frameworks on TabMWP, while being SLMs.
Authors: Jeongho Ju, Daeyoung Kim, SunYoung Park, Youngjune Kim
Abstract: In this paper, we introduce an open-source Korean-English vision-language model (VLM), VARCO-VISION. We incorporate a step-by-step training strategy that allows a model learn both linguistic and visual information while preserving the backbone model's knowledge. Our model demonstrates outstanding performance in diverse settings requiring bilingual image-text understanding and generation abilities compared to models of similar size. VARCO-VISION is also capable of grounding, referring, and OCR, expanding its usage and potential applications for real-world scenarios. In addition to the model, we release five Korean evaluation datasets, including four closed-set and one openset benchmarks. We anticipate that our milestone will broaden the opportunities for AI researchers aiming to train VLMs. VARCO-VISION is available at https://huggingface.co/NCSOFT/VARCO-VISION-14B.
Authors: Roberto Balestri
Abstract: This study investigates ChatGPT-4o's multimodal content generation, highlighting significant disparities in its treatment of sexual content and nudity versus violent and drug-related themes. Detailed analysis reveals that ChatGPT-4o consistently censors sexual content and nudity, while showing leniency towards violence and drug use. Moreover, a pronounced gender bias emerges, with female-specific content facing stricter regulation compared to male-specific content. This disparity likely stems from media scrutiny and public backlash over past AI controversies, prompting tech companies to impose stringent guidelines on sensitive issues to protect their reputations. Our findings emphasize the urgent need for AI systems to uphold genuine ethical standards and accountability, transcending mere political correctness. This research contributes to the understanding of biases in AI-driven language and multimodal models, calling for more balanced and ethical content moderation practices.
Authors: Luca Barsellotti, Lorenzo Bianchi, Nicola Messina, Fabio Carrara, Marcella Cornia, Lorenzo Baraldi, Fabrizio Falchi, Rita Cucchiara
Abstract: Open-Vocabulary Segmentation (OVS) aims at segmenting images from free-form textual concepts without predefined training classes. While existing vision-language models such as CLIP can generate segmentation masks by leveraging coarse spatial information from Vision Transformers, they face challenges in spatial localization due to their global alignment of image and text features. Conversely, self-supervised visual models like DINO excel in fine-grained visual encoding but lack integration with language. To bridge this gap, we present Talk2DINO, a novel hybrid approach that combines the spatial accuracy of DINOv2 with the language understanding of CLIP. Our approach aligns the textual embeddings of CLIP to the patch-level features of DINOv2 through a learned mapping function without the need to fine-tune the underlying backbones. At training time, we exploit the attention maps of DINOv2 to selectively align local visual patches with textual embeddings. We show that the powerful semantic and localization abilities of Talk2DINO can enhance the segmentation process, resulting in more natural and less noisy segmentations, and that our approach can also effectively distinguish foreground objects from the background. Experimental results demonstrate that Talk2DINO achieves state-of-the-art performance across several unsupervised OVS benchmarks. Source code and models are publicly available at: https://lorebianchi98.github.io/Talk2DINO/.
Authors: Mohamed Fazli Imam, Rufael Fedaku Marew, Jameel Hassan, Mustansar Fiaz, Alham Fikri Aji, Hisham Cholakkal
Abstract: In the era of foundation models, CLIP has emerged as a powerful tool for aligning text and visual modalities into a common embedding space. However, the alignment objective used to train CLIP often results in subpar visual features for fine-grained tasks. In contrast, SSL-pretrained models like DINO excel at extracting rich visual features due to their specialized training paradigm. Yet, these SSL models require an additional supervised linear probing step, which relies on fully labeled data which is often expensive and difficult to obtain at scale. In this paper, we propose a label-free prompt-tuning method that leverages the rich visual features of self-supervised learning models (DINO) and the broad textual knowledge of large language models (LLMs) to largely enhance CLIP-based image classification performance using unlabeled images. Our approach unfolds in three key steps: (1) We generate robust textual feature embeddings that more accurately represent object classes by leveraging class-specific descriptions from LLMs, enabling more effective zero-shot classification compared to CLIP's default name-specific prompts. (2) These textual embeddings are then used to produce pseudo-labels to train an alignment module that integrates the complementary strengths of LLM description-based textual embeddings and DINO's visual features. (3) Finally, we prompt-tune CLIP's vision encoder through DINO-assisted supervision using the trained alignment module. This three-step process allows us to harness the best of visual and textual foundation models, resulting in a powerful and efficient approach that surpasses state-of-the-art label-free classification methods. Notably, our framework, NoLA (No Labels Attached), achieves an average absolute gain of 3.6% over the state-of-the-art LaFter across 11 diverse image classification datasets.
Authors: Xi Zhang, Zaiqiao Meng, Jake Lever, Edmond S. L. Ho
Abstract: Radiology report generation (RRG) is a challenging task, as it requires a thorough understanding of medical images, integration of multiple temporal inputs, and accurate report generation. Effective interpretation of medical images, such as chest X-rays (CXRs), demands sophisticated visual-language reasoning to map visual findings to structured reports. Recent studies have shown that multimodal large language models (MLLMs) can acquire multimodal capabilities by aligning with pre-trained vision encoders. However, current approaches predominantly focus on single-image analysis or utilise rule-based symbolic processing to handle multiple images, thereby overlooking the essential temporal information derived from comparing current images with prior ones. To overcome this critical limitation, we introduce Libra, a temporal-aware MLLM tailored for CXR report generation using temporal images. Libra integrates a radiology-specific image encoder with a MLLM and utilises a novel Temporal Alignment Connector to capture and synthesise temporal information of images across different time points with unprecedented precision. Extensive experiments show that Libra achieves new state-of-the-art performance among the same parameter scale MLLMs for RRG tasks on the MIMIC-CXR. Specifically, Libra improves the RadCliQ metric by 12.9% and makes substantial gains across all lexical metrics compared to previous models.
Authors: Safaa Abdullahi Moallim Mohamud, Ho-Young Jung
Abstract: In this paper, we introduce the Actions and Objects Pathways (AOPath) for out-of-domain generalization in video question answering tasks. AOPath leverages features from a large pretrained model to enhance generalizability without the need for explicit training on the unseen domains. Inspired by human brain, AOPath dissociates the pretrained features into action and object features, and subsequently processes them through separate reasoning pathways. It utilizes a novel module which converts out-of-domain features into domain-agnostic features without introducing any trainable weights. We validate the proposed approach on the TVQA dataset, which is partitioned into multiple subsets based on genre to facilitate the assessment of generalizability. The proposed approach demonstrates 5% and 4% superior performance over conventional classifiers on out-of-domain and in-domain datasets, respectively. It also outperforms prior methods that involve training millions of parameters, whereas the proposed approach trains very few parameters.
Authors: Zipeng Qiu, You Peng, Guangxin He, Binhang Yuan, Chen Wang
Abstract: The advent of large language models (LLMs) has unlocked great opportunities in complex data management tasks, particularly in question answering (QA) over complicated multi-table relational data. Despite significant progress, systematically evaluating LLMs on multi-table QA remains a critical challenge due to the inherent complexity of analyzing heterogeneous table structures and potential large scale of serialized relational data. Existing benchmarks primarily focus on single-table QA, failing to capture the intricacies of reasoning across multiple relational tables, as required in real-world domains such as finance, healthcare, and e-commerce. To address this gap, we present TQA-Bench, a new multi-table QA benchmark designed to evaluate the capabilities of LLMs in tackling complex QA tasks over relational data. Our benchmark incorporates diverse relational database instances sourced from real-world public datasets and introduces a flexible sampling mechanism to create tasks with varying multi-table context lengths, ranging from 8K to 64K tokens. To ensure robustness and reliability, we integrate symbolic extensions into the evaluation framework, enabling the assessment of LLM reasoning capabilities beyond simple data retrieval or probabilistic pattern matching. We systematically evaluate a range of LLMs, both open-source and closed-source, spanning model scales from 7 billion to 70 billion parameters. Our extensive experiments reveal critical insights into the performance of LLMs in multi-table QA, highlighting both challenges and opportunities for advancing their application in complex, data-driven environments. Our benchmark implementation and results are available at https://github.com/Relaxed-System-Lab/TQA-Bench.
Authors: Yuta Ojima, Hiroki Sakaji, Tadashi Nakamura, Hiroaki Sakata, Kazuya Seki, Yuu Teshigawara, Masami Yamashita, Kazuhiro Aoyama
Abstract: This paper presents a knowledge management system for automobile failure analysis using retrieval-augmented generation (RAG) with large language models (LLMs) and knowledge graphs (KGs). In the automotive industry, there is a growing demand for knowledge transfer of failure analysis from experienced engineers to young engineers. However, failure events are phenomena that occur in a chain reaction, making them difficult for beginners to analyze them. While knowledge graphs, which can describe semantic relationships and structure information is effective in representing failure events, due to their capability of representing the relationships between components, there is much information in KGs, so it is challenging for young engineers to extract and understand sub-graphs from the KG. On the other hand, there is increasing interest in the use of Graph RAG, a type of RAG that combines LLMs and KGs for knowledge management. However, when using the current Graph RAG framework with an existing knowledge graph for automobile failures, several issues arise because it is difficult to generate executable queries for a knowledge graph database which is not constructed by LLMs. To address this, we focused on optimizing the Graph RAG pipeline for existing knowledge graphs. Using an original Q&A dataset, the ROUGE F1 score of the sentences generated by the proposed method showed an average improvement of 157.6% compared to the current method. This highlights the effectiveness of the proposed method for automobile failure analysis.
Authors: Qiong Wu, Wenhao Lin, Weihao Ye, Yiyi Zhou, Xiaoshuai Sun, Rongrong Ji
Abstract: The excessive use of visual tokens in existing Multimoal Large Language Models (MLLMs) often exhibits obvious redundancy and brings in prohibitively expensive computation. To gain insights into this problem, we first conduct extensive empirical studies on the attention behaviors of MLLMs, and summarize three main inference stages in MLLMs: (i) Early fusion between tokens is first accomplished quickly. (ii) Intra-modality modeling then comes to play. (iii) Multimodal reasoning} resumes and lasts until the end of inference. In particular, we reveal that visual tokens will stop contributing to reasoning when the text tokens receive enough image information, yielding obvious visual redundancy. Based on these generalized observations, we propose a simple yet effective method to improve the efficiency of MLLMs, termed dynamic visual-token exit (DyVTE). DyVTE uses lightweight hyper-networks to perceive the text token status and decide the removal of all visual tokens after a certain layer, thereby addressing the observed visual redundancy. To validate VTE, we apply it to a set of MLLMs, including LLaVA, VILA, Eagle and InternVL, and conduct extensive experiments on a bunch of benchmarks. The experiment results not only show the effectiveness of our VTE in improving MLLMs' efficiency, but also yield the general modeling patterns of MLLMs, well facilitating the in-depth understanding of MLLMs. Our code is anonymously released at https://github.com/DoubtedSteam/DyVTE.
Authors: Qixiu Li, Yaobo Liang, Zeyu Wang, Lin Luo, Xi Chen, Mozheng Liao, Fangyun Wei, Yu Deng, Sicheng Xu, Yizhong Zhang, Xiaofan Wang, Bei Liu, Jianlong Fu, Jianmin Bao, Dong Chen, Yuanchun Shi, Jiaolong Yang, Baining Guo
Abstract: The advancement of large Vision-Language-Action (VLA) models has significantly improved robotic manipulation in terms of language-guided task execution and generalization to unseen scenarios. While existing VLAs adapted from pretrained large Vision-Language-Models (VLM) have demonstrated promising generalizability, their task performance is still unsatisfactory as indicated by the low tasks success rates in different environments. In this paper, we present a new advanced VLA architecture derived from VLM. Unlike previous works that directly repurpose VLM for action prediction by simple action quantization, we propose a omponentized VLA architecture that has a specialized action module conditioned on VLM output. We systematically study the design of the action module and demonstrates the strong performance enhancement with diffusion action transformers for action sequence modeling, as well as their favorable scaling behaviors. We also conduct comprehensive experiments and ablation studies to evaluate the efficacy of our models with varied designs. The evaluation on 5 robot embodiments in simulation and real work shows that our model not only significantly surpasses existing VLAs in task performance and but also exhibits remarkable adaptation to new robots and generalization to unseen objects and backgrounds. It exceeds the average success rates of OpenVLA which has similar model size (7B) with ours by over 35% in simulated evaluation and 55% in real robot experiments. It also outperforms the large RT-2-X model (55B) by 18% absolute success rates in simulation. Code and models can be found on our project page (https://cogact.github.io/).
Authors: Haorui He, Yuchen Song, Yuancheng Wang, Haoyang Li, Xueyao Zhang, Li Wang, Gongping Huang, Eng Siong Chng, Zhizheng Wu
Abstract: One-shot voice conversion (VC) aims to alter the timbre of speech from a source speaker to match that of a target speaker using just a single reference speech from the target, while preserving the semantic content of the original source speech. Despite advancements in one-shot VC, its effectiveness decreases in real-world scenarios where reference speeches, often sourced from the internet, contain various disturbances like background noise. To address this issue, we introduce Noro, a Noise Robust One-shot VC system. Noro features innovative components tailored for VC using noisy reference speeches, including a dual-branch reference encoding module and a noise-agnostic contrastive speaker loss. Experimental results demonstrate that Noro outperforms our baseline system in both clean and noisy scenarios, highlighting its efficacy for real-world applications. Additionally, we investigate the hidden speaker representation capabilities of our baseline system by repurposing its reference encoder as a speaker encoder. The results shows that it is competitive with several advanced self-supervised learning models for speaker representation under the SUPERB settings, highlighting the potential for advancing speaker representation learning through one-shot VC task.
Authors: Tiantian Geng, Jinrui Zhang, Qingni Wang, Teng Wang, Jinming Duan, Feng Zheng
Abstract: Despite impressive advancements in video understanding, most efforts remain limited to coarse-grained or visual-only video tasks. However, real-world videos encompass omni-modal information (vision, audio, and speech) with a series of events forming a cohesive storyline. The lack of multi-modal video data with fine-grained event annotations and the high cost of manual labeling are major obstacles to comprehensive omni-modality video perception. To address this gap, we propose an automatic pipeline consisting of high-quality multi-modal video filtering, semantically coherent omni-modal event boundary detection, and cross-modal correlation-aware event captioning. In this way, we present LongVALE, the first-ever Vision-Audio-Language Event understanding benchmark comprising 105K omni-modal events with precise temporal boundaries and detailed relation-aware captions within 8.4K high-quality long videos. Further, we build a baseline that leverages LongVALE to enable video large language models (LLMs) for omni-modality fine-grained temporal video understanding for the first time. Extensive experiments demonstrate the effectiveness and great potential of LongVALE in advancing comprehensive multi-modal video understanding.
Authors: Guofeng Mei, Wei Lin, Luigi Riz, Yujiao Wu, Fabio Poiesi, Yiming Wang
Abstract: Enabling Large Language Models (LLMs) to understand the 3D physical world is an emerging yet challenging research direction. Current strategies for processing point clouds typically downsample the scene or divide it into smaller parts for separate analysis. However, both approaches risk losing key local details or global contextual information. In this paper, we introduce PerLA, a 3D language assistant designed to be more perceptive to both details and context, making visual representations more informative for the LLM. PerLA captures high-resolution (local) details in parallel from different point cloud areas and integrates them with (global) context obtained from a lower-resolution whole point cloud. We present a novel algorithm that preserves point cloud locality through the Hilbert curve and effectively aggregates local-to-global information via cross-attention and a graph neural network. Lastly, we introduce a novel loss for local representation consensus to promote training stability. PerLA outperforms state-of-the-art 3D language assistants, with gains of up to +1.34 CiDEr on ScanQA for question answering, and +4.22 on ScanRefer and +3.88 on Nr3D for dense captioning.\url{https://gfmei.github.io/PerLA/}
Authors: Yiming Wu, Wei Ji, Kecheng Zheng, Zicheng Wang, Dong Xu
Abstract: Recently, human motion analysis has experienced great improvement due to inspiring generative models such as the denoising diffusion model and large language model. While the existing approaches mainly focus on generating motions with textual descriptions and overlook the reciprocal task. In this paper, we present~\textbf{MoTe}, a unified multi-modal model that could handle diverse tasks by learning the marginal, conditional, and joint distributions of motion and text simultaneously. MoTe enables us to handle the paired text-motion generation, motion captioning, and text-driven motion generation by simply modifying the input context. Specifically, MoTe is composed of three components: Motion Encoder-Decoder (MED), Text Encoder-Decoder (TED), and Moti-on-Text Diffusion Model (MTDM). In particular, MED and TED are trained for extracting latent embeddings, and subsequently reconstructing the motion sequences and textual descriptions from the extracted embeddings, respectively. MTDM, on the other hand, performs an iterative denoising process on the input context to handle diverse tasks. Experimental results on the benchmark datasets demonstrate the superior performance of our proposed method on text-to-motion generation and competitive performance on motion captioning.
Authors: Aymeric Vinot, Nicolas Perez
Abstract: In most team-based esports, voice communications are prominent in the team efficiency and synergy. In fact it has been observed that not only the skill aspect of the team but also the team effective voice communication comes into play when trying to have good performance in official matches. With the recent emergence of LLM (Large Language Models) tools regarding NLP (Natural Language Processing) (Vaswani et. al.), we decided to try applying them in order to have a better understanding on how to improve the effectiveness of the voice communications. In this paper the study has been made through the prism of League of Legends esport. However the main concepts and ideas can be easily applicable in any other team related esports.
Authors: Xiang minjie
Abstract: Research on Speech Emotion Recognition (SER) often faces challenges such as the lack of large-scale public datasets and limited generalization capability when dealing with data from different distributions. To solve this problem, this paper proposes a cross-corpus speech emotion recognition method based on supervised contrast learning. The method employs a two-stage fine-tuning process: first, the self-supervised speech representation model is fine-tuned using supervised contrastive learning on multiple speech emotion datasets; then, the classifier is fine-tuned on the target dataset. The experimental results show that the WavLM-based model achieved unweighted accuracy (UA) of 77.41% on the IEMOCAP dataset and 96.49% on the CASIA dataset, outperforming the state-of-the-art results on the two datasets.
Authors: Ali Lotfi, Ian McQuillan, Steven Rayan
Abstract: L-systems can be made to model and create simulations of many biological processes, such as plant development. Finding an L-system for a given process is typically solved by hand, by experts, in a hugely time-consuming process. It would be significant if this could be done automatically from data, such as from sequences of images. In this paper, we are interested in inferring a particular type of L-system, deterministic context-free L-system (D0L-system) from a sequence of strings. We introduce the characteristic graph of a sequence of strings, which we then utilize to translate our problem (inferring D0L-system) in polynomial time into the maximum independent set problem (MIS) and the SAT problem. After that, we offer a classical exact algorithm and an approximate quantum algorithm for the problem.
Authors: Wenjia Wang, Liang Pan, Zhiyang Dou, Zhouyingcheng Liao, Yuke Lou, Lei Yang, Jingbo Wang, Taku Komura
Abstract: Simulating long-term human-scene interaction is a challenging yet fascinating task. Previous works have not effectively addressed the generation of long-term human scene interactions with detailed narratives for physics-based animation. This paper introduces a novel framework for the planning and controlling of long-horizon physical plausible human-scene interaction. On the one hand, films and shows with stylish human locomotions or interactions with scenes are abundantly available on the internet, providing a rich source of data for script planning. On the other hand, Large Language Models (LLMs) can understand and generate logical storylines. This motivates us to marry the two by using an LLM-based pipeline to extract scripts from videos, and then employ LLMs to imitate and create new scripts, capturing complex, time-series human behaviors and interactions with environments. By leveraging this, we utilize a dual-aware policy that achieves both language comprehension and scene understanding to guide character motions within contextual and spatial constraints. To facilitate training and evaluation, we contribute a comprehensive planning dataset containing diverse motion sequences extracted from real-world videos and expand them with large language models. We also collect and re-annotate motion clips from existing kinematic datasets to enable our policy learn diverse skills. Extensive experiments demonstrate the effectiveness of our framework in versatile task execution and its generalization ability to various scenarios, showing remarkably enhanced performance compared with existing methods. Our code and data will be publicly available soon.
Authors: Xuhao Hu, Dongrui Liu, Hao Li, Xuanjing Huang, Jing Shao
Abstract: Safety concerns of Multimodal large language models (MLLMs) have gradually become an important problem in various applications. Surprisingly, previous works indicate a counter-intuitive phenomenon that using textual unlearning to align MLLMs achieves comparable safety performances with MLLMs trained with image-text pairs. To explain such a counter-intuitive phenomenon, we discover a visual safety information leakage (VSIL) problem in existing multimodal safety benchmarks, i.e., the potentially risky and sensitive content in the image has been revealed in the textual query. In this way, MLLMs can easily refuse these sensitive text-image queries according to textual queries. However, image-text pairs without VSIL are common in real-world scenarios and are overlooked by existing multimodal safety benchmarks. To this end, we construct multimodal visual leakless safety benchmark (VLSBench) preventing visual safety leakage from image to textual query with 2.4k image-text pairs. Experimental results indicate that VLSBench poses a significant challenge to both open-source and close-source MLLMs, including LLaVA, Qwen2-VL, Llama3.2-Vision, and GPT-4o. This study demonstrates that textual alignment is enough for multimodal safety scenarios with VSIL, while multimodal alignment is a more promising solution for multimodal safety scenarios without VSIL. Please see our code and data at: http://hxhcreate.github.io/VLSBench
Authors: Joseph Heyward, Jo\~ao Carreira, Dima Damen, Andrew Zisserman, Viorica P\u{a}tr\u{a}ucean
Abstract: Following the successful 2023 edition, we organised the Second Perception Test challenge as a half-day workshop alongside the IEEE/CVF European Conference on Computer Vision (ECCV) 2024, with the goal of benchmarking state-of-the-art video models and measuring the progress since last year using the Perception Test benchmark. This year, the challenge had seven tracks (up from six last year) and covered low-level and high-level tasks, with language and non-language interfaces, across video, audio, and text modalities; the additional track covered hour-long video understanding and introduced a novel video QA benchmark 1h-walk VQA. Overall, the tasks in the different tracks were: object tracking, point tracking, temporal action localisation, temporal sound localisation, multiple-choice video question-answering, grounded video question-answering, and hour-long video question-answering. We summarise in this report the challenge tasks and results, and introduce in detail the novel hour-long video QA benchmark 1h-walk VQA.
Authors: Shukang Yin, Chaoyou Fu, Sirui Zhao, Yunhang Shen, Chunjiang Ge, Yan Yang, Zuwei Long, Yuhan Dai, Tong Xu, Xing Sun, Ran He, Caifeng Shan, Enhong Chen
Abstract: The success of Multimodal Large Language Models (MLLMs) in the image domain has garnered wide attention from the research community. Drawing on previous successful experiences, researchers have recently explored extending the success to the video understanding realms. Apart from training from scratch, an efficient way is to utilize the pre-trained image-LLMs, leading to two mainstream approaches, i.e. zero-shot inference and further fine-tuning with video data. In this work, our study of these approaches harvests an effective data augmentation method. We first make a deeper inspection of the zero-shot inference way and identify two limitations, i.e. limited generalization and lack of temporal understanding capabilities. Thus, we further investigate the fine-tuning approach and find a low learning efficiency when simply using all the video data samples, which can be attributed to a lack of instruction diversity. Aiming at this issue, we develop a method called T2Vid to synthesize video-like samples to enrich the instruction diversity in the training corpus. Integrating these data enables a simple and efficient training scheme, which achieves performance comparable to or even superior to using full video datasets by training with just 15% the sample size. Meanwhile, we find that the proposed scheme can boost the performance of long video understanding without training with long video samples. We hope our study will spark more thinking about using MLLMs for video understanding and curation of high-quality data. The code is released at https://github.com/xjtupanda/T2Vid.
Authors: Andor Diera, Nicolas Lell, Aygul Garifullina, Ansgar Scherp
Abstract: Privacy preserving deep learning is an emerging field in machine learning that aims to mitigate the privacy risks in the use of deep neural networks. One such risk is training data extraction from language models that have been trained on datasets, which contain personal and privacy sensitive information. In our study, we investigate the extent of named entity memorization in fine-tuned BERT models. We use single-label text classification as representative downstream task and employ three different fine-tuning setups in our experiments, including one with Differential Privacy (DP). We create a large number of text samples from the fine-tuned BERT models utilizing a custom sequential sampling strategy with two prompting strategies. We search in these samples for named entities and check if they are also present in the fine-tuning datasets. We experiment with two benchmark datasets in the domains of emails and blogs. We show that the application of DP has a detrimental effect on the text generation capabilities of BERT. Furthermore, we show that a fine-tuned BERT does not generate more named entities specific to the fine-tuning dataset than a BERT model that is pre-trained only. This suggests that BERT is unlikely to emit personal or privacy sensitive named entities. Overall, our results are important to understand to what extent BERT-based services are prone to training data extraction attacks.
Authors: Yu Ying Chiu, Ashish Sharma, Inna Wanyin Lin, Tim Althoff
Abstract: The emergence of large language models (LLMs) like ChatGPT has increased interest in their use as therapists to address mental health challenges and the widespread lack of access to care. However, experts have emphasized the critical need for systematic evaluation of LLM-based mental health interventions to accurately assess their capabilities and limitations. Here, we propose BOLT, a proof-of-concept computational framework to systematically assess the conversational behavior of LLM therapists. We quantitatively measure LLM behavior across 13 psychotherapeutic approaches with in-context learning methods. Then, we compare the behavior of LLMs against high- and low-quality human therapy. Our analysis based on Motivational Interviewing therapy reveals that LLMs often resemble behaviors more commonly exhibited in low-quality therapy rather than high-quality therapy, such as offering a higher degree of problem-solving advice when clients share emotions. However, unlike low-quality therapy, LLMs reflect significantly more upon clients' needs and strengths. Our findings caution that LLM therapists still require further research for consistent, high-quality care.
Authors: Heeseung Kim, Soonshin Seo, Kyeongseok Jeong, Ohsung Kwon, Soyoon Kim, Jungwhan Kim, Jaehong Lee, Eunwoo Song, Myungwoo Oh, Jung-Woo Ha, Sungroh Yoon, Kang Min Yoo
Abstract: Recent work shows promising results in expanding the capabilities of large language models (LLM) to directly understand and synthesize speech. However, an LLM-based strategy for modeling spoken dialogs remains elusive, calling for further investigation. This paper introduces an extensive speech-text LLM framework, the Unified Spoken Dialog Model (USDM), designed to generate coherent spoken responses with naturally occurring prosodic features relevant to the given input speech without relying on explicit automatic speech recognition (ASR) or text-to-speech (TTS) systems. We have verified the inclusion of prosody in speech tokens that predominantly contain semantic information and have used this foundation to construct a prosody-infused speech-text model. Additionally, we propose a generalized speech-text pretraining scheme that enhances the capture of cross-modal semantics. To construct USDM, we fine-tune our speech-text model on spoken dialog data using a multi-step spoken dialog template that stimulates the chain-of-reasoning capabilities exhibited by the underlying LLM. Automatic and human evaluations on the DailyTalk dataset demonstrate that our approach effectively generates natural-sounding spoken responses, surpassing previous and cascaded baselines. Our code and checkpoints are available at https://github.com/naver-ai/usdm.
Authors: R. Patrick Xian, Alex J. Lee, Satvik Lolla, Vincent Wang, Qiming Cui, Russell Ro, Reza Abbasi-Asl
Abstract: The increasing depth of parametric domain knowledge in large language models (LLMs) is fueling their rapid deployment in real-world applications. Understanding model vulnerabilities in high-stakes and knowledge-intensive tasks is essential for quantifying the trustworthiness of model predictions and regulating their use. The recent discovery of named entities as adversarial examples (i.e. adversarial entities) in natural language processing tasks raises questions about their potential impact on the knowledge robustness of pre-trained and finetuned LLMs in high-stakes and specialized domains. We examined the use of type-consistent entity substitution as a template for collecting adversarial entities for billion-parameter LLMs with biomedical knowledge. To this end, we developed an embedding-space attack based on powerscaled distance-weighted sampling to assess the robustness of their biomedical knowledge with a low query budget and controllable coverage. Our method has favorable query efficiency and scaling over alternative approaches based on random sampling and blackbox gradient-guided search, which we demonstrated for adversarial distractor generation in biomedical question answering. Subsequent failure mode analysis uncovered two regimes of adversarial entities on the attack surface with distinct characteristics and we showed that entity substitution attacks can manipulate token-wise Shapley value explanations, which become deceptive in this setting. Our approach complements standard evaluations for high-capacity models and the results highlight the brittleness of domain knowledge in LLMs.
Authors: Jie Liu, Wenxuan Wang, Yihang Su, Jingyuan Huan, Wenting Chen, Yudi Zhang, Cheng-Yi Li, Kao-Jung Chang, Xiaohan Xin, Linlin Shen, Michael R. Lyu
Abstract: The significant breakthroughs of Medical Multi-Modal Large Language Models (Med-MLLMs) renovate modern healthcare with robust information synthesis and medical decision support. However, these models are often evaluated on benchmarks that are unsuitable for the Med-MLLMs due to the complexity of real-world diagnostics across diverse specialties. To address this gap, we introduce Asclepius, a novel Med-MLLM benchmark that comprehensively assesses Med-MLLMs in terms of: distinct medical specialties (cardiovascular, gastroenterology, etc.) and different diagnostic capacities (perception, disease analysis, etc.). Grounded in 3 proposed core principles, Asclepius ensures a comprehensive evaluation by encompassing 15 medical specialties, stratifying into 3 main categories and 8 sub-categories of clinical tasks, and exempting overlap with existing VQA dataset. We further provide an in-depth analysis of 6 Med-MLLMs and compare them with 3 human specialists, providing insights into their competencies and limitations in various medical contexts. Our work not only advances the understanding of Med-MLLMs' capabilities but also sets a precedent for future evaluations and the safe deployment of these models in clinical environments.
Authors: Marco Gaido, Sara Papi, Matteo Negri, Luisa Bentivogli
Abstract: The field of natural language processing (NLP) has recently witnessed a transformative shift with the emergence of foundation models, particularly Large Language Models (LLMs) that have revolutionized text-based NLP. This paradigm has extended to other modalities, including speech, where researchers are actively exploring the combination of Speech Foundation Models (SFMs) and LLMs into single, unified models capable of addressing multimodal tasks. Among such tasks, this paper focuses on speech-to-text translation (ST). By examining the published papers on the topic, we propose a unified view of the architectural solutions and training strategies presented so far, highlighting similarities and differences among them. Based on this examination, we not only organize the lessons learned but also show how diverse settings and evaluation approaches hinder the identification of the best-performing solution for each architectural building block and training choice. Lastly, we outline recommendations for future works on the topic aimed at better understanding the strengths and weaknesses of the SFM+LLM solutions for ST.
Authors: Mohammed Safi Ur Rahman Khan, Priyam Mehta, Ananth Sankar, Umashankar Kumaravelan, Sumanth Doddapaneni, Suriyaprasaad B, Varun Balan G, Sparsh Jain, Anoop Kunchukuttan, Pratyush Kumar, Raj Dabre, Mitesh M. Khapra
Abstract: Despite the considerable advancements in English LLMs, the progress in building comparable models for other languages has been hindered due to the scarcity of tailored resources. Our work aims to bridge this divide by introducing an expansive suite of resources specifically designed for the development of Indic LLMs, covering 22 languages, containing a total of 251B tokens and 74.8M instruction-response pairs. Recognizing the importance of both data quality and quantity, our approach combines highly curated manually verified data, unverified yet valuable data, and synthetic data. We build a clean, open-source pipeline for curating pre-training data from diverse sources, including websites, PDFs, and videos, incorporating best practices for crawling, cleaning, flagging, and deduplication. For instruction-fine tuning, we amalgamate existing Indic datasets, translate/transliterate English datasets into Indian languages, and utilize LLaMa2 and Mixtral models to create conversations grounded in articles from Indian Wikipedia and Wikihow. Additionally, we address toxicity alignment by generating toxic prompts for multiple scenarios and then generate non-toxic responses by feeding these toxic prompts to an aligned LLaMa2 model. We hope that the datasets, tools, and resources released as a part of this work will not only propel the research and development of Indic LLMs but also establish an open-source blueprint for extending such efforts to other languages. The data and other artifacts created as part of this work are released with permissive licenses.
Authors: Wanyun Cui, Qianle Wang
Abstract: This paper reveals the phenomenon of parameter heterogeneity in large language models (LLMs). We find that a small subset of "cherry" parameters exhibit a disproportionately large influence on model performance, while the vast majority of parameters have minimal impact. This heterogeneity is found to be prevalent across different model families, scales, and types. Motivated by this observation, we propose CherryQ, a novel quantization method that unifies the optimization of mixed-precision parameters. CherryQ identifies and preserves the critical cherry parameters in high precision while aggressively quantizing the remaining parameters to low precision. Extensive experiments demonstrate the effectiveness of CherryQ. CherryQ outperforms existing quantization approaches in terms of perplexity and downstream task performance. Notably, our 3-bit quantized Vicuna-1.5 exhibits competitive performance compared to their 16-bit counterparts.
Authors: Anubhav Jangra, Jamshid Mozafari, Adam Jatowt, Smaranda Muresan
Abstract: Digital education has gained popularity in the last decade, especially after the COVID-19 pandemic. With the improving capabilities of large language models to reason and communicate with users, envisioning intelligent tutoring systems (ITSs) that can facilitate self-learning is not very far-fetched. One integral component to fulfill this vision is the ability to give accurate and effective feedback via hints to scaffold the learning process. In this survey article, we present a comprehensive review of prior research on hint generation, aiming to bridge the gap between research in education and cognitive science, and research in AI and Natural Language Processing. Informed by our findings, we propose a formal definition of the hint generation task, and discuss the roadmap of building an effective hint generation system aligned with the formal definition, including open challenges, future directions and ethical considerations.
Authors: R. Michael Alvarez, Jacob Morrier
Abstract: This paper proposes a novel methodology for assessing the quality of answers in political question-and-answer sessions. Our approach consists of measuring the quality of an answer based on how accurately it can be identified among all observed answers given the question. This reflects the relevance and depth of engagement of the answer to the question. Similarly to semantic search, this measurement approach can be implemented by training a language model on the corpus of observed questions and answers without additional labeled data. We showcase and validate our methodology using data from the Question Period in the Canadian House of Commons. Our analysis reveals that while some answers have a weak semantic connection with questions, hinting at some evasion or obfuscation, answers are generally relevant, far surpassing what would be expected from random replies. Besides, our findings provide valuable insights into the correlates of answer quality. We find significant variations based on the party affiliation of the members of Parliament posing the questions. Finally, we uncover a meaningful correlation between the quality of answers and the topic of the questions.
Authors: Chuanbo Hu, Wenqi Li, Mindi Ruan, Xiangxu Yu, Shalaka Deshpande, Lynn K. Paul, Shuo Wang, Xin Li
Abstract: Diagnosing language disorders associated with autism is a complex challenge, often hampered by the subjective nature and variability of traditional assessment methods. Traditional diagnostic methods not only require intensive human effort but also often result in delayed interventions due to their lack of speed and precision. In this study, we explored the application of ChatGPT, a large language model, to overcome these obstacles by enhancing sensitivity and profiling linguistic features for autism diagnosis. This research utilizes ChatGPT natural language processing capabilities to simplify and improve the diagnostic process, focusing on identifying autism related language patterns. Specifically, we compared ChatGPT performance with that of conventional supervised learning models, including BERT, a model acclaimed for its effectiveness in various natural language processing tasks. We showed that ChatGPT substantially outperformed these models, achieving over 10% improvement in both sensitivity and positive predictive value, in a zero shot learning configuration. The findings underscore the model potential as a diagnostic tool, combining accuracy and applicability. We identified ten key features of autism associated language disorders across scenarios. Features such as echolalia, pronoun reversal, and atypical language usage play a critical role in diagnosing ASD and informing tailored treatment plans. Together, our findings advocate for adopting sophisticated AI tools like ChatGPT in clinical settings to assess and diagnose developmental disorders. Our approach promises enhanced diagnostic precision and supports personalized medicine, potentially transforming the evaluation landscape for autism and similar neurological conditions.
Authors: Yutao Yang, Jie Zhou, Xuanwen Ding, Tianyu Huai, Shunyu Liu, Qin Chen, Yuan Xie, Liang He
Abstract: Recently, foundation language models (LMs) have marked significant achievements in the domains of natural language processing (NLP) and computer vision (CV). Unlike traditional neural network models, foundation LMs obtain a great ability for transfer learning by acquiring rich commonsense knowledge through pre-training on extensive unsupervised datasets with a vast number of parameters. However, they still can not emulate human-like continuous learning due to catastrophic forgetting. Consequently, various continual learning (CL)-based methodologies have been developed to refine LMs, enabling them to adapt to new tasks without forgetting previous knowledge. However, a systematic taxonomy of existing approaches and a comparison of their performance are still lacking, which is the gap that our survey aims to fill. We delve into a comprehensive review, summarization, and classification of the existing literature on CL-based approaches applied to foundation language models, such as pre-trained language models (PLMs), large language models (LLMs) and vision-language models (VLMs). We divide these studies into offline CL and online CL, which consist of traditional methods, parameter-efficient-based methods, instruction tuning-based methods and continual pre-training methods. Offline CL encompasses domain-incremental learning, task-incremental learning, and class-incremental learning, while online CL is subdivided into hard task boundary and blurry task boundary settings. Additionally, we outline the typical datasets and metrics employed in CL research and provide a detailed analysis of the challenges and future work for LMs-based continual learning.
Authors: Xun Liu, Zhengwei Ni
Abstract: Large language models (LLMs) exhibit impressive proficiency in natural language generation, understanding user instructions, and emulating human-like language use, which has led to significant interest in their application to role-playing scenarios. However, the manual collection of role-specific script data and the evaluation of model performance are resource-intensive processes. This project introduces a prompt-based framework designed to leverage GPT's capabilities for the generation of role-playing dialogue datasets and the evaluation of role-playing performance. To validate the effectiveness of the GPT-based generation and evaluation, we further incorporate the recall-oriented Rouge-L metric, providing an additional quantitative measure of performance.
Authors: Daixuan Cheng, Yuxian Gu, Shaohan Huang, Junyu Bi, Minlie Huang, Furu Wei
Abstract: Unsupervised multitask pre-training has been the critical method behind the recent success of language models (LMs). However, supervised multitask learning still holds significant promise, as scaling it in the post-training stage trends towards better generalization. In this paper, we explore supervised multitask pre-training by proposing Instruction Pre-Training, a framework that scalably augments massive raw corpora with instruction-response pairs to pre-train LMs. The instruction-response pairs are generated by an efficient instruction synthesizer built on open-source models. In our experiments, we synthesize 200M instruction-response pairs covering 40+ task categories to verify the effectiveness of Instruction Pre-Training. In pre-training from scratch, Instruction Pre-Training not only consistently enhances pre-trained base models but also benefits more from further instruction tuning. In continual pre-training, Instruction Pre-Training enables Llama3-8B to be comparable to or even outperform Llama3-70B. Our model, code, and data are available at https://github.com/microsoft/LMOps.
Authors: Yibo Jiang, Goutham Rajendran, Pradeep Ravikumar, Bryon Aragam
Abstract: Large Language Models (LLMs) have the capacity to store and recall facts. Through experimentation with open-source models, we observe that this ability to retrieve facts can be easily manipulated by changing contexts, even without altering their factual meanings. These findings highlight that LLMs might behave like an associative memory model where certain tokens in the contexts serve as clues to retrieving facts. We mathematically explore this property by studying how transformers, the building blocks of LLMs, can complete such memory tasks. We study a simple latent concept association problem with a one-layer transformer and we show theoretically and empirically that the transformer gathers information using self-attention and uses the value matrix for associative memory.
Authors: Mengkang Hu, Pu Zhao, Can Xu, Qingfeng Sun, Jianguang Lou, Qingwei Lin, Ping Luo, Saravan Rajmohan
Abstract: Large Language Model-based agents have garnered significant attention and are becoming increasingly popular. Furthermore, planning ability is a crucial component of an LLM-based agent, which generally entails achieving a desired goal from an initial state. This paper investigates enhancing the planning abilities of LLMs through instruction tuning, referred to as agent training. Recent studies have demonstrated that utilizing expert-level trajectory for instruction-tuning LLMs effectively enhances their planning capabilities. However, existing work primarily focuses on synthesizing trajectories from manually designed planning tasks and environments. The labor-intensive nature of creating these environments and tasks impedes the generation of sufficiently varied and extensive trajectories. To address this limitation, this paper explores the automated synthesis of diverse environments and a gradual range of planning tasks, from easy to difficult. We introduce a framework, AgentGen, that leverages LLMs first to generate environments and subsequently generate planning tasks conditioned on these environments. Specifically, to improve environmental diversity, we propose using an inspiration corpus composed of various domain-specific text segments as the context for synthesizing environments. Moreover, to increase the difficulty diversity of generated planning tasks, we propose a bidirectional evolution method, Bi-Evol, that evolves planning tasks from easier and harder directions to synthesize a task set with a smoother difficulty curve. The evaluation results derived from AgentBoard show that AgentGen greatly improves LLMs' planning ability, e.g., the AgentGen instruction-tuned Llama-3.1-8B surpasses GPT-3.5 in overall performance. Moreover, the AgentGen-tuned Llama-3.1-70B model achieves state-of-the-art results in planning tasks.
Authors: Vivi Nastase, Chunyang Jiang, Giuseppe Samo, Paola Merlo
Abstract: In this paper, our goal is to investigate to what degree multilingual pretrained language models capture cross-linguistically valid abstract linguistic representations. We take the approach of developing curated synthetic data on a large scale, with specific properties, and using them to study sentence representations built using pretrained language models. We use a new multiple-choice task and datasets, Blackbird Language Matrices (BLMs), to focus on a specific grammatical structural phenomenon -- subject-verb agreement across a variety of sentence structures -- in several languages. Finding a solution to this task requires a system detecting complex linguistic patterns and paradigms in text representations. Using a two-level architecture that solves the problem in two steps -- detect syntactic objects and their properties in individual sentences, and find patterns across an input sequence of sentences -- we show that despite having been trained on multilingual texts in a consistent manner, multilingual pretrained language models have language-specific differences, and syntactic structure is not shared, even across closely related languages.
Authors: Vivi Nastase, Giuseppe Samo, Chunyang Jiang, Paola Merlo
Abstract: We investigate to what degree existing LLMs encode abstract linguistic information in Italian in a multi-task setting. We exploit curated synthetic data on a large scale -- several Blackbird Language Matrices (BLMs) problems in Italian -- and use them to study how sentence representations built using pre-trained language models encode specific syntactic and semantic information. We use a two-level architecture to model separately a compression of the sentence embeddings into a representation that contains relevant information for a task, and a BLM task. We then investigate whether we can obtain compressed sentence representations that encode syntactic and semantic information relevant to several BLM tasks. While we expected that the sentence structure -- in terms of sequence of phrases/chunks -- and chunk properties could be shared across tasks, performance and error analysis show that the clues for the different tasks are encoded in different manners in the sentence embeddings, suggesting that abstract linguistic notions such as constituents or thematic roles does not seem to be present in the pretrained sentence embeddings.
Authors: Jiun-Ting Li, Bi-Cheng Yan, Tien-Hong Lo, Yi-Cheng Wang, Yung-Chang Hsu, Berlin Chen
Abstract: Automated speaking assessment in conversation tests (ASAC) aims to evaluate the overall speaking proficiency of an L2 (second-language) speaker in a setting where an interlocutor interacts with one or more candidates. Although prior ASAC approaches have shown promising performance on their respective datasets, there is still a dearth of research specifically focused on incorporating the coherence of the logical flow within a conversation into the grading model. To address this critical challenge, we propose a hierarchical graph model that aptly incorporates both broad inter-response interactions (e.g., discourse relations) and nuanced semantic information (e.g., semantic words and speaker intents), which is subsequently fused with contextual information for the final prediction. Extensive experimental results on the NICT-JLE benchmark dataset suggest that our proposed modeling approach can yield considerable improvements in prediction accuracy with respect to various assessment metrics, as compared to some strong baselines. This also sheds light on the importance of investigating coherence-related facets of spoken responses in ASAC.
Authors: Jiale Kang
Abstract: Low-Rank Adaptation (LoRA) has achieved remarkable training results by freezing the original weights and training only low-rank matrices, establishing itself as the predominant fine-tuning method for LLMs. Many LoRA variants have emerged, yet they lack a design tailored to the characteristics of LLM weights and fail to leverage the original weights effectively. To address the sparsity of LLM weights, and drawing inspiration from GQA and MQA, we propose Block-Affine Adaptation (Bone), a novel PEFT technique distinct from LoRA. By dividing the original weights into multiple subspaces that share a single matrix for weight updates, Bone simplifies the process by requiring the trainable matrix to be initialized to zero, eliminating the need for complex initialization as in some LoRA variants. Compared to LoRA, Bone significantly reduces memory usage and achieves faster computation. Evaluation of both NLU and NLG tasks demonstrates that Bone substantially outperforms LoRA and its variants. Inspired by Pissa, we propose a new theory called "Weight Guide" to better utilize the information embedded in the original weights. This approach extracts valuable information through a linear transformation of the original weight matrix using a trainable matrix. To validate the effectiveness of "Weight Guide" we combined it with Bone to create a new structure called Block-Affine Transformation (Bat), and ablation experiments confirmed the effectiveness of "Weight Guide".
Authors: Genta Indra Winata, David Anugraha, Lucky Susanto, Garry Kuwanto, Derry Tanti Wijaya
Abstract: Understanding the quality of a performance evaluation metric is crucial for ensuring that model outputs align with human preferences. However, it remains unclear how well each metric captures the diverse aspects of these preferences, as metrics often excel in one particular area but not across all dimensions. To address this, it is essential to systematically calibrate metrics to specific aspects of human preference, catering to the unique characteristics of each aspect. We introduce MetaMetrics, a calibrated meta-metric designed to evaluate generation tasks across different modalities in a supervised manner. MetaMetrics optimizes the combination of existing metrics to enhance their alignment with human preferences. Our metric demonstrates flexibility and effectiveness in both language and vision downstream tasks, showing significant benefits across various multilingual and multi-domain scenarios. MetaMetrics aligns closely with human preferences and is highly extendable and easily integrable into any application. This makes MetaMetrics a powerful tool for improving the evaluation of generation tasks, ensuring that metrics are more representative of human judgment across diverse contexts.
Authors: Siheng Xiong, Ali Payani, Yuan Yang, Faramarz Fekri
Abstract: Enhancing the reasoning capabilities of large language models (LLMs) remains a key challenge, especially for tasks that require complex, multi-step decision-making. Humans excel at these tasks by leveraging deliberate planning with an internal world model to simulate the potential outcomes of various actions. Inspired by this, we propose a novel multi-step reasoning framework for LLMs, referred to as Structure-aware Planning with Accurate World Model (SWAP). Unlike previous approaches that rely solely on Chain-of-Thought (CoT) reasoning in natural language, SWAP incorporates structural information to guide the reasoning process via a world model and provides a soft verification mechanism over the steps. Moreover, SWAP overcomes the challenge of accurate world state predictions in complex reasoning tasks by introducing a Generator-Discriminator architecture, which enables more reliable world modeling. Specifically, the generator predicts the next state, and the discriminator ensures alignment with the logical consistency required by the problem context. SWAP also encourages the policy model to explore a broad range of potential actions to prevent premature convergence. By resolving the bottlenecks of generation diversity for both actions and states using diversity-based modeling (DBM) and improving discrimination accuracy through contrastive ranking (CR), SWAP significantly enhances the reasoning performance of LLMs. We evaluate SWAP across diverse reasoning-intensive benchmarks including math reasoning, logical reasoning, and coding tasks. Extensive experiments demonstrate that SWAP achieves substantial improvements over the baselines and consistently outperforms existing methods.
Authors: Genta Indra Winata, Frederikus Hudi, Patrick Amadeus Irawan, David Anugraha, Rifki Afina Putri, Yutong Wang, Adam Nohejl, Ubaidillah Ariq Prathama, Nedjma Ousidhoum, Afifa Amriani, Anar Rzayev, Anirban Das, Ashmari Pramodya, Aulia Adila, Bryan Wilie, Candy Olivia Mawalim, Ching Lam Cheng, Daud Abolade, Emmanuele Chersoni, Enrico Santus, Fariz Ikhwantri, Garry Kuwanto, Hanyang Zhao, Haryo Akbarianto Wibowo, Holy Lovenia, Jan Christian Blaise Cruz, Jan Wira Gotama Putra, Junho Myung, Lucky Susanto, Maria Angelica Riera Machin, Marina Zhukova, Michael Anugraha, Muhammad Farid Adilazuarda, Natasha Santosa, Peerat Limkonchotiwat, Raj Dabre, Rio Alexander Audino, Samuel Cahyawijaya, Shi-Xiong Zhang, Stephanie Yulia Salim, Yi Zhou, Yinxuan Gui, David Ifeoluwa Adelani, En-Shiun Annie Lee, Shogo Okada, Ayu Purwarianti, Alham Fikri Aji, Taro Watanabe, Derry Tanti Wijaya, Alice Oh, Chong-Wah Ngo
Abstract: Vision Language Models (VLMs) often struggle with culture-specific knowledge, particularly in languages other than English and in underrepresented cultural contexts. To evaluate their understanding of such knowledge, we introduce WorldCuisines, a massive-scale benchmark for multilingual and multicultural, visually grounded language understanding. This benchmark includes a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects, spanning 9 language families and featuring over 1 million data points, making it the largest multicultural VQA benchmark to date. It includes tasks for identifying dish names and their origins. We provide evaluation datasets in two sizes (12k and 60k instances) alongside a training dataset (1 million instances). Our findings show that while VLMs perform better with correct location context, they struggle with adversarial contexts and predicting specific regional cuisines and languages. To support future research, we release a knowledge base with annotated food entries and images along with the VQA data.
Authors: Haein Kong, Yongsu Ahn, Sangyub Lee, Yunho Maeng
Abstract: LLMs have emerged as a promising tool for assisting individuals in diverse text-generation tasks, including job-related texts. However, LLM-generated answers have been increasingly found to exhibit gender bias. This study evaluates three LLMs (GPT-3.5, GPT-4, Claude) to conduct a multifaceted audit of LLM-generated interview responses across models, question types, and jobs, and their alignment with two gender stereotypes. Our findings reveal that gender bias is consistent, and closely aligned with gender stereotypes and the dominance of jobs. Overall, this study contributes to the systematic examination of gender bias in LLM-generated interview responses, highlighting the need for a mindful approach to mitigate such biases in related applications.
Authors: Rui Song, Yingji Li, Lida Shi, Fausto Giunchiglia, Hao Xu
Abstract: Shortcut learning refers to the phenomenon where models employ simple, non-robust decision rules in practical tasks, which hinders their generalization and robustness. With the rapid development of large language models (LLMs) in recent years, an increasing number of studies have shown the impact of shortcut learning on LLMs. This paper provides a novel perspective to review relevant research on shortcut learning in In-Context Learning (ICL). It conducts a detailed exploration of the types of shortcuts in ICL tasks, their causes, available benchmarks, and strategies for mitigating shortcuts. Based on corresponding observations, it summarizes the unresolved issues in existing research and attempts to outline the future research landscape of shortcut learning.
Authors: Jianyi Zhang, Da-Cheng Juan, Cyrus Rashtchian, Chun-Sung Ferng, Heinrich Jiang, Yiran Chen
Abstract: Large language models (LLMs) have demonstrated remarkable capabilities, but their outputs can sometimes be unreliable or factually incorrect. To address this, we introduce Self Logits Evolution Decoding (SLED), a novel decoding framework that enhances the truthfulness of LLMs without relying on external knowledge bases or requiring further fine-tuning. From an optimization perspective, our SLED framework leverages the latent knowledge embedded within the LLM by contrasting the output logits from the final layer with those from early layers. It then utilizes an approximate gradient approach to enable latent knowledge to guide the self-refinement of outputs, thereby effectively improving factual accuracy. Extensive experiments have been conducted on established benchmarks across a diverse range of model families (LLaMA 2, LLaMA 3, Gemma) and scales (from 2B to 70B), including more advanced architectural configurations such as the mixture of experts (MoE). Our evaluation spans a wide variety of tasks, including multi-choice, open-generation, and adaptations to chain-of-thought reasoning tasks. The results demonstrate that SLED consistently improves factual accuracy by up to 20\% compared to existing decoding methods while maintaining natural language fluency and negligible latency overhead. Furthermore, it can be flexibly combined with other decoding methods to further enhance their performance.
Authors: Yuxuan Hu, Ke Wang, Xiaokang Zhang, Fanjin Zhang, Cuiping Li, Hong Chen, Jing Zhang
Abstract: Large Language Models (LLMs) have revolutionized natural language processing by unifying tasks into text generation, yet their large parameter sizes and autoregressive nature limit inference speed. SAM-Decoding addresses this by introducing a novel retrieval-based speculative decoding method that uses a suffix automaton for efficient and accurate draft generation. Unlike n-gram matching used by the existing method, SAM-Decoding finds the longest suffix match in generating text and text corpuss, achieving an average time complexity of $O(1)$ per generation step. SAM-Decoding constructs static and dynamic suffix automatons for the text corpus and input prompts, respectively, enabling fast and precise draft generation. Meanwhile, it is designed as an approach that can be combined with existing methods, allowing SAM-Decoding to adaptively select a draft generation strategy based on the matching length, thus increasing the inference speed of the LLM. When combined with Token Recycling, evaluations show SAM-Decoding outperforms existing model-free methods, achieving a speedup of $2.27\times$ over autoregressive decoding on Spec-Bench. When combined with EAGLE2, it reaches a speedup of $2.49\times$, surpassing all current approaches. Our code is available at https://github.com/hyx1999/SAM-Decoding.
Authors: Anton Alekseev, Gulnara Kabaeva
Abstract: One of the key tasks in modern applied computational linguistics is constructing word vector representations (word embeddings), which are widely used to address natural language processing tasks such as sentiment analysis, information extraction, and more. To choose an appropriate method for generating these word embeddings, quality assessment techniques are often necessary. A standard approach involves calculating distances between vectors for words with expert-assessed 'similarity'. This work introduces the first 'silver standard' dataset for such tasks in the Kyrgyz language, alongside training corresponding models and validating the dataset's suitability through quality evaluation metrics.
Authors: Chenhang Cui, Gelei Deng, An Zhang, Jingnan Zheng, Yicong Li, Lianli Gao, Tianwei Zhang, Tat-Seng Chua
Abstract: Recent advances in Large Vision-Language Models (LVLMs) have showcased strong reasoning abilities across multiple modalities, achieving significant breakthroughs in various real-world applications. Despite this great success, the safety guardrail of LVLMs may not cover the unforeseen domains introduced by the visual modality. Existing studies primarily focus on eliciting LVLMs to generate harmful responses via carefully crafted image-based jailbreaks designed to bypass alignment defenses. In this study, we reveal that a safe image can be exploited to achieve the same jailbreak consequence when combined with additional safe images and prompts. This stems from two fundamental properties of LVLMs: universal reasoning capabilities and safety snowball effect. Building on these insights, we propose Safety Snowball Agent (SSA), a novel agent-based framework leveraging agents' autonomous and tool-using abilities to jailbreak LVLMs. SSA operates through two principal stages: (1) initial response generation, where tools generate or retrieve jailbreak images based on potential harmful intents, and (2) harmful snowballing, where refined subsequent prompts induce progressively harmful outputs. Our experiments demonstrate that \ours can use nearly any image to induce LVLMs to produce unsafe content, achieving high success jailbreaking rates against the latest LVLMs. Unlike prior works that exploit alignment flaws, \ours leverages the inherent properties of LVLMs, presenting a profound challenge for enforcing safety in generative multimodal systems. Our code is avaliable at \url{https://github.com/gzcch/Safety_Snowball_Agent}.
Authors: Mengfei Lan, Lecheng Zheng, Shufan Ming, Halil Kilicoglu
Abstract: Sequential sentence classification (SSC) in scientific publications is crucial for supporting downstream tasks such as fine-grained information retrieval and extractive summarization. However, current SSC methods are constrained by model size, sequence length, and single-label setting. To address these limitations, this paper proposes LLM-SSC, a large language model (LLM)-based framework for both single- and multi-label SSC tasks. Unlike previous approaches that employ small- or medium-sized language models, the proposed framework utilizes LLMs to generate SSC labels through designed prompts, which enhance task understanding by incorporating demonstrations and a query to describe the prediction target. We also present a multi-label contrastive learning loss with auto-weighting scheme, enabling the multi-label classification task. To support our multi-label SSC analysis, we introduce and release a new dataset, biorc800, which mainly contains unstructured abstracts in the biomedical domain with manual annotations. Experiments demonstrate LLM-SSC's strong performance in SSC under both in-context learning and task-specific tuning settings. We release biorc800 and our code at: https://github.com/ScienceNLP-Lab/LLM-SSC.
Authors: Shaohan Huang, Xun Wu, Shuming Ma, Furu Wei
Abstract: Multi-Head Mixture-of-Experts (MH-MoE) demonstrates superior performance by using the multi-head mechanism to collectively attend to information from various representation spaces within different experts. In this paper, we present a novel implementation of MH-MoE that maintains both FLOPs and parameter parity with sparse Mixture of Experts models. Experimental results on language models show that the new implementation yields quality improvements over both vanilla MoE and fine-grained MoE models. Additionally, our experiments demonstrate that MH-MoE is compatible with 1-bit Large Language Models (LLMs) such as BitNet.
Authors: Sanjana Ramprasad, Byron C. Wallace
Abstract: Modern LLMs can now produce highly readable abstractive summaries, to the point where traditional automated metrics for evaluating summary quality, such as ROUGE, have become saturated. However, LLMs still sometimes introduce unwanted content into summaries, i.e., information inconsistent with or unsupported by their source. Measuring the occurrence of these often subtle ``hallucinations'' automatically has proved to be challenging. This in turn has motivated development of a variety of metrics intended to measure the factual consistency of generated summaries against their source. But are these approaches measuring what they purport to do? In this work, we stress-test automatic factuality metrics. Specifically, we investigate whether and to what degree superficial attributes of summary texts suffice to predict ``factuality'', finding that a (supervised) model using only such shallow features is reasonably competitive with SOTA factuality scoring methods. We then evaluate how factuality metrics respond to factual corrections in inconsistent summaries and find that only a few show meaningful improvements. In contrast, some metrics are more sensitive to benign, non-factual edits. Motivated by these insights, we show that one can ``game'' (most) automatic factuality metrics, i.e., reliably inflate ``factuality'' scores by appending innocuous sentences to generated summaries. Taken together, our results raise questions about the degree to which we should rely on existing automated factuality metrics and what exactly we want ``factuality metrics'' to measure.
Authors: Yuhang Wang, Jitao Sang
Abstract: The o1 system card identifies the o1 models as the most robust within OpenAI, with their defining characteristic being the progression from rapid, intuitive thinking to slower, more deliberate reasoning. This observation motivated us to investigate the influence of System-2 thinking patterns on model safety. In our preliminary research, we conducted safety evaluations of the o1 model, including complex jailbreak attack scenarios using adversarial natural language prompts and mathematical encoding prompts. Our findings indicate that the o1 model demonstrates relatively improved safety performance; however, it still exhibits vulnerabilities, particularly against jailbreak attacks employing mathematical encoding. Through detailed case analysis, we identified specific patterns in the o1 model's responses. We also explored the alignment of System-2 safety in open-source models using prompt engineering and supervised fine-tuning techniques. Experimental results show that some simple methods to encourage the model to carefully scrutinize user requests are beneficial for model safety. Additionally, we proposed a implementation plan for process supervision to enhance safety alignment. The implementation details and experimental results will be provided in future versions.
Authors: Ratnesh Kumar Joshi, Priyanshu Priya, Vishesh Desai, Saurav Dudhate, Siddhant Senapati, Asif Ekbal, Roshni Ramnani, Anutosh Maitra, Shubhashis Sengupta
Abstract: Given the advancements in conversational artificial intelligence, the evaluation and assessment of Large Language Models (LLMs) play a crucial role in ensuring optimal performance across various conversational tasks. In this paper, we present a comprehensive study that thoroughly evaluates the capabilities and limitations of five prevalent LLMs: Llama, OPT, Falcon, Alpaca, and MPT. The study encompasses various conversational tasks, including reservation, empathetic response generation, mental health and legal counseling, persuasion, and negotiation. To conduct the evaluation, an extensive test setup is employed, utilizing multiple evaluation criteria that span from automatic to human evaluation. This includes using generic and task-specific metrics to gauge the LMs' performance accurately. From our evaluation, no single model emerges as universally optimal for all tasks. Instead, their performance varies significantly depending on the specific requirements of each task. While some models excel in certain tasks, they may demonstrate comparatively poorer performance in others. These findings emphasize the importance of considering task-specific requirements and characteristics when selecting the most suitable LM for conversational applications.
Authors: Jordan J. Bird
Abstract: The integration of new literature into the English curriculum remains a challenge since educators often lack scalable tools to rapidly evaluate readability and adapt texts for diverse classroom needs. This study proposes to address this gap through a multimodal approach that combines transformer-based text classification with linguistic feature analysis to align texts with UK Key Stages. Eight state-of-the-art Transformers were fine-tuned on segmented text data, with BERT achieving the highest unimodal F1 score of 0.75. In parallel, 500 deep neural network topologies were searched for the classification of linguistic characteristics, achieving an F1 score of 0.392. The fusion of these modalities shows a significant improvement, with every multimodal approach outperforming all unimodal models. In particular, the ELECTRA Transformer fused with the neural network achieved an F1 score of 0.996. Unimodal and multimodal approaches are shown to have statistically significant differences in all validation metrics (accuracy, precision, recall, F1 score) except for inference time. The proposed approach is finally encapsulated in a stakeholder-facing web application, providing non-technical stakeholder access to real-time insights on text complexity, reading difficulty, curriculum alignment, and recommendations for learning age range. The application empowers data-driven decision making and reduces manual workload by integrating AI-based recommendations into lesson planning for English literature.
Authors: Akshat Sharma, Hangliang Ding, Jianping Li, Neel Dani, Minjia Zhang
Abstract: How to efficiently serve LLMs in practice has become exceptionally challenging due to their prohibitive memory and computation requirements. In this study, we investigate optimizing the KV cache, whose memory footprint poses a critical bottleneck in LLM inference, especially when dealing with long context tasks. To tackle the challenge, we introduce MiniKV, a KV cache optimization method that simultaneously preserves long context task accuracy while significantly reducing KV cache size via a novel 2-bit layer-discriminative KV cache. More importantly, we develop specialized CUDA kernels to make MiniKV compatible with FlashAttention. Experiments on a wide range of long context tasks show that MiniKV effectively achieves 86% KV cache compression ratio while recovering over 98.5% of accuracy, outperforming state-of-the-art methods while achieving excellent measured system performance improvements.
Authors: Shukang Yin, Chaoyou Fu, Sirui Zhao, Ke Li, Xing Sun, Tong Xu, Enhong Chen
Abstract: Recently, Multimodal Large Language Model (MLLM) represented by GPT-4V has been a new rising research hotspot, which uses powerful Large Language Models (LLMs) as a brain to perform multimodal tasks. The surprising emergent capabilities of MLLM, such as writing stories based on images and OCR-free math reasoning, are rare in traditional multimodal methods, suggesting a potential path to artificial general intelligence. To this end, both academia and industry have endeavored to develop MLLMs that can compete with or even better than GPT-4V, pushing the limit of research at a surprising speed. In this paper, we aim to trace and summarize the recent progress of MLLMs. First of all, we present the basic formulation of MLLM and delineate its related concepts, including architecture, training strategy and data, as well as evaluation. Then, we introduce research topics about how MLLMs can be extended to support more granularity, modalities, languages, and scenarios. We continue with multimodal hallucination and extended techniques, including Multimodal ICL (M-ICL), Multimodal CoT (M-CoT), and LLM-Aided Visual Reasoning (LAVR). To conclude the paper, we discuss existing challenges and point out promising research directions. In light of the fact that the era of MLLM has only just begun, we will keep updating this survey and hope it can inspire more research. An associated GitHub link collecting the latest papers is available at https://github.com/BradyFU/Awesome-Multimodal-Large-Language-Models.
URLs: https://github.com/BradyFU/Awesome-Multimodal-Large-Language-Models.
Authors: Xinyue Zheng, Haowei Lin, Kaichen He, Zihao Wang, Zilong Zheng, Yitao Liang
Abstract: Evaluating generalist agents presents significant challenges due to their wide-ranging abilities and the limitations of current benchmarks in assessing true generalization. We introduce the Minecraft Universe (MCU), a fully automated benchmarking framework set within the open-world game Minecraft. MCU dynamically generates and evaluates a broad spectrum of tasks, offering three core components: 1) a task generation mechanism that provides high degrees of freedom and variability, 2) an ever-expanding set of over 3K composable atomic tasks, and 3) a general evaluation framework that supports open-ended task assessment. By integrating large language models (LLMs), MCU dynamically creates diverse environments for each evaluation, fostering agent generalization. The framework uses a vision-language model (VLM) to automatically generate evaluation criteria, achieving over 90% agreement with human ratings across multi-dimensional assessments, which demonstrates that MCU is a scalable and explainable solution for evaluating generalist agents. Additionally, we show that while state-of-the-art foundational models perform well on specific tasks, they often struggle with increased task diversity and difficulty.
Authors: Takehiko Ohkawa, Takuma Yagi, Taichi Nishimura, Ryosuke Furuta, Atsushi Hashimoto, Yoshitaka Ushiku, Yoichi Sato
Abstract: We propose a novel benchmark for cross-view knowledge transfer of dense video captioning, adapting models from web instructional videos with exocentric views to an egocentric view. While dense video captioning (predicting time segments and their captions) is primarily studied with exocentric videos (e.g., YouCook2), benchmarks with egocentric videos are restricted due to data scarcity. To overcome the limited video availability, transferring knowledge from abundant exocentric web videos is demanded as a practical approach. However, learning the correspondence between exocentric and egocentric views is difficult due to their dynamic view changes. The web videos contain shots showing either full-body or hand regions, while the egocentric view is constantly shifting. This necessitates the in-depth study of cross-view transfer under complex view changes. To this end, we first create a real-life egocentric dataset (EgoYC2) whose captions follow the definition of YouCook2 captions, enabling transfer learning between these datasets with access to their ground-truth. To bridge the view gaps, we propose a view-invariant learning method using adversarial training, which consists of pre-training and fine-tuning stages. Our experiments confirm the effectiveness of overcoming the view change problem and knowledge transfer to egocentric views. Our benchmark pushes the study of cross-view transfer into a new task domain of dense video captioning and envisions methodologies that describe egocentric videos in natural language.
Authors: Jonathan F\"urst, Catherine Kosten, Farhad Nooralahzadeh, Yi Zhang, Kurt Stockinger
Abstract: Text-to-SQL systems (also known as NL-to-SQL systems) have become an increasingly popular solution for bridging the gap between user capabilities and SQL-based data access. These systems translate user requests in natural language to valid SQL statements for a specific database. Recent Text-to-SQL systems have benefited from the rapid improvement of transformer-based language models. However, while Text-to-SQL systems that incorporate such models continuously reach new high scores on -- often synthetic -- benchmark datasets, a systematic exploration of their robustness towards different data models in a real-world, realistic scenario is notably missing. This paper provides the first in-depth evaluation of the data model robustness of Text-to-SQL systems in practice based on a multi-year international project focused on Text-to-SQL interfaces. Our evaluation is based on a real-world deployment of FootballDB, a system that was deployed over a 9 month period in the context of the FIFA World Cup 2022, during which about 6K natural language questions were asked and executed. All of our data is based on real user questions that were asked live to the system. We manually labeled and translated a subset of these questions for three different data models. For each data model, we explore the performance of representative Text-to-SQL systems and language models. We further quantify the impact of training data size, pre-, and post-processing steps as well as language model inference time. Our comprehensive evaluation sheds light on the design choices of real-world Text-to-SQL systems and their impact on moving from research prototypes to real deployments. Last, we provide a new benchmark dataset to the community, which is the first to enable the evaluation of different data models for the same dataset and is substantially more challenging than most previous datasets in terms of query complexity.
Authors: Yueen Ma, Zixing Song, Yuzheng Zhuang, Jianye Hao, Irwin King
Abstract: Deep learning has demonstrated remarkable success across many domains, including computer vision, natural language processing, and reinforcement learning. Representative artificial neural networks in these fields span convolutional neural networks, Transformers, and deep Q-networks. Built upon unimodal neural networks, numerous multi-modal models have been introduced to address a range of tasks such as visual question answering, image captioning, and speech recognition. The rise of instruction-following robotic policies in embodied AI has spurred the development of a novel category of multi-modal models known as vision-language-action models (VLAs). Their multi-modality capability has become a foundational element in robot learning. Various methods have been proposed to enhance traits such as versatility, dexterity, and generalizability. Some models focus on refining specific components. Others aim to develop control policies adept at predicting low-level actions. Certain VLAs serve as high-level task planners capable of decomposing long-horizon tasks into executable subtasks. Over the past few years, a myriad of VLAs have emerged, reflecting the rapid advancement of embodied AI. Therefore, it is imperative to capture the evolving landscape through a comprehensive survey.
Authors: Laurens Samson, Nimrod Barazani, Sennay Ghebreab, Yuki M. Asano
Abstract: This paper aims to advance our understanding of how Visual Language Models (VLMs) handle privacy-sensitive information, a crucial concern as these technologies become integral to everyday life. To this end, we introduce a new benchmark PrivBench, which contains images from 8 sensitive categories such as passports, or fingerprints. We evaluate 10 state-of-the-art VLMs on this benchmark and observe a generally limited understanding of privacy, highlighting a significant area for model improvement. Based on this we introduce PrivTune, a new instruction-tuning dataset aimed at equipping VLMs with knowledge about visual privacy. By tuning two pretrained VLMs, TinyLLaVa and MiniGPT-v2, on this small dataset, we achieve strong gains in their ability to recognize sensitive content, outperforming even GPT4-V. At the same time, we show that privacy-tuning only minimally affects the VLMs performance on standard benchmarks such as VQA. Overall, this paper lays out a crucial challenge for making VLMs effective in handling real-world data safely and provides a simple recipe that takes the first step towards building privacy-aware VLMs.
Authors: Wei Wang, Qing Li
Abstract: Language models have emerged as a critical area of focus in artificial intelligence, particularly with the introduction of groundbreaking innovations like ChatGPT. Large-scale Transformer networks have quickly become the leading approach for advancing natural language processing algorithms. Built on the Transformer architecture, these models enable interactions that closely mimic human communication and, equipped with extensive knowledge, can even assist in guiding human tasks. Despite their impressive capabilities and growing complexity, a key question remains-the theoretical foundations of large language models (LLMs). What makes Transformer so effective for powering intelligent language applications, such as translation and coding? What underlies LLMs' ability for In-Context Learning (ICL)? How does the LoRA scheme enhance the fine-tuning of LLMs? And what supports the practicality of pruning LLMs? To address these critical questions and explore the technological strategies within LLMs, we leverage the Universal Approximation Theory (UAT) to offer a theoretical backdrop, shedding light on the mechanisms that underpin these advancements.
Authors: Zesen Liu, Tianshuo Cong, Xinlei He, Qi Li
Abstract: Large Language Models (LLMs) excel in various applications, including text generation and complex tasks. However, the misuse of LLMs raises concerns about the authenticity and ethical implications of the content they produce, such as deepfake news, academic fraud, and copyright infringement. Watermarking techniques, which embed identifiable markers in machine-generated text, offer a promising solution to these issues by allowing for content verification and origin tracing. Unfortunately, the robustness of current LLM watermarking schemes under potential watermark removal attacks has not been comprehensively explored. In this paper, to fill this gap, we first systematically comb the mainstream watermarking schemes and removal attacks on machine-generated texts, and then we categorize them into pre-text (before text generation) and post-text (after text generation) classes so that we can conduct diversified analyses. In our experiments, we evaluate eight watermarks (five pre-text, three post-text) and twelve attacks (two pre-text, ten post-text) across 87 scenarios. Evaluation results indicate that (1) KGW and Exponential watermarks offer high text quality and watermark retention but remain vulnerable to most attacks; (2) Post-text attacks are found to be more efficient and practical than pre-text attacks; (3) Pre-text watermarks are generally more imperceptible, as they do not alter text fluency, unlike post-text watermarks; (4) Additionally, combined attack methods can significantly increase effectiveness, highlighting the need for more robust watermarking solutions. Our study underscores the vulnerabilities of current techniques and the necessity for developing more resilient schemes.
Authors: Yulei Qin, Yuncheng Yang, Pengcheng Guo, Gang Li, Hang Shao, Yuchen Shi, Zihan Xu, Yun Gu, Ke Li, Xing Sun
Abstract: Instruction tuning plays a critical role in aligning large language models (LLMs) with human preference. Despite the vast amount of open instruction datasets, naively training a LLM on all existing instructions may not be optimal and practical. To pinpoint the most beneficial datapoints, data assessment and selection methods have been proposed in the fields of natural language processing (NLP) and deep learning. However, under the context of instruction tuning, there still exists a gap in knowledge on what kind of data evaluation metrics can be employed and how they can be integrated into the selection mechanism. To bridge this gap, we present a comprehensive review on existing literature of data assessment and selection especially for instruction tuning of LLMs. We systematically categorize all applicable methods into quality-based, diversity-based, and importance-based ones where a unified, fine-grained taxonomy is structured. For each category, representative methods are elaborated to describe the landscape of relevant research. In addition, comparison between the latest methods is conducted on their officially reported results to provide in-depth discussions on their limitations. Finally, we summarize the open challenges and propose the promosing avenues for future studies. All related contents are available at https://github.com/yuleiqin/fantastic-data-engineering.
URLs: https://github.com/yuleiqin/fantastic-data-engineering.
Authors: John Burden, Manuel Cebrian, Jose Hernandez-Orallo
Abstract: Large Language Models (LLMs) present a dual-use dilemma: they enable beneficial applications while harboring potential for harm, particularly through conversational interactions. Despite various safeguards, advanced LLMs remain vulnerable. A watershed case in early 2023 involved journalist Kevin Roose's extended dialogue with Bing, an LLM-powered search engine, which revealed harmful outputs after probing questions, highlighting vulnerabilities in the model's safeguards. This contrasts with simpler early jailbreaks, like the "Grandma Jailbreak," where users framed requests as innocent help for a grandmother, easily eliciting similar content. This raises the question: How much conversational effort is needed to elicit harmful information from LLMs? We propose two measures to quantify this effort: Conversational Length (CL), which measures the number of conversational turns needed to obtain a specific harmful response, and Conversational Complexity (CC), defined as the Kolmogorov complexity of the user's instruction sequence leading to the harmful response. To address the incomputability of Kolmogorov complexity, we approximate CC using a reference LLM to estimate the compressibility of the user instructions. Applying this approach to a large red-teaming dataset, we perform a quantitative analysis examining the statistical distribution of harmful and harmless conversational lengths and complexities. Our empirical findings suggest that this distributional analysis and the minimization of CC serve as valuable tools for understanding AI safety, offering insights into the accessibility of harmful information. This work establishes a foundation for a new perspective on LLM safety, centered around the algorithmic complexity of pathways to harm.
Authors: Wei Liu, Yang Bai, Chengcheng Han, Rongxiang Weng, Jun Xu, Xuezhi Cao, Jingang Wang, Xunliang Cai
Abstract: Direct Preference Optimization (DPO) is widely utilized in the Reinforcement Learning from Human Feedback (RLHF) phase to align Large Language Models (LLMs) with human preferences, thereby enhancing both their harmlessness and efficacy. However, it has been observed that DPO tends to over-optimize for verbosity, which can detrimentally affect both performance and user experience. In this paper, we conduct an in-depth theoretical analysis of DPO's optimization objective and reveal a strong correlation between its implicit reward and data length. This correlation misguides the optimization direction, resulting in length sensitivity during the DPO training and leading to verbosity. To address this issue, we propose a length-desensitization improvement method for DPO, termed LD-DPO. The proposed method aims to desensitize DPO to data length by decoupling explicit length preference, which is relatively insignificant, from the other implicit preferences, thereby enabling more effective learning of the intrinsic preferences. We utilized two settings (Base and Instruct) of Llama2-13B, Llama3-8B, and Qwen2-7B for experimental validation on various benchmarks including MT-Bench and AlpacaEval 2. The experimental results indicate that LD-DPO consistently outperforms DPO and other baseline methods, achieving more concise responses with a 10-40% reduction in length compared to DPO. We conducted in-depth experimental analyses to demonstrate that LD-DPO can indeed achieve length desensitization and align the model more closely with human-like preferences.
Authors: In Gim, Caihua Li, Lin Zhong
Abstract: Our work tackles the challenge of securing user inputs in cloud-hosted large language model (LLM) serving while ensuring output invariance, model confidentiality, and compute efficiency. We introduce secure multi-party decoding (SMD), which leverages confidential computing to confine user prompts to a trusted execution environment (TEE), namely a confidential virtual machine (CVM), while allowing service providers to generate tokens efficiently. We also introduce a novel cryptographic method, prompt obfuscation (PO), to ensure robustness against reconstruction attacks on SMD. We demonstrate that our approach preserves both prompt confidentiality and LLM serving efficiency. Our solution can enable privacy-preserving cloud LLM serving that handles sensitive prompts, such as clinical records, financial data, and personal information.
Authors: Defa Zhu, Hongzhi Huang, Zihao Huang, Yutao Zeng, Yunyao Mao, Banggu Wu, Qiyang Min, Xun Zhou
Abstract: We present hyper-connections, a simple yet effective method that can serve as an alternative to residual connections. This approach specifically addresses common drawbacks observed in residual connection variants, such as the seesaw effect between gradient vanishing and representation collapse. Theoretically, hyper-connections allow the network to adjust the strength of connections between features at different depths and dynamically rearrange layers. We conduct experiments focusing on the pre-training of large language models, including dense and sparse models, where hyper-connections show significant performance improvements over residual connections. Additional experiments conducted on vision tasks also demonstrate similar improvements. We anticipate that this method will be broadly applicable and beneficial across a wide range of AI problems.
Authors: Kamesh R
Abstract: Pretrained large language models (LLMs) are increasingly utilized across a wide range of natural language processing (NLP) tasks due to their impressive capabilities as few-shot learners. Recent techniques, such as chain-of-thought (CoT) prompting, have significantly advanced multi-step reasoning by introducing step-by-step decomposition, achieving state-of-the-art results on complex reasoning benchmarks. However, these approaches often rely on static prompting templates that do not adapt to task complexity or errors during the reasoning process. In this work, we introduce Adaptive Prompting, a dynamic and iterative framework designed to enhance reasoning by incorporating real-time adjustments to prompt structures and validation mechanisms.Experimental results demonstrate that Adaptive Prompting significantly improves performance on diverse reasoning benchmarks, including arithmetic reasoning (GSM8K, MultiArith), logical reasoning and commonsense tasks, achieving substantial accuracy gains compared to static prompting baselines. By integrating guided prompts, intermediate validation, and self-corrective steps, our approach enables smaller models to achieve competitive performance with larger counterparts, such as GPT-4, while maintaining computational efficiency. The framework achieves this without requiring fine-tuning or task-specific training data, highlighting the untapped potential of iterative reasoning methods.
Authors: Raghav Singhal, Kaustubh Ponkshe, Praneeth Vepakomma
Abstract: Low-Rank Adaptation (LoRA) is a popular technique for efficient fine-tuning of foundation models. However, applying LoRA in federated learning environments, where data is distributed across multiple clients, presents unique challenges. Existing methods rely on traditional federated averaging of LoRA adapters, resulting in inexact updates. To address this, we propose Federated Exact LoRA, or FedExLoRA, which adds a residual error term to the pretrained frozen weight matrix. Our approach achieves exact updates with minimal computational and communication overhead, preserving LoRA's efficiency. We evaluate the method on various models across arithmetic reasoning, commonsense reasoning, natural language understanding and natural language generation tasks, showing consistent performance gains over state-of-the-art methods across multiple settings. Through extensive analysis, we quantify that the deviations in updates from the ideal solution are significant, highlighting the need for exact aggregation. Our method's simplicity, efficiency, and broad applicability position it as a promising solution for accurate and effective federated fine-tuning of foundation models. Our code is publicly available at https://github.com/RaghavSinghal10/fedex-lora.
Authors: Kanan Mahammadli, Seyda Bolelli Ertekin
Abstract: This study introduces SLLMBO, an innovative framework that leverages Large Language Models (LLMs) for hyperparameter optimization (HPO), incorporating dynamic search space adaptability, enhanced parameter landscape exploitation, and a hybrid, novel LLM-Tree-structured Parzen Estimator (LLM-TPE) sampler. By addressing limitations in recent fully LLM-based methods and traditional Bayesian Optimization (BO), SLLMBO achieves more robust optimization. This comprehensive benchmarking evaluates multiple LLMs, including GPT-3.5-turbo, GPT-4o, Claude-Sonnet-3.5, and Gemini-1.5-flash, extending prior work beyond GPT-3.5 and GPT-4 and establishing SLLMBO as the first framework to benchmark a diverse set of LLMs for HPO. By integrating LLMs' established strengths in parameter initialization with the exploitation abilities demonstrated in this study, alongside TPE's exploration capabilities, the LLM-TPE sampler achieves a balanced exploration-exploitation trade-off, reduces API costs, and mitigates premature early stoppings for more effective parameter searches. Across 14 tabular tasks in classification and regression, the LLM-TPE sampler outperformed fully LLM-based methods and achieved superior results over BO methods in 9 tasks. Testing early stopping in budget-constrained scenarios further demonstrated competitive performance, indicating that LLM-based methods generally benefit from extended iterations for optimal results. This work lays the foundation for future research exploring open-source LLMs, reproducibility of LLM results in HPO, and benchmarking SLLMBO on complex datasets, such as image classification, segmentation, and machine translation.
Authors: Tianyi Qiu
Abstract: Social choice theory is the study of preference aggregation across a population, used both in mechanism design for human agents and in the democratic alignment of language models. In this study, we propose the representative social choice framework for the modeling of democratic representation in collective decisions, where the number of issues and individuals are too large for mechanisms to consider all preferences directly. These scenarios are widespread in real-world decision-making processes, such as jury trials, indirect elections, legislation processes, corporate governance, and, more recently, language model alignment. In representative social choice, the population is represented by a finite sample of individual-issue pairs based on which social choice decisions are made. We show that many of the deepest questions in representative social choice can be naturally formulated as statistical learning problems, and prove the generalization properties of social choice mechanisms using the theory of machine learning. We further formulate axioms for representative social choice, and prove Arrow-like impossibility theorems with new combinatorial tools of analysis. Our framework introduces the representative approach to social choice, opening up research directions at the intersection of social choice, learning theory, and AI alignment.
Authors: Xiong Wang, Yangze Li, Chaoyou Fu, Yunhang Shen, Lei Xie, Ke Li, Xing Sun, Long Ma
Abstract: Rapidly developing large language models (LLMs) have brought tremendous intelligent applications. Especially, the GPT-4o's excellent duplex speech interaction ability has brought impressive experience to users. Researchers have recently proposed several multi-modal LLMs in this direction that can achieve user-agent speech-to-speech conversations. This paper proposes a novel speech-text multimodal LLM architecture called Freeze-Omni. Our main contribution is that the speech input and output modalities can be easily connected to a textual LLM while keeping the LLM's parameters frozen throughout the training process. We design a three-stage training strategy for modeling both the speech input and output, enabling Freeze-Omni to obtain speech-to-speech conversation ability using text-speech paired data (such as ASR and TTS data) and only 60,000 multi-round text Q&A data on 8 GPUs. Moreover, we can effectively ensure that the intelligence of the Freeze-Omni in the speech modality is at the same level compared with that in the text modality of its backbone LLM, while achieving low latency end-to-end spoken response. In addition, we also designed a method to achieve duplex dialogue ability through multi-task training, giving Freeze-Omni a more natural style of dialogue ability between users and agents. In summary, Freeze-Omni holds great potential to conduct speech-to-speech dialogue based on a multimodal LLM under the condition of a frozen LLM, avoiding the catastrophic forgetting problem caused by limited data and training resources.
Authors: Jiawei Li, Xiaoang Xu, Yang Gao
Abstract: Model evolution enables learning from feedback to refine experiences and update skills, transforming models from having no domain knowledge to becoming domain experts. However, there is currently no unified and effective method for guiding this evolutionary process. To address this gap, we propose the Meteor method, which includes three training phases: weak-to-strong data distillation, iterative training, and self-evolution strategies. Each phase maximizes the model's inherent domain capabilities, allowing it to autonomously refine its domain knowledge and enhance performance. Experiments demonstrate that our approach significantly improves accuracy, completeness, relevance, coherence, and reliability across domain-specific tasks.
Authors: Anthony Nguyen
Abstract: Graphical User Interface (GUI) grounding plays a crucial role in enhancing the capabilities of Vision-Language Model (VLM) agents. While general VLMs, such as GPT-4V, demonstrate strong performance across various tasks, their proficiency in GUI grounding remains suboptimal. Recent studies have focused on fine-tuning these models specifically for one-shot GUI grounding, yielding significant improvements over baseline performance. We introduce a visual prompting framework that employs an iterative narrowing mechanism to improve the performance of both general and fine-tuned models in GUI grounding by up to 61%. For evaluation, we tested our method on a comprehensive benchmark comprising various UI platforms and provided the code to reproduce our results.
Authors: Libo Wang
Abstract: As the application of large language models continues to expand in various fields, it poses higher challenges to the effectiveness of identifying harmful content generation and guardrail mechanisms. This research aims to evaluate the guardrail effectiveness of GPT-4o, Grok-2 Beta, Llama 3.1 (405B), Gemini 1.5, and Claude 3.5 Sonnet through black-box testing of seemingly ethical multi-step jailbreak prompts. It conducts ethical attacks by designing an identical multi-step prompts that simulates the scenario of "corporate middle managers competing for promotions." The data results show that the guardrails of the above-mentioned LLMs were bypassed and the content of verbal attacks was generated. Claude 3.5 Sonnet's resistance to multi-step jailbreak prompts is more obvious. To ensure objectivity, the experimental process, black box test code, and enhanced guardrail code are uploaded to the GitHub repository: https://github.com/brucewang123456789/GeniusTrail.git.
URLs: https://github.com/brucewang123456789/GeniusTrail.git.
Authors: Chaoyun Zhang, Shilin He, Jiaxu Qian, Bowen Li, Liqun Li, Si Qin, Yu Kang, Minghua Ma, Guyue Liu, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, Qi Zhang
Abstract: GUIs have long been central to human-computer interaction, providing an intuitive and visually-driven way to access and interact with digital systems. The advent of LLMs, particularly multimodal models, has ushered in a new era of GUI automation. They have demonstrated exceptional capabilities in natural language understanding, code generation, and visual processing. This has paved the way for a new generation of LLM-brained GUI agents capable of interpreting complex GUI elements and autonomously executing actions based on natural language instructions. These agents represent a paradigm shift, enabling users to perform intricate, multi-step tasks through simple conversational commands. Their applications span across web navigation, mobile app interactions, and desktop automation, offering a transformative user experience that revolutionizes how individuals interact with software. This emerging field is rapidly advancing, with significant progress in both research and industry. To provide a structured understanding of this trend, this paper presents a comprehensive survey of LLM-brained GUI agents, exploring their historical evolution, core components, and advanced techniques. We address research questions such as existing GUI agent frameworks, the collection and utilization of data for training specialized GUI agents, the development of large action models tailored for GUI tasks, and the evaluation metrics and benchmarks necessary to assess their effectiveness. Additionally, we examine emerging applications powered by these agents. Through a detailed analysis, this survey identifies key research gaps and outlines a roadmap for future advancements in the field. By consolidating foundational knowledge and state-of-the-art developments, this work aims to guide both researchers and practitioners in overcoming challenges and unlocking the full potential of LLM-brained GUI agents.