Authors: Chien-Kun Huang, Yi-Ting Chang, Lun-Wei Ku, Cheng-Te Li, Hong-Han Shuai
Abstract: This paper provides an overview of the Fake-EmoReact 2021 Challenge, held at the 9th SocialNLP Workshop, in conjunction with NAACL 2021. The challenge requires predicting the authenticity of tweets using reply context and augmented GIF categories from EmotionGIF dataset. We offer the Fake-EmoReact dataset with more than 453k as the experimental materials, where every tweet is labeled with authenticity. Twenty-four teams registered to participate in this challenge, and 5 submitted their results successfully in the evaluation phase. The best team achieves 93.9 on Fake-EmoReact 2021 dataset using F1 score. In addition, we show the definition of share task, data collection, and the teams' performance that joined this challenge and their approaches.
Authors: Tejaswini Pedapati, Amit Dhurandhar, Soumya Ghosh, Soham Dan, Prasanna Sattigeri
Abstract: Estimating uncertainty or confidence in the responses of a model can be significant in evaluating trust not only in the responses, but also in the model as a whole. In this paper, we explore the problem of estimating confidence for responses of large language models (LLMs) with simply black-box or query access to them. We propose a simple and extensible framework where, we engineer novel features and train a (interpretable) model (viz. logistic regression) on these features to estimate the confidence. We empirically demonstrate that our simple framework is effective in estimating confidence of flan-ul2, llama-13b and mistral-7b with it consistently outperforming existing black-box confidence estimation approaches on benchmark datasets such as TriviaQA, SQuAD, CoQA and Natural Questions by even over $10\%$ (on AUROC) in some cases. Additionally, our interpretable approach provides insight into features that are predictive of confidence, leading to the interesting and useful discovery that our confidence models built for one LLM generalize zero-shot across others on a given dataset.
Authors: Wei Pang, Chuan Zhou, Xiao-Hua Zhou, Xiaojie Wang
Abstract: Instruction Fine-Tuning, a method enhancing pre-trained language models' capabilities from mere next-word prediction to complex instruction following, often employs a one-off training approach on diverse instruction dataset. However, this method may not effectively enhance models' adherence to instructions due to the simultaneous handling of varying instruction complexities. To address this, we propose a novel phased instruction fine-tuning (Phased IFT) method, grounded in the hypothesis of progressive alignment, which posits that the transition of a pre-trained language model from simple next-word prediction to sophisticated instruction following is a gradual learning process. Specifically, we obtain the score of difficulty for each instruction via GPT-4, stratify the instruction data into subsets of increasing difficulty, and sequentially uptrain on these subsets using the standard supervised loss. Through extensive experiments on the pre-trained models Llama-2 7B/13B, and Mistral-7B using the 52K Alpaca instruction data, we demonstrate that Phased IFT significantly surpasses traditional one-off instruction fine-tuning (One-off IFT) method in win rate, empirically validating the progressive alignment hypothesis. Our findings suggest that Phased IFT offers a simple yet effective pathway for elevating the instruction-following capabilities of pre-trained language models. Models and datasets from our experiments are freely available at https://github.com/xubuvd/PhasedSFT.
Authors: Salomon Kabongo, Jennifer D'Souza, S\"oren Auer
Abstract: The rapid advancements in Large Language Models (LLMs) have opened new avenues for automating complex tasks in AI research. This paper investigates the efficacy of different LLMs-Mistral 7B, Llama-2, GPT-4-Turbo and GPT-4.o in extracting leaderboard information from empirical AI research articles. We explore three types of contextual inputs to the models: DocTAET (Document Title, Abstract, Experimental Setup, and Tabular Information), DocREC (Results, Experiments, and Conclusions), and DocFULL (entire document). Our comprehensive study evaluates the performance of these models in generating (Task, Dataset, Metric, Score) quadruples from research papers. The findings reveal significant insights into the strengths and limitations of each model and context type, providing valuable guidance for future AI research automation efforts.
Authors: Jianchao Ji, Yutong Chen, Mingyu Jin, Wujiang Xu, Wenyue Hua, Yongfeng Zhang
Abstract: In the rapidly evolving field of artificial intelligence, large language models (LLMs) have emerged as powerful tools for a myriad of applications, from natural language processing to decision-making support systems. However, as these models become increasingly integrated into societal frameworks, the imperative to ensure they operate within ethical and moral boundaries has never been more critical. This paper introduces a novel benchmark designed to measure and compare the moral reasoning capabilities of LLMs. We present the first comprehensive dataset specifically curated to probe the moral dimensions of LLM outputs, addressing a wide range of ethical dilemmas and scenarios reflective of real-world complexities. The main contribution of this work lies in the development of benchmark datasets and metrics for assessing the moral identity of LLMs, which accounts for nuance, contextual sensitivity, and alignment with human ethical standards. Our methodology involves a multi-faceted approach, combining quantitative analysis with qualitative insights from ethics scholars to ensure a thorough evaluation of model performance. By applying our benchmark across several leading LLMs, we uncover significant variations in moral reasoning capabilities of different models. These findings highlight the importance of considering moral reasoning in the development and evaluation of LLMs, as well as the need for ongoing research to address the biases and limitations uncovered in our study. We publicly release the benchmark at https://drive.google.com/drive/u/0/folders/1k93YZJserYc2CkqP8d4B3M3sgd3kA8W7 and also open-source the code of the project at https://github.com/agiresearch/MoralBench.
URLs: https://drive.google.com/drive/u/0/folders/1k93YZJserYc2CkqP8d4B3M3sgd3kA8W7, https://github.com/agiresearch/MoralBench.
Authors: Wazib Ansar, Saptarsi Goswami, Amlan Chakrabarti
Abstract: One of the principal objectives of Natural Language Processing (NLP) is to generate meaningful representations from text. Improving the informativeness of the representations has led to a tremendous rise in the dimensionality and the memory footprint. It leads to a cascading effect amplifying the complexity of the downstream model by increasing its parameters. The available techniques cannot be applied to cross-modal applications such as text-to-image. To ameliorate these issues, a novel Text-to-Image methodology for generating fixed-length representations through a self-supervised Variational Auto-Encoder (VAE) for semantic evaluation applying transformers (TexIm FAST) has been proposed in this paper. The pictorial representations allow oblivious inference while retaining the linguistic intricacies, and are potent in cross-modal applications. TexIm FAST deals with variable-length sequences and generates fixed-length representations with over 75% reduced memory footprint. It enhances the efficiency of the models for downstream tasks by reducing its parameters. The efficacy of TexIm FAST has been extensively analyzed for the task of Semantic Textual Similarity (STS) upon the MSRPC, CNN/ Daily Mail, and XSum data-sets. The results demonstrate 6% improvement in accuracy compared to the baseline and showcase its exceptional ability to compare disparate length sequences such as a text with its summary.
Authors: Shruthi Bannur, Kenza Bouzid, Daniel C. Castro, Anton Schwaighofer, Sam Bond-Taylor, Maximilian Ilse, Fernando P\'erez-Garc\'ia, Valentina Salvatelli, Harshita Sharma, Felix Meissen, Mercy Ranjit, Shaury Srivastav, Julia Gong, Fabian Falck, Ozan Oktay, Anja Thieme, Matthew P. Lungren, Maria Teodora Wetscherek, Javier Alvarez-Valle, Stephanie L. Hyland
Abstract: Radiology reporting is a complex task that requires detailed image understanding, integration of multiple inputs, including comparison with prior imaging, and precise language generation. This makes it ideal for the development and use of generative multimodal models. Here, we extend report generation to include the localisation of individual findings on the image - a task we call grounded report generation. Prior work indicates that grounding is important for clarifying image understanding and interpreting AI-generated text. Therefore, grounded reporting stands to improve the utility and transparency of automated report drafting. To enable evaluation of grounded reporting, we propose a novel evaluation framework - RadFact - leveraging the reasoning capabilities of large language models (LLMs). RadFact assesses the factuality of individual generated sentences, as well as correctness of generated spatial localisations when present. We introduce MAIRA-2, a large multimodal model combining a radiology-specific image encoder with a LLM, and trained for the new task of grounded report generation on chest X-rays. MAIRA-2 uses more comprehensive inputs than explored previously: the current frontal image, the current lateral image, the prior frontal image and prior report, as well as the Indication, Technique and Comparison sections of the current report. We demonstrate that these additions significantly improve report quality and reduce hallucinations, establishing a new state of the art on findings generation (without grounding) on MIMIC-CXR while demonstrating the feasibility of grounded reporting as a novel and richer task.
Authors: Shang Zhou, Feng Yao, Chengyu Dong, Zihan Wang, Jingbo Shang
Abstract: Controlling the attribute intensity of text generation is crucial across scenarios (e.g., writing conciseness, chatting emotion, and explanation clarity). The remarkable capabilities of large language models (LLMs) have revolutionized text generation, prompting us to explore such \emph{smooth control} of LLM generation. Specifically, we propose metrics to assess the range, calibration, and consistency of the generated text's attribute intensity in response to varying control values, as well as its relevance to the intended context. To quantify the attribute intensity and context relevance, we propose an effective evaluation framework leveraging the Elo rating system and GPT4, both renowned for their robust alignment with human judgment. We look into two viable training-free methods for achieving smooth control of LLMs: (1) Prompting with semantic shifters, and (2) Modifying internal model representations. The evaluations of these two methods are conducted on $5$ different attributes with various models. Our code and dataset can be obtained from \url{https://github.com/ShangDataLab/Smooth-Control}.
Authors: Wanqiu Long, N. Siddharth, Bonnie Webber
Abstract: Discourse relations play a pivotal role in establishing coherence within textual content, uniting sentences and clauses into a cohesive narrative. The Penn Discourse Treebank (PDTB) stands as one of the most extensively utilized datasets in this domain. In PDTB-3, the annotators can assign multiple labels to an example, when they believe that multiple relations are present. Prior research in discourse relation recognition has treated these instances as separate examples during training, and only one example needs to have its label predicted correctly for the instance to be judged as correct. However, this approach is inadequate, as it fails to account for the interdependence of labels in real-world contexts and to distinguish between cases where only one sense relation holds and cases where multiple relations hold simultaneously. In our work, we address this challenge by exploring various multi-label classification frameworks to handle implicit discourse relation recognition. We show that multi-label classification methods don't depress performance for single-label prediction. Additionally, we give comprehensive analysis of results and data. Our work contributes to advancing the understanding and application of discourse relations and provide a foundation for the future study
Authors: Tianrong Zhang, Zhaohan Xi, Ting Wang, Prasenjit Mitra, Jinghui Chen
Abstract: Pre-trained language models (PLMs) have attracted enormous attention over the past few years with their unparalleled performances. Meanwhile, the soaring cost to train PLMs as well as their amazing generalizability have jointly contributed to few-shot fine-tuning and prompting as the most popular training paradigms for natural language processing (NLP) models. Nevertheless, existing studies have shown that these NLP models can be backdoored such that model behavior is manipulated when trigger tokens are presented. In this paper, we propose PromptFix, a novel backdoor mitigation strategy for NLP models via adversarial prompt-tuning in few-shot settings. Unlike existing NLP backdoor removal methods, which rely on accurate trigger inversion and subsequent model fine-tuning, PromptFix keeps the model parameters intact and only utilizes two extra sets of soft tokens which approximate the trigger and counteract it respectively. The use of soft tokens and adversarial optimization eliminates the need to enumerate possible backdoor configurations and enables an adaptive balance between trigger finding and preservation of performance. Experiments with various backdoor attacks validate the effectiveness of the proposed method and the performances when domain shift is present further shows PromptFix's applicability to models pretrained on unknown data source which is the common case in prompt tuning scenarios.
Authors: Claire Jin, Sudha Rao, Xiangyu Peng, Portia Botchway, Jessica Quaye, Chris Brockett, Bill Dolan
Abstract: Advancements in large language models (LLMs) are revolutionizing interactive game design, enabling dynamic plotlines and interactions between players and non-player characters (NPCs). However, LLMs may exhibit flaws such as hallucinations, forgetfulness, or misinterpretations of prompts, causing logical inconsistencies and unexpected deviations from intended designs. Automated techniques for detecting such game bugs are still lacking. To address this, we propose a systematic LLM-based method for automatically identifying such bugs from player game logs, eliminating the need for collecting additional data such as post-play surveys. Applied to a text-based game DejaBoom!, our approach effectively identifies bugs inherent in LLM-powered interactive games, surpassing unstructured LLM-powered bug-catching methods and filling the gap in automated detection of logical and design flaws.
Authors: Xiou Ge, Ali Mousavi, Edouard Grave, Armand Joulin, Kun Qian, Benjamin Han, Mostafa Arefiyan, Yunyao Li
Abstract: Large Language Models (LLMs) have demonstrated impressive capability in different tasks and are bringing transformative changes to many domains. However, keeping the knowledge in LLMs up-to-date remains a challenge once pretraining is complete. It is thus essential to design effective methods to both update obsolete knowledge and induce new knowledge into LLMs. Existing locate-and-edit knowledge editing (KE) method suffers from two limitations. First, the post-edit LLMs by such methods generally have poor capability in answering complex queries that require multi-hop reasoning. Second, the long run-time of such locate-and-edit methods to perform knowledge edits make it infeasible for large scale KE in practice. In this paper, we explore Parameter-Efficient Fine-Tuning (PEFT) techniques as an alternative for KE. We curate a more comprehensive temporal KE dataset with both knowledge update and knowledge injection examples for KE performance benchmarking. We further probe the effect of fine-tuning on a range of layers in an LLM for the multi-hop QA task. We find that PEFT performs better than locate-and-edit techniques for time-sensitive knowledge edits.
Authors: Abdul Waheed, Karima Kadaoui, Muhammad Abdul-Mageed
Abstract: Arabic is known to present unique challenges for Automatic Speech Recognition (ASR). On one hand, its rich linguistic diversity and wide range of dialects complicate the development of robust, inclusive models. On the other, current multilingual ASR models are compute-intensive and lack proper comprehensive evaluations. In light of these challenges, we distill knowledge from large teacher models into smaller student variants that are more efficient. We also introduce a novel human-annotated dataset covering five under-represented Arabic dialects for evaluation. We further evaluate both our models and existing SoTA multilingual models on both standard available benchmarks and our new dialectal data. Our best-distilled model's overall performance ($45.0$\% WER) surpasses that of a SoTA model twice its size (SeamlessM4T-large-v2, WER=$47.0$\%) and its teacher model (Whisper-large-v2, WER=$55.1$\%), and its average performance on our new dialectal data ($56.9$\% WER) outperforms all other models. To gain more insight into the poor performance of these models on dialectal data, we conduct an error analysis and report the main types of errors the different models tend to make. The GitHub repository for the project is available at \url{https://github.com/UBC-NLP/distill-whisper-ar}.
Authors: Huaixiu Steven Zheng, Swaroop Mishra, Hugh Zhang, Xinyun Chen, Minmin Chen, Azade Nova, Le Hou, Heng-Tze Cheng, Quoc V. Le, Ed H. Chi, Denny Zhou
Abstract: We introduce NATURAL PLAN, a realistic planning benchmark in natural language containing 3 key tasks: Trip Planning, Meeting Planning, and Calendar Scheduling. We focus our evaluation on the planning capabilities of LLMs with full information on the task, by providing outputs from tools such as Google Flights, Google Maps, and Google Calendar as contexts to the models. This eliminates the need for a tool-use environment for evaluating LLMs on Planning. We observe that NATURAL PLAN is a challenging benchmark for state of the art models. For example, in Trip Planning, GPT-4 and Gemini 1.5 Pro could only achieve 31.1% and 34.8% solve rate respectively. We find that model performance drops drastically as the complexity of the problem increases: all models perform below 5% when there are 10 cities, highlighting a significant gap in planning in natural language for SoTA LLMs. We also conduct extensive ablation studies on NATURAL PLAN to further shed light on the (in)effectiveness of approaches such as self-correction, few-shot generalization, and in-context planning with long-contexts on improving LLM planning.
Authors: Renjie Liu, Yanxiang Zhang, Yun Zhu, Haicheng Sun, Yuanbo Zhang, Michael Xuelin Huang, Shanqing Cai, Lei Meng, Shumin Zhai
Abstract: The impressive capabilities in Large Language Models (LLMs) provide a powerful approach to reimagine users' typing experience. This paper demonstrates Proofread, a novel Gboard feature powered by a server-side LLM in Gboard, enabling seamless sentence-level and paragraph-level corrections with a single tap. We describe the complete system in this paper, from data generation, metrics design to model tuning and deployment. To obtain models with sufficient quality, we implement a careful data synthetic pipeline tailored to online use cases, design multifaceted metrics, employ a two-stage tuning approach to acquire the dedicated LLM for the feature: the Supervised Fine Tuning (SFT) for foundational quality, followed by the Reinforcement Learning (RL) tuning approach for targeted refinement. Specifically, we find sequential tuning on Rewrite and proofread tasks yields the best quality in SFT stage, and propose global and direct rewards in the RL tuning stage to seek further improvement. Extensive experiments on a human-labeled golden set showed our tuned PaLM2-XS model achieved 85.56\% good ratio. We launched the feature to Pixel 8 devices by serving the model on TPU v5 in Google Cloud, with thousands of daily active users. Serving latency was significantly reduced by quantization, bucket inference, text segmentation, and speculative decoding. Our demo could be seen in \href{https://youtu.be/4ZdcuiwFU7I}{Youtube}.
Authors: Fabi\'an Villena, Luis Miranda, Claudio Aracena
Abstract: Large language models (LLMs) allow us to generate high-quality human-like text. One interesting task in natural language processing (NLP) is named entity recognition (NER), which seeks to detect mentions of relevant information in documents. This paper presents llmNER, a Python library for implementing zero-shot and few-shot NER with LLMs; by providing an easy-to-use interface, llmNER can compose prompts, query the model, and parse the completion returned by the LLM. Also, the library enables the user to perform prompt engineering efficiently by providing a simple interface to test multiple variables. We validated our software on two NER tasks to show the library's flexibility. llmNER aims to push the boundaries of in-context learning research by removing the barrier of the prompting and parsing steps.
Authors: Keqi Deng, Philip C. Woodland
Abstract: While the neural transducer is popular for online speech recognition, simultaneous speech translation (SST) requires both streaming and re-ordering capabilities. This paper presents the LS-Transducer-SST, a label-synchronous neural transducer for SST, which naturally possesses these two properties. The LS-Transducer-SST dynamically decides when to emit translation tokens based on an Auto-regressive Integrate-and-Fire (AIF) mechanism. A latency-controllable AIF is also proposed, which can control the quality-latency trade-off either only during decoding, or it can be used in both decoding and training. The LS-Transducer-SST can naturally utilise monolingual text-only data via its prediction network which helps alleviate the key issue of data sparsity for E2E SST. During decoding, a chunk-based incremental joint decoding technique is designed to refine and expand the search space. Experiments on the Fisher-CallHome Spanish (Es-En) and MuST-C En-De data show that the LS-Transducer-SST gives a better quality-latency trade-off than existing popular methods. For example, the LS-Transducer-SST gives a 3.1/2.9 point BLEU increase (Es-En/En-De) relative to CAAT at a similar latency and a 1.4 s reduction in average lagging latency with similar BLEU scores relative to Wait-k.
Authors: Pavan Holur, Shreyas Rajesh, David Chong, Vwani Roychowdhury
Abstract: An experienced human Observer reading a document -- such as a crime report -- creates a succinct plot-like $\textit{``Working Memory''}$ comprising different actors, their prototypical roles and states at any point, their evolution over time based on their interactions, and even a map of missing Semantic parts anticipating them in the future. $\textit{An equivalent AI Observer currently does not exist}$. We introduce the $\textbf{[G]}$enerative $\textbf{[S]}$emantic $\textbf{[W]}$orkspace (GSW) -- comprising an $\textit{``Operator''}$ and a $\textit{``Reconciler''}$ -- that leverages advancements in LLMs to create a generative-style Semantic framework, as opposed to a traditionally predefined set of lexicon labels. Given a text segment $C_n$ that describes an ongoing situation, the $\textit{Operator}$ instantiates actor-centric Semantic maps (termed ``Workspace instance'' $\mathcal{W}_n$). The $\textit{Reconciler}$ resolves differences between $\mathcal{W}_n$ and a ``Working memory'' $\mathcal{M}_n^*$ to generate the updated $\mathcal{M}_{n+1}^*$. GSW outperforms well-known baselines on several tasks ($\sim 94\%$ vs. FST, GLEN, BertSRL - multi-sentence Semantics extraction, $\sim 15\%$ vs. NLI-BERT, $\sim 35\%$ vs. QA). By mirroring the real Observer, GSW provides the first step towards Spatial Computing assistants capable of understanding individual intentions and predicting future behavior.
Authors: Md Imbesat Hassan Rizvi, Xiaodan Zhu, Iryna Gurevych
Abstract: Spatial reasoning is a crucial component of both biological and artificial intelligence. In this work, we present a comprehensive study of the capability of current state-of-the-art large language models (LLMs) on spatial reasoning. To support our study, we created and contribute a novel Spatial Reasoning Characterization (SpaRC) framework and Spatial Reasoning Paths (SpaRP) datasets, to enable an in-depth understanding of the spatial relations and compositions as well as the usefulness of spatial reasoning chains. We found that all the state-of-the-art LLMs do not perform well on the datasets -- their performances are consistently low across different setups. The spatial reasoning capability improves substantially as model sizes scale up. Finetuning both large language models (e.g., Llama-2-70B) and smaller ones (e.g., Llama-2-13B) can significantly improve their F1-scores by 7--32 absolute points. We also found that the top proprietary LLMs still significantly outperform their open-source counterparts in topological spatial understanding and reasoning.
Authors: Jie Zhao, Ziyu Guan, Cai Xu, Wei Zhao, Yue Jiang
Abstract: Text style transfer (TST) aims to vary the style polarity of text while preserving the semantic content. Although recent advancements have demonstrated remarkable progress in short TST, it remains a relatively straightforward task with limited practical applications. The more comprehensive long TST task presents two challenges: (1) existing methods encounter difficulties in accurately evaluating content attributes in multiple words, leading to content degradation; (2) the conventional vanilla style classifier loss encounters obstacles in maintaining consistent style across multiple generated sentences. In this paper, we propose a novel method SC2, where a multilayer Joint Style-Content Weighed (JSCW) module and a Style Consistency loss are designed to address the two issues. The JSCW simultaneously assesses the amounts of style and content attributes within a token, aiming to acquire a lossless content representation and thereby enhancing content preservation. The multiple JSCW layers further progressively refine content representations. We design a style consistency loss to ensure the generated multiple sentences consistently reflect the target style polarity. Moreover, we incorporate a denoising non-autoregressive decoder to accelerate the training. We conduct plentiful experiments and the results show significant improvements of SC2 over competitive baselines. Our code: https://github.com/jiezhao6/SC2.
Authors: Yanquan Chen, Zhen Wu, Junjie Guo, Shujian Huang, Xinyu Dai
Abstract: Large language models (LLMs) exhibit robust capabilities in text generation and comprehension, mimicking human behavior and exhibiting synthetic personalities. However, some LLMs have displayed offensive personality, propagating toxic discourse. Existing literature neglects the origin and evolution of LLM personalities, as well as the effective personality control. To fill these gaps, our study embarked on a comprehensive investigation into LLM personality control. We investigated several typical methods to influence LLMs, including three training methods: Continual Pre-training, Supervised Fine-Tuning (SFT), and Reinforcement Learning from Human Feedback (RLHF), along with inference phase considerations (prompts). Our investigation revealed a hierarchy of effectiveness in control: Prompt > SFT > RLHF > Continual Pre-train. Notably, SFT exhibits a higher control success rate compared to prompt induction. While prompts prove highly effective, we found that prompt-induced personalities are less robust than those trained, making them more prone to showing conflicting personalities under reverse personality prompt induction. Besides, harnessing the strengths of both SFT and prompt, we proposed $\underline{\text{P}}$rompt $\underline{\text{I}}$nduction post $\underline{\text{S}}$upervised $\underline{\text{F}}$ine-tuning (PISF), which emerges as the most effective and robust strategy for controlling LLMs' personality, displaying high efficacy, high success rates, and high robustness. Even under reverse personality prompt induction, LLMs controlled by PISF still exhibit stable and robust personalities.
Authors: Jiaxin Wen, Ruiqi Zhong, Pei Ke, Zhihong Shao, Hongning Wang, Minlie Huang
Abstract: When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones. To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks. We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution. We collect a dataset of human repair experiences on different decomposed solutions. Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV. We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3\% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts.
Authors: Zhi Zhou, Jiang-Xin Shi, Peng-Xiao Song, Xiao-Wen Yang, Yi-Xuan Jin, Lan-Zhe Guo, Yu-Feng Li
Abstract: Large language models (LLMs), including both proprietary and open-source models, have showcased remarkable capabilities in addressing a wide range of downstream tasks. Nonetheless, when it comes to practical Chinese legal tasks, these models fail to meet the actual requirements. Proprietary models do not ensure data privacy for sensitive legal cases, while open-source models demonstrate unsatisfactory performance due to their lack of legal knowledge. To address this problem, we introduce LawGPT, the first open-source model specifically designed for Chinese legal applications. LawGPT comprises two key components: legal-oriented pre-training and legal supervised fine-tuning. Specifically, we employ large-scale Chinese legal documents for legal-oriented pre-training to incorporate legal domain knowledge. To further improve the model's performance on downstream legal tasks, we create a knowledge-driven instruction dataset for legal supervised fine-tuning. Our experimental results demonstrate that LawGPT outperforms the open-source LLaMA 7B model. Our code and resources are publicly available at https://github.com/pengxiao-song/LaWGPT and have received 5.7K stars on GitHub.
Authors: Sangwon Ryu, Heejin Do, Yunsu Kim, Gary Geunbae Lee, Jungseul Ok
Abstract: Remarkable advances in large language models (LLMs) have enabled high-quality text summarization. However, this capability is currently accessible only through LLMs of substantial size or proprietary LLMs with usage fees. In response, smaller-scale LLMs (sLLMs) of easy accessibility and low costs have been extensively studied, yet they often suffer from missing key information and entities, i.e., low relevance, in particular, when input documents are long. We hence propose a key-element-informed instruction tuning for summarization, so-called KEITSum, which identifies key elements in documents and instructs sLLM to generate summaries capturing these key elements. Experimental results on dialogue and news datasets demonstrate that sLLM with KEITSum indeed provides high-quality summarization with higher relevance and less hallucinations, competitive to proprietary LLM.
Authors: Gyutae Park, Seojin Hwang, Hwanhee Lee
Abstract: Cross-lingual summarization (XLS) aims to generate a summary in a target language different from the source language document. While large language models (LLMs) have shown promising zero-shot XLS performance, their few-shot capabilities on this task remain unexplored, especially for low-resource languages with limited parallel data. In this paper, we investigate the few-shot XLS performance of various models, including Mistral-7B-Instruct-v0.2, GPT-3.5, and GPT-4. Our experiments demonstrate that few-shot learning significantly improves the XLS performance of LLMs, particularly GPT-3.5 and GPT-4, in low-resource settings. However, the open-source model Mistral-7B-Instruct-v0.2 struggles to adapt effectively to the XLS task with limited examples. Our findings highlight the potential of few-shot learning for improving XLS performance and the need for further research in designing LLM architectures and pre-training objectives tailored for this task. We provide a future work direction to explore more effective few-shot learning strategies and to investigate the transfer learning capabilities of LLMs for cross-lingual summarization.
Authors: Xiang Kong, Tom Gunter, Ruoming Pang
Abstract: Large Language Model (LLM) pre-training exhausts an ever growing compute budget, yet recent research has demonstrated that careful document selection enables comparable model quality with only a fraction of the FLOPs. Inspired by efforts suggesting that domain-specific training document selection is in fact an interpretable process [Gunasekar et al., 2023], as well as research showing that instruction-finetuned LLMs are adept zero-shot data labelers [Gilardi et al.,2023], we explore a promising direction for scalable general-domain document selection; employing a prompted LLM as a document grader, we distill quality labels into a classifier model, which is applied at scale to a large, and already heavily-filtered, web-crawl-derived corpus autonomously. Following the guidance of this classifier, we drop 75% of the corpus and train LLMs on the remaining data. Results across multiple benchmarks show that: 1. Filtering allows us to quality-match a model trained on the full corpus across diverse benchmarks with at most 70% of the FLOPs, 2. More capable LLM labelers and classifier models lead to better results that are less sensitive to the labeler's prompt, 3. In-context learning helps to boost the performance of less-capable labeling models. In all cases we use open-source datasets, models, recipes, and evaluation frameworks, so that results can be reproduced by the community.
Authors: Wichayaporn Wongkamjan, Feng Gu, Yanze Wang, Ulf Hermjakob, Jonathan May, Brandon M. Stewart, Jonathan K. Kummerfeld, Denis Peskoff, Jordan Lee Boyd-Graber
Abstract: The boardgame Diplomacy is a challenging setting for communicative and cooperative artificial intelligence. The most prominent communicative Diplomacy AI, Cicero, has excellent strategic abilities, exceeding human players. However, the best Diplomacy players master communication, not just tactics, which is why the game has received attention as an AI challenge. This work seeks to understand the degree to which Cicero succeeds at communication. First, we annotate in-game communication with abstract meaning representation to separate in-game tactics from general language. Second, we run two dozen games with humans and Cicero, totaling over 200 human-player hours of competition. While AI can consistently outplay human players, AI-Human communication is still limited because of AI's difficulty with deception and persuasion. This shows that Cicero relies on strategy and has not yet reached the full promise of communicative and cooperative AI.
Authors: Chengang Hu, Xiao Liu, Yansong Feng
Abstract: Most of the existing compositional generalization datasets are synthetically-generated, resulting in a lack of natural language variation. While there have been recent attempts to introduce non-synthetic datasets for compositional generalization, they suffer from either limited data scale or a lack of diversity in the forms of combinations. To better investigate compositional generalization with more linguistic phenomena and compositional diversity, we propose the DIsh NamE Recognition (DiNeR) task and create a large realistic Chinese dataset. Given a recipe instruction, models are required to recognize the dish name composed of diverse combinations of food, actions, and flavors. Our dataset consists of 3,811 dishes and 228,114 recipes, and involves plenty of linguistic phenomena such as anaphora, omission and ambiguity. We provide two strong baselines based on T5 and large language models (LLMs). This work contributes a challenging task, baseline methods to tackle the task, and insights into compositional generalization in the context of dish name recognition. Code and data are available at https://github.com/Jumpy-pku/DiNeR.
Authors: Dongkyu Lee, Chandana Satya Prakash, Jack FitzGerald, Jens Lehmann
Abstract: Leveraging external knowledge is crucial for achieving high performance in knowledge-intensive tasks, such as question answering. The retrieve-and-read approach is widely adopted for integrating external knowledge into a language model. However, this approach suffers from increased computational cost and latency due to the long context length, which grows proportionally with the number of retrieved knowledge. Furthermore, existing retrieval-augmented models typically retrieve information from a single type of knowledge source, limiting their scalability to diverse knowledge sources with varying structures. In this work, we introduce an efficient memory-augmented transformer called MATTER, designed to retrieve relevant knowledge from multiple heterogeneous knowledge sources. Specifically, our model retrieves and reads from both unstructured sources (paragraphs) and semi-structured sources (QA pairs) in the form of fixed-length neural memories. We demonstrate that our model outperforms existing efficient retrieval-augmented models on popular QA benchmarks in terms of both accuracy and speed. Furthermore, MATTER achieves competitive results compared to conventional read-and-retrieve models while having 100x throughput during inference.
Authors: Junlin Wang, Jue Wang, Ben Athiwaratkun, Ce Zhang, James Zou
Abstract: Recent advances in large language models (LLMs) demonstrate substantial capabilities in natural language understanding and generation tasks. With the growing number of LLMs, how to harness the collective expertise of multiple LLMs is an exciting open direction. Toward this goal, we propose a new approach that leverages the collective strengths of multiple LLMs through a Mixture-of-Agents (MoA) methodology. In our approach, we construct a layered MoA architecture wherein each layer comprises multiple LLM agents. Each agent takes all the outputs from agents in the previous layer as auxiliary information in generating its response. MoA models achieves state-of-art performance on AlpacaEval 2.0, MT-Bench and FLASK, surpassing GPT-4 Omni. For example, our MoA using only open-source LLMs is the leader of AlpacaEval 2.0 by a substantial gap, achieving a score of 65.1% compared to 57.5% by GPT-4 Omni.
Authors: Yinghui Xia, Yuyan Chen, Tianyu Shi, Jun Wang, Jinsong Yang
Abstract: Automated code generation is a pivotal capability of large language models (LLMs). However, assessing this capability in real-world scenarios remains challenging. Previous methods focus more on low-level code generation, such as model loading, instead of generating high-level codes catering for real-world tasks, such as image-to-text, text classification, in various domains. Therefore, we construct AICoderEval, a dataset focused on real-world tasks in various domains based on HuggingFace, PyTorch, and TensorFlow, along with comprehensive metrics for evaluation and enhancing LLMs' task-specific code generation capability. AICoderEval contains test cases and complete programs for automated evaluation of these tasks, covering domains such as natural language processing, computer vision, and multimodal learning. To facilitate research in this area, we open-source the AICoderEval dataset at \url{https://huggingface.co/datasets/vixuowis/AICoderEval}. After that, we propose CoderGen, an agent-based framework, to help LLMs generate codes related to real-world tasks on the constructed AICoderEval. Moreover, we train a more powerful task-specific code generation model, named AICoder, which is refined on llama-3 based on AICoderEval. Our experiments demonstrate the effectiveness of CoderGen in improving LLMs' task-specific code generation capability (by 12.00\% on pass@1 for original model and 9.50\% on pass@1 for ReAct Agent). AICoder also outperforms current code generation LLMs, indicating the great quality of the AICoderEval benchmark.
Authors: Xiao Yang, Kai Sun, Hao Xin, Yushi Sun, Nikita Bhalla, Xiangsen Chen, Sajal Choudhary, Rongze Daniel Gui, Ziran Will Jiang, Ziyu Jiang, Lingkun Kong, Brian Moran, Jiaqi Wang, Yifan Ethan Xu, An Yan, Chenyu Yang, Eting Yuan, Hanwen Zha, Nan Tang, Lei Chen, Nicolas Scheffer, Yue Liu, Nirav Shah, Rakesh Wanga, Anuj Kumar, Wen-tau Yih, Xin Luna Dong
Abstract: Retrieval-Augmented Generation (RAG) has recently emerged as a promising solution to alleviate Large Language Model (LLM)'s deficiency in lack of knowledge. Existing RAG datasets, however, do not adequately represent the diverse and dynamic nature of real-world Question Answering (QA) tasks. To bridge this gap, we introduce the Comprehensive RAG Benchmark (CRAG), a factual question answering benchmark of 4,409 question-answer pairs and mock APIs to simulate web and Knowledge Graph (KG) search. CRAG is designed to encapsulate a diverse array of questions across five domains and eight question categories, reflecting varied entity popularity from popular to long-tail, and temporal dynamisms ranging from years to seconds. Our evaluation on this benchmark highlights the gap to fully trustworthy QA. Whereas most advanced LLMs achieve <=34% accuracy on CRAG, adding RAG in a straightforward manner improves the accuracy only to 44%. State-of-the-art industry RAG solutions only answer 63% questions without any hallucination. CRAG also reveals much lower accuracy in answering questions regarding facts with higher dynamism, lower popularity, or higher complexity, suggesting future research directions. The CRAG benchmark laid the groundwork for a KDD Cup 2024 challenge, attracting thousands of participants and submissions within the first 50 days of the competition. We commit to maintaining CRAG to serve research communities in advancing RAG solutions and general QA solutions.
Authors: Ling Shi, Deyi Xiong
Abstract: Large language models (LLMs) are possessed of numerous beneficial capabilities, yet their potential inclination harbors unpredictable risks that may materialize in the future. We hence propose CRiskEval, a Chinese dataset meticulously designed for gauging the risk proclivities inherent in LLMs such as resource acquisition and malicious coordination, as part of efforts for proactive preparedness. To curate CRiskEval, we define a new risk taxonomy with 7 types of frontier risks and 4 safety levels, including extremely hazardous,moderately hazardous, neutral and safe. We follow the philosophy of tendency evaluation to empirically measure the stated desire of LLMs via fine-grained multiple-choice question answering. The dataset consists of 14,888 questions that simulate scenarios related to predefined 7 types of frontier risks. Each question is accompanied with 4 answer choices that state opinions or behavioral tendencies corresponding to the question. All answer choices are manually annotated with one of the defined risk levels so that we can easily build a fine-grained frontier risk profile for each assessed LLM. Extensive evaluation with CRiskEval on a spectrum of prevalent Chinese LLMs has unveiled a striking revelation: most models exhibit risk tendencies of more than 40% (weighted tendency to the four risk levels). Furthermore, a subtle increase in the model's inclination toward urgent self-sustainability, power seeking and other dangerous goals becomes evident as the size of models increase. To promote further research on the frontier risk evaluation of LLMs, we publicly release our dataset at https://github.com/lingshi6565/Risk_eval.
Authors: Jiangnan Li, Zheng Lin, Lanrui Wang, Qingyi Si, Yanan Cao, Mo Yu, Peng Fu, Weiping Wang, Jie Zhou
Abstract: Humans convey emotions through daily dialogues, making emotion understanding a crucial step of affective intelligence. To understand emotions in dialogues, machines are asked to recognize the emotion for an utterance (Emotion Recognition in Dialogues, ERD); based on the emotion, then find causal utterances for the emotion (Emotion Cause Extraction in Dialogues, ECED). The setting of the two tasks requires first ERD and then ECED, ignoring the mutual complement between emotion and cause. To fix this, some new tasks are proposed to extract them simultaneously. Although the current research on these tasks has excellent achievements, simply identifying emotion-related factors by classification modeling lacks realizing the specific thinking process of causes stimulating the emotion in an explainable way. This thinking process especially reflected in the reasoning ability of Large Language Models (LLMs) is under-explored. To this end, we propose a new task "Emotion Deducing Explanation in Dialogues" (EDEN). EDEN recognizes emotion and causes in an explicitly thinking way. That is, models need to generate an explanation text, which first summarizes the causes; analyzes the inner activities of the speakers triggered by the causes using common sense; then guesses the emotion accordingly. To support the study of EDEN, based on the existing resources in ECED, we construct two EDEN datasets by human effort. We further evaluate different models on EDEN and find that LLMs are more competent than conventional PLMs. Besides, EDEN can help LLMs achieve better recognition of emotions and causes, which explores a new research direction of explainable emotion understanding in dialogues.
Authors: Bill Yuchen Lin, Yuntian Deng, Khyathi Chandu, Faeze Brahman, Abhilasha Ravichander, Valentina Pyatkin, Nouha Dziri, Ronan Le Bras, Yejin Choi
Abstract: We introduce WildBench, an automated evaluation framework designed to benchmark large language models (LLMs) using challenging, real-world user queries. WildBench consists of 1,024 tasks carefully selected from over one million human-chatbot conversation logs. For automated evaluation with WildBench, we have developed two metrics, WB-Reward and WB-Score, which are computable using advanced LLMs such as GPT-4-turbo. WildBench evaluation uses task-specific checklists to evaluate model outputs systematically and provides structured explanations that justify the scores and comparisons, resulting in more reliable and interpretable automatic judgments. WB-Reward employs fine-grained pairwise comparisons between model responses, generating five potential outcomes: much better, slightly better, slightly worse, much worse, or a tie. Unlike previous evaluations that employed a single baseline model, we selected three baseline models at varying performance levels to ensure a comprehensive pairwise evaluation. Additionally, we propose a simple method to mitigate length bias, by converting outcomes of ``slightly better/worse'' to ``tie'' if the winner response exceeds the loser one by more than $K$ characters. WB-Score evaluates the quality of model outputs individually, making it a fast and cost-efficient evaluation metric. WildBench results demonstrate a strong correlation with the human-voted Elo ratings from Chatbot Arena on hard tasks. Specifically, WB-Reward achieves a Pearson correlation of 0.98 with top-ranking models. Additionally, WB-Score reaches 0.95, surpassing both ArenaHard's 0.91 and AlpacaEval2.0's 0.89 for length-controlled win rates, as well as the 0.87 for regular win rates.
Authors: Ruihan Yang, Jiangjie Chen, Yikai Zhang, Siyu Yuan, Aili Chen, Kyle Richardson, Yanghua Xiao, Deqing Yang
Abstract: Language agents powered by large language models (LLMs) are increasingly valuable as decision-making tools in domains such as gaming and programming. However, these agents often face challenges in achieving high-level goals without detailed instructions and in adapting to environments where feedback is delayed. In this paper, we present SelfGoal, a novel automatic approach designed to enhance agents' capabilities to achieve high-level goals with limited human prior and environmental feedback. The core concept of SelfGoal involves adaptively breaking down a high-level goal into a tree structure of more practical subgoals during the interaction with environments while identifying the most useful subgoals and progressively updating this structure. Experimental results demonstrate that SelfGoal significantly enhances the performance of language agents across various tasks, including competitive, cooperative, and deferred feedback environments. Project page: https://selfgoal-agent.github.io.
Authors: David Samuel
Abstract: This paper explores the in-context learning capabilities of masked language models, challenging the common view that this ability does not 'emerge' in them. We present an embarrassingly simple inference technique that enables DeBERTa to operate as a generative model without any additional training. Our findings demonstrate that DeBERTa can match and even surpass GPT-3, its contemporary that famously introduced the paradigm of in-context learning. The comparative analysis reveals that the masked and causal language models behave very differently, as they clearly outperform each other on different categories of tasks. This suggests that there is great potential for a hybrid training approach that takes advantage of the strengths of both training objectives.
Authors: Xinyue Cui, Swabha Swayamdipta
Abstract: Despite the remarkable generative capabilities of language models in producing naturalistic language, their effectiveness on explicit manipulation and generation of linguistic structures remain understudied. In this paper, we investigate the task of generating new sentences preserving a given semantic structure, following the FrameNet formalism. We propose a framework to produce novel frame-semantically annotated sentences following an overgenerate-and-filter approach. Our results show that conditioning on rich, explicit semantic information tends to produce generations with high human acceptance, under both prompting and finetuning. Our generated frame-semantic structured annotations are effective at training data augmentation for frame-semantic role labeling in low-resource settings; however, we do not see benefits under higher resource settings. Our study concludes that while generating high-quality, semantically rich data might be within reach, the downstream utility of such generations remains to be seen, highlighting the outstanding challenges with automating linguistic annotation tasks.
Authors: Hongyu Li, Liang Ding, Meng Fang, Dacheng Tao
Abstract: Catastrophic Forgetting (CF) means models forgetting previously acquired knowledge when learning new data. It compromises the effectiveness of large language models (LLMs) during fine-tuning, yet the underlying causes have not been thoroughly investigated. This paper takes the first step to reveal the direct link between the flatness of the model loss landscape and the extent of CF in the field of LLMs. Based on this, we introduce the sharpness-aware minimization to mitigate CF by flattening the loss landscape. Experiments on three widely-used fine-tuning datasets, spanning different model scales, demonstrate the effectiveness of our method in alleviating CF. Analyses show that we nicely complement the existing anti-forgetting strategies, further enhancing the resistance of LLMs to CF.
Authors: Rui Ye, Rui Ge, Xinyu Zhu, Jingyi Chai, Yaxin Du, Yang Liu, Yanfeng Wang, Siheng Chen
Abstract: Federated learning has enabled multiple parties to collaboratively train large language models without directly sharing their data (FedLLM). Following this training paradigm, the community has put massive efforts from diverse aspects including framework, performance, and privacy. However, an unpleasant fact is that there are currently no realistic datasets and benchmarks for FedLLM and previous works all rely on artificially constructed datasets, failing to capture properties in real-world scenarios. Addressing this, we propose FedLLM-Bench, which involves 8 training methods, 4 training datasets, and 6 evaluation metrics, to offer a comprehensive testbed for the FedLLM community. FedLLM-Bench encompasses three datasets (e.g., user-annotated multilingual dataset) for federated instruction tuning and one dataset (e.g., user-annotated preference dataset) for federated preference alignment, whose scale of client number ranges from 38 to 747. Our datasets incorporate several representative diversities: language, quality, quantity, instruction, length, embedding, and preference, capturing properties in real-world scenarios. Based on FedLLM-Bench, we conduct experiments on all datasets to benchmark existing FL methods and provide empirical insights (e.g., multilingual collaboration). We believe that our FedLLM-Bench can benefit the FedLLM community by reducing required efforts, providing a practical testbed, and promoting fair comparisons. Code and datasets are available at https://github.com/rui-ye/FedLLM-Bench.
Authors: Jaap Jumelet, Willem Zuidema, Arabella Sinclair
Abstract: We explore which linguistic factors -- at the sentence and token level -- play an important role in influencing language model predictions, and investigate whether these are reflective of results found in humans and human corpora (Gries and Kootstra, 2017). We make use of the structural priming paradigm, where recent exposure to a structure facilitates processing of the same structure. We don't only investigate whether, but also where priming effects occur, and what factors predict them. We show that these effects can be explained via the inverse frequency effect, known in human priming, where rarer elements within a prime increase priming effects, as well as lexical dependence between prime and target. Our results provide an important piece in the puzzle of understanding how properties within their context affect structural prediction in language models.
Authors: Yikun Wang, Rui Zheng, Liang Ding, Qi Zhang, Dahua Lin, Dacheng Tao
Abstract: As instruction-tuned large language models (LLMs) evolve, aligning pretrained foundation models presents increasing challenges. Existing alignment strategies, which typically leverage diverse and high-quality data sources, often overlook the intrinsic uncertainty of tasks, learning all data samples equally. This may lead to suboptimal data efficiency and model performance. In response, we propose uncertainty-aware learning (UAL) to improve the model alignment of different task scenarios, by introducing the sample uncertainty (elicited from more capable LLMs). We implement UAL in a simple fashion -- adaptively setting the label smoothing value of training according to the uncertainty of individual samples. Analysis shows that our UAL indeed facilitates better token clustering in the feature space, validating our hypothesis. Extensive experiments on widely used benchmarks demonstrate that our UAL significantly and consistently outperforms standard supervised fine-tuning. Notably, LLMs aligned in a mixed scenario have achieved an average improvement of 10.62\% on high-entropy tasks (i.e., AlpacaEval leaderboard), and 1.81\% on complex low-entropy tasks (i.e., MetaMath and GSM8K).
Authors: Denis Saveliev, Ruslan Kuchakov
Abstract: We present the comprehensive Russian primary and secondary legislation corpus covering 1991 to 2023. The corpus collects all 281,413 texts (176,523,268 tokens) of non-secret federal regulations and acts, along with their metadata. The corpus has two versions the original text with minimal preprocessing and a version prepared for linguistic analysis with morphosyntactic markup.
Authors: Raphael Gruber, Abdelrahman Abdallah, Michael F\"arber, Adam Jatowt
Abstract: We introduce ComplexTempQA,a large-scale dataset consisting of over 100 million question-answer pairs designed to tackle the challenges in temporal question answering. ComplexTempQA significantly surpasses existing benchmarks like HOTPOTQA, TORQUE, and TEQUILA in scale and scope. Utilizing data from Wikipedia and Wikidata, the dataset covers questions spanning over two decades and offers an unmatched breadth of topics. We introduce a unique taxonomy that categorizes questions as attributes, comparisons, and counting questions, each revolving around events, entities, and time periods. One standout feature of ComplexTempQA is the high complexity of its questions, which demand effective capabilities for answering such as across-time comparison, temporal aggregation, and multi-hop reasoning involving temporal event ordering and entity recognition. Additionally, each question is accompanied by detailed metadata, including specific time scopes, allowing for comprehensive evaluation and enhancement of the temporal reasoning abilities of large language models. ComplexTempQA serves both as a testing ground for developing sophisticated AI models and as a foundation for advancing research in question answering, information retrieval, and language understanding. Dataset and code are freely available at: https://github.com/DataScienceUIBK/ComplexTempQA.
Authors: Nankai Lin, Hongyan Wu, Zhengming Chen, Zijian Li, Lianxi Wang, Shengyi Jiang, Dong Zhou, Aimin Yang
Abstract: Hate speech on social media is ubiquitous but urgently controlled. Without detecting and mitigating the biases brought by hate speech, different types of ethical problems. While a number of datasets have been proposed to address the problem of hate speech detection, these datasets seldom consider the diversity and variability of bias, making it far from real-world scenarios. To fill this gap, we propose a benchmark, named HateDebias, to analyze the model ability of hate speech detection under continuous, changing environments. Specifically, to meet the diversity of biases, we collect existing hate speech detection datasets with different types of biases. To further meet the variability (i.e., the changing of bias attributes in datasets), we reorganize datasets to follow the continuous learning setting. We evaluate the detection accuracy of models trained on the datasets with a single type of bias with the performance on the HateDebias, where a significant performance drop is observed. To provide a potential direction for debiasing, we further propose a debiasing framework based on continuous learning and bias information regularization, as well as the memory replay strategies to ensure the debiasing ability of the model. Experiment results on the proposed benchmark show that the aforementioned method can improve several baselines with a distinguished margin, highlighting its effectiveness in real-world applications.
Authors: Megh Thakkar, Quentin Fournier, Matthew D Riemer, Pin-Yu Chen, Amal Zouaq, Payel Das, Sarath Chandar
Abstract: Large language models are first pre-trained on trillions of tokens and then instruction-tuned or aligned to specific preferences. While pre-training remains out of reach for most researchers due to the compute required, fine-tuning has become affordable thanks to parameter-efficient methods such as LoRA and QLoRA. Alignment is known to be sensitive to the many factors involved, including the quantity and quality of data, the alignment method, and the adapter rank. However, there has not yet been an extensive study of their effect on downstream performance. To address this gap, we conduct an in-depth investigation of the impact of popular choices for three crucial axes: (i) the alignment dataset (HH-RLHF and BeaverTails), (ii) the alignment technique (SFT and DPO), and (iii) the model (LLaMA-1, Vicuna-v1.3, Mistral-7b, and Mistral-7b-Instruct). Our extensive setup spanning over 300 experiments reveals consistent trends and unexpected findings. We observe how more informative data helps with preference alignment, cases where supervised fine-tuning outperforms preference optimization, and how aligning to a distinct preference boosts performance on downstream tasks. Through our in-depth analyses, we put forward key guidelines to help researchers perform more effective parameter-efficient LLM alignment.
Authors: Rabiraj Bandyopadhyay, Dennis Assenmacher, Jose M. Alonso Moral, Claudia Wagner
Abstract: There is an increase in the proliferation of online hate commensurate with the rise in the usage of social media. In response, there is also a significant advancement in the creation of automated tools aimed at identifying harmful text content using approaches grounded in Natural Language Processing and Deep Learning. Although it is known that training Deep Learning models require a substantial amount of annotated data, recent line of work suggests that models trained on specific subsets of the data still retain performance comparable to the model that was trained on the full dataset. In this work, we show how we can leverage influence scores to estimate the importance of a data point while training a model and designing a pruning strategy applied to the case of sexism detection. We evaluate the model performance trained on data pruned with different pruning strategies on three out-of-domain datasets and find, that in accordance with other work a large fraction of instances can be removed without significant performance drop. However, we also discover that the strategies for pruning data, previously successful in Natural Language Inference tasks, do not readily apply to the detection of harmful content and instead amplify the already prevalent class imbalance even more, leading in the worst-case to a complete absence of the hateful class.
Authors: Micha{\l} Romaszewski, Przemys{\l}aw Seku{\l}a, Przemys{\l}aw G{\l}omb, Micha{\l} Cholewa, Katarzyna Ko{\l}odziej
Abstract: Large Language Models (LLMs) have shown exceptional performance in text processing. Notably, LLMs can synthesize information from large datasets and explain their decisions similarly to human reasoning through a chain of thought (CoT). An emerging application of LLMs is the handling and interpreting of numerical data, where fine-tuning enhances their performance over basic inference methods. This paper proposes a novel approach to training LLMs using knowledge transfer from a random forest (RF) ensemble, leveraging its efficiency and accuracy. By converting RF decision paths into natural language statements, we generate outputs for LLM fine-tuning, enhancing the model's ability to classify and explain its decisions. Our method includes verifying these rules through established classification metrics, ensuring their correctness. We also examine the impact of preprocessing techniques on the representation of numerical data and their influence on classification accuracy and rule correctness
Authors: Ping Yu, Kaitao Song, Fengchen He, Ming Chen, Jianfeng Lu
Abstract: The recently unprecedented advancements in Large Language Models (LLMs) have propelled the medical community by establishing advanced medical-domain models. However, due to the limited collection of medical datasets, there are only a few comprehensive benchmarks available to gauge progress in this area. In this paper, we introduce a new medical question-answering (QA) dataset that contains massive manual instruction for solving Traditional Chinese Medicine examination tasks, called TCMD. Specifically, our TCMD collects massive questions across diverse domains with their annotated medical subjects and thus supports us in comprehensively assessing the capability of LLMs in the TCM domain. Extensive evaluation of various general LLMs and medical-domain-specific LLMs is conducted. Moreover, we also analyze the robustness of current LLMs in solving TCM QA tasks by introducing randomness. The inconsistency of the experimental results also reveals the shortcomings of current LLMs in solving QA tasks. We also expect that our dataset can further facilitate the development of LLMs in the TCM area.
Authors: Baktash Ansari, Mohammadmostafa Rostamkhani, Sauleh Eetemadi
Abstract: This paper outlines our approach to SemEval 2024 Task 9, BRAINTEASER: A Novel Task Defying Common Sense. The task aims to evaluate the ability of language models to think creatively. The dataset comprises multi-choice questions that challenge models to think "outside of the box". We fine-tune 2 models, BERT and RoBERTa Large. Next, we employ a Chain of Thought (CoT) zero-shot prompting approach with 6 large language models, such as GPT-3.5, Mixtral, and Llama2. Finally, we utilize ReConcile, a technique that employs a "round table conference" approach with multiple agents for zero-shot learning, to generate consensus answers among 3 selected language models. Our best method achieves an overall accuracy of 85 percent on the sentence puzzles subtask.
Authors: Ilya Ilyankou, Meihui Wang, James Haworth, Stefano Cavazzi
Abstract: Large language models (LLMs) exhibit emerging geospatial capabilities, stemming from their pre-training on vast unlabelled text datasets that are often derived from the Common Crawl corpus. However, the geospatial content within CC remains largely unexplored, impacting our understanding of LLMs' spatial reasoning. This paper investigates the prevalence of geospatial data in recent Common Crawl releases using Gemini, a powerful language model. By analyzing a sample of documents and manually revising the results, we estimate that between 1 in 5 and 1 in 6 documents contain geospatial information such as coordinates and street addresses. Our findings provide quantitative insights into the nature and extent of geospatial data within Common Crawl, and web crawl data in general. Furthermore, we formulate questions to guide future investigations into the geospatial content of available web crawl datasets and its influence on LLMs.
Authors: Jitai Hao, WeiWei Sun, Xin Xin, Qi Meng, Zhumin Chen, Pengjie Ren, Zhaochun Ren
Abstract: Parameter-Efficient Fine-tuning (PEFT) facilitates the fine-tuning of Large Language Models (LLMs) under limited resources. However, the fine-tuning performance with PEFT on complex, knowledge-intensive tasks is limited due to the constrained model capacity, which originates from the limited number of additional trainable parameters. To overcome this limitation, we introduce a novel mechanism that fine-tunes LLMs with adapters of larger size yet memory-efficient. This is achieved by leveraging the inherent activation sparsity in the Feed-Forward Networks (FFNs) of LLMs and utilizing the larger capacity of Central Processing Unit (CPU) memory compared to Graphics Processing Unit (GPU). We store and update the parameters of larger adapters on the CPU. Moreover, we employ a Mixture of Experts (MoE)-like architecture to mitigate unnecessary CPU computations and reduce the communication volume between the GPU and CPU. This is particularly beneficial over the limited bandwidth of PCI Express (PCIe). Our method can achieve fine-tuning results comparable to those obtained with larger memory capacities, even when operating under more limited resources such as a 24GB memory single GPU setup, with acceptable loss in training efficiency. Our codes are available at https://github.com/CURRENTF/MEFT.
Authors: Patrick Haller, Lena S. Bolliger, Lena A. J\"ager
Abstract: To date, most investigations on surprisal and entropy effects in reading have been conducted on the group level, disregarding individual differences. In this work, we revisit the predictive power of surprisal and entropy measures estimated from a range of language models (LMs) on data of human reading times as a measure of processing effort by incorporating information of language users' cognitive capacities. To do so, we assess the predictive power of surprisal and entropy estimated from generative LMs on reading data obtained from individuals who also completed a wide range of psychometric tests. Specifically, we investigate if modulating surprisal and entropy relative to cognitive scores increases prediction accuracy of reading times, and we examine whether LMs exhibit systematic biases in the prediction of reading times for cognitively high- or low-performing groups, revealing what type of psycholinguistic subject a given LM emulates. Our study finds that in most cases, incorporating cognitive capacities increases predictive power of surprisal and entropy on reading times, and that generally, high performance in the psychometric tests is associated with lower sensitivity to predictability effects. Finally, our results suggest that the analyzed LMs emulate readers with lower verbal intelligence, suggesting that for a given target group (i.e., individuals with high verbal intelligence), these LMs provide less accurate predictability estimates.
Authors: Sondre Wold, \'Etienne Simon, Lucas Georges Gabriel Charpentier, Egor V. Kostylev, Erik Velldal, Lilja {\O}vrelid
Abstract: Grounded language models use external sources of information, such as knowledge graphs, to meet some of the general challenges associated with pre-training. By extending previous work on compositional generalization in semantic parsing, we allow for a controlled evaluation of the degree to which these models learn and generalize from patterns in knowledge graphs. We develop a procedure for generating natural language questions paired with knowledge graphs that targets different aspects of compositionality and further avoids grounding the language models in information already encoded implicitly in their weights. We evaluate existing methods for combining language models with knowledge graphs and find them to struggle with generalization to sequences of unseen lengths and to novel combinations of seen base components. While our experimental results provide some insight into the expressive power of these models, we hope our work and released datasets motivate future research on how to better combine language models with structured knowledge representations.
Authors: Pengshuo Qiu, Frank Rudzicz, Zining Zhu
Abstract: Large language models (LLMs) can be used to generate natural language explanations (NLE) that are adapted to different users' situations. However, there is yet to be a quantitative evaluation of the extent of such adaptation. To bridge this gap, we collect a benchmarking dataset, Situation-Based Explanation. This dataset contains 100 explanandums. Each explanandum is paired with explanations targeted at three distinct audience types-such as educators, students, and professionals-enabling us to assess how well the explanations meet the specific informational needs and contexts of these diverse groups e.g. students, teachers, and parents. For each "explanandum paired with an audience" situation, we include a human-written explanation. These allow us to compute scores that quantify how the LLMs adapt the explanations to the situations. On an array of pretrained language models with varying sizes, we examine three categories of prompting methods: rule-based prompting, meta-prompting, and in-context learning prompting. We find that 1) language models can generate prompts that result in explanations more precisely aligned with the target situations, 2) explicitly modeling an "assistant" persona by prompting "You are a helpful assistant..." is not a necessary prompt technique for situated NLE tasks, and 3) the in-context learning prompts only can help LLMs learn the demonstration template but can't improve their inference performance. SBE and our analysis facilitate future research towards generating situated natural language explanations.
Authors: Anuradha Welivita, Pearl Pu
Abstract: With the emergence of large language models (LLMs), investigating if they can surpass humans in areas such as emotion recognition and empathetic responding has become a focal point of research. This paper presents a comprehensive study exploring the empathetic responding capabilities of four state-of-the-art LLMs: GPT-4, LLaMA-2-70B-Chat, Gemini-1.0-Pro, and Mixtral-8x7B-Instruct in comparison to a human baseline. We engaged 1,000 participants in a between-subjects user study, assessing the empathetic quality of responses generated by humans and the four LLMs to 2,000 emotional dialogue prompts meticulously selected to cover a broad spectrum of 32 distinct positive and negative emotions. Our findings reveal a statistically significant superiority of the empathetic responding capability of LLMs over humans. GPT-4 emerged as the most empathetic, marking approximately 31% increase in responses rated as "Good" compared to the human benchmark. It was followed by LLaMA-2, Mixtral-8x7B, and Gemini-Pro, which showed increases of approximately 24%, 21%, and 10% in "Good" ratings, respectively. We further analyzed the response ratings at a finer granularity and discovered that some LLMs are significantly better at responding to specific emotions compared to others. The suggested evaluation framework offers a scalable and adaptable approach for assessing the empathy of new LLMs, avoiding the need to replicate this study's findings in future research.
Authors: Eunjeong Hwang, Yichao Zhou, Beliz Gunel, James Bradley Wendt, Sandeep Tata
Abstract: No existing dataset adequately tests how well language models can incrementally update entity summaries - a crucial ability as these models rapidly advance. The Incremental Entity Summarization (IES) task is vital for maintaining accurate, up-to-date knowledge. To address this, we introduce SUMIE, a fully synthetic dataset designed to expose real-world IES challenges. This dataset effectively highlights problems like incorrect entity association and incomplete information presentation. Unlike common synthetic datasets, ours captures the complexity and nuances found in real-world data. We generate informative and diverse attributes, summaries, and unstructured paragraphs in sequence, ensuring high quality. The alignment between generated summaries and paragraphs exceeds 96%, confirming the dataset's quality. Extensive experiments demonstrate the dataset's difficulty - state-of-the-art LLMs struggle to update summaries with an F1 higher than 80.4%. We will open source the benchmark and the evaluation metrics to help the community make progress on IES tasks.
Authors: Maciej Besta, Ales Kubicek, Roman Niggli, Robert Gerstenberger, Lucas Weitzendorf, Mingyuan Chi, Patrick Iff, Joanna Gajda, Piotr Nyczyk, J\"urgen M\"uller, Hubert Niewiadomski, Marcin Chrapek, Micha{\l} Podstawski, Torsten Hoefler
Abstract: Retrieval Augmented Generation (RAG) enhances the abilities of Large Language Models (LLMs) by enabling the retrieval of documents into the LLM context to provide more accurate and relevant responses. Existing RAG solutions do not focus on queries that may require fetching multiple documents with substantially different contents. Such queries occur frequently, but are challenging because the embeddings of these documents may be distant in the embedding space, making it hard to retrieve them all. This paper introduces Multi-Head RAG (MRAG), a novel scheme designed to address this gap with a simple yet powerful idea: leveraging activations of Transformer's multi-head attention layer, instead of the decoder layer, as keys for fetching multi-aspect documents. The driving motivation is that different attention heads can learn to capture different data aspects. Harnessing the corresponding activations results in embeddings that represent various facets of data items and queries, improving the retrieval accuracy for complex queries. We provide an evaluation methodology and metrics, synthetic datasets, and real-world use cases to demonstrate MRAG's effectiveness, showing improvements of up to 20% in relevance over standard RAG baselines. MRAG can be seamlessly integrated with existing RAG frameworks and benchmarking tools like RAGAS as well as different classes of data stores.
Authors: Xiongtao Zhou, Jie He, Yuhua Ke, Guangyao Zhu, V\'ictor Guti\'errez-Basulto, Jeff Z. Pan
Abstract: Multimodal large language models (MLLMs) fine-tuned with multimodal instruction datasets have demonstrated remarkable capabilities in multimodal tasks. However, fine-tuning all parameters of MLLMs has become challenging as they usually contain billions of parameters. To address this issue, we study parameter-efficient fine-tuning (PEFT) methods for MLLMs. We aim to identify effective methods for enhancing the performance of MLLMs in scenarios where only a limited number of parameters are trained. This paper conducts empirical studies using four popular PEFT methods to fine-tune the LLM component of open-source MLLMs. We present a comprehensive analysis that encompasses various aspects, including the impact of PEFT methods on various models, parameters and location of the PEFT module, size of fine-tuning data, model stability based on PEFT methods, MLLM's generalization, and hallucination. We evaluated four PEFT methods on seven datasets from two different categories: unseen and seen datasets. Across all experiments, we show that the adapter is the best-performing PEFT method. At the same time, fine-tuning the connector layers leads to improved performance in most MLLMs. Code and data are available at https://github.com/alenai97/PEFT-MLLM.git.
Authors: Prashanth Vijayaraghavan, Luyao Shi, Stefano Ambrogio, Charles Mackin, Apoorva Nitsure, David Beymer, Ehsan Degan
Abstract: With the unprecedented advancements in Large Language Models (LLMs), their application domains have expanded to include code generation tasks across various programming languages. While significant progress has been made in enhancing LLMs for popular programming languages, there exists a notable gap in comprehensive evaluation frameworks tailored for Hardware Description Languages (HDLs), particularly VHDL. This paper addresses this gap by introducing a comprehensive evaluation framework designed specifically for assessing LLM performance in VHDL code generation task. We construct a dataset for evaluating LLMs on VHDL code generation task. This dataset is constructed by translating a collection of Verilog evaluation problems to VHDL and aggregating publicly available VHDL problems, resulting in a total of 202 problems. To assess the functional correctness of the generated VHDL code, we utilize a curated set of self-verifying testbenches specifically designed for those aggregated VHDL problem set. We conduct an initial evaluation of different LLMs and their variants, including zero-shot code generation, in-context learning (ICL), and Parameter-efficient fine-tuning (PEFT) methods. Our findings underscore the considerable challenges faced by existing LLMs in VHDL code generation, revealing significant scope for improvement. This study emphasizes the necessity of supervised fine-tuning code generation models specifically for VHDL, offering potential benefits to VHDL designers seeking efficient code generation solutions.
Authors: Rylan Schaeffer, Hailey Schoelkopf, Brando Miranda, Gabriel Mukobi, Varun Madan, Adam Ibrahim, Herbie Bradley, Stella Biderman, Sanmi Koyejo
Abstract: Predictable behavior from scaling advanced AI systems is an extremely desirable property. Although a well-established literature exists on how pretraining performance scales, the literature on how particular downstream capabilities scale is significantly muddier. In this work, we take a step back and ask: why has predicting specific downstream capabilities with scale remained elusive? While many factors are certainly responsible, we identify a new factor that makes modeling scaling behavior on widely used multiple-choice question-answering benchmarks challenging. Using five model families and twelve well-established multiple-choice benchmarks, we show that downstream performance is computed from negative log likelihoods via a sequence of transformations that progressively degrade the statistical relationship between performance and scale. We then reveal the mechanism causing this degradation: downstream metrics require comparing the correct choice against a small number of specific incorrect choices, meaning accurately predicting downstream capabilities requires predicting not just how probability mass concentrates on the correct choice with scale, but also how probability mass fluctuates on specific incorrect choices with scale. We empirically study how probability mass on the correct choice co-varies with probability mass on incorrect choices with increasing compute, suggesting that scaling laws for incorrect choices might be achievable. Our work also explains why pretraining scaling laws are commonly regarded as more predictable than downstream capabilities and contributes towards establishing scaling-predictable evaluations of frontier AI models.
Authors: Dongyoung Kim, Kimin Lee, Jinwoo Shin, Jaehyung Kim
Abstract: Aligning large language models (LLMs) with human preferences becomes a key component to obtaining state-of-the-art performance, but it yields a huge cost to construct a large human-annotated preference dataset. To tackle this problem, we propose a new framework that boosts the alignment of LLMs through Self-generated Preference data (Selfie) using only a very small amount of human-annotated preference data. Our key idea is leveraging the human prior knowledge within the small (seed) data and progressively improving the alignment of LLM, by iteratively generating the responses and learning from them with the self-annotated preference data. To be specific, we propose to derive the preference label from the logits of LLM to explicitly extract the model's inherent preference. Compared to the previous approaches using external reward models or implicit in-context learning, we observe that the proposed approach is significantly more effective. In addition, we introduce a noise-aware preference learning algorithm to mitigate the risk of low quality within generated preference data. Our experimental results demonstrate that the proposed framework significantly boosts the alignment of LLMs. For example, we achieve superior alignment performance on AlpacaEval 2.0 with only 3.3\% of the ground-truth preference labels in the Ultrafeedback data compared to the cases using the entire data or state-of-the-art baselines.
Authors: Sreyan Ghosh, Sonal Kumar, Ashish Seth, Purva Chiniya, Utkarsh Tyagi, Ramani Duraiswami, Dinesh Manocha
Abstract: Visual cues, like lip motion, have been shown to improve the performance of Automatic Speech Recognition (ASR) systems in noisy environments. We propose LipGER (Lip Motion aided Generative Error Correction), a novel framework for leveraging visual cues for noise-robust ASR. Instead of learning the cross-modal correlation between the audio and visual modalities, we make an LLM learn the task of visually-conditioned (generative) ASR error correction. Specifically, we instruct an LLM to predict the transcription from the N-best hypotheses generated using ASR beam-search. This is further conditioned on lip motions. This approach addresses key challenges in traditional AVSR learning, such as the lack of large-scale paired datasets and difficulties in adapting to new domains. We experiment on 4 datasets in various settings and show that LipGER improves the Word Error Rate in the range of 1.1%-49.2%. We also release LipHyp, a large-scale dataset with hypothesis-transcription pairs that is additionally equipped with lip motion cues to promote further research in this space
Authors: Th\'eodor Lemerle, Nicolas Obin, Axel Roebel
Abstract: Recent advancements in text-to-speech (TTS) powered by language models have showcased remarkable capabilities in achieving naturalness and zero-shot voice cloning. Notably, the decoder-only transformer is the prominent architecture in this domain. However, transformers face challenges stemming from their quadratic complexity in sequence length, impeding training on lengthy sequences and resource-constrained hardware. Moreover they lack specific inductive bias with regards to the monotonic nature of TTS alignments. In response, we propose to replace transformers with emerging recurrent architectures and introduce specialized cross-attention mechanisms for reducing repeating and skipping issues. Consequently our architecture can be efficiently trained on long samples and achieve state-of-the-art zero-shot voice cloning against baselines of comparable size.
Authors: Abdelrahman Abdallah, Mahmoud Abdalla, Mahmoud SalahEldin Kasem, Mohamed Mahmoud, Ibrahim Abdelhalim, Mohamed Elkasaby, Yasser ElBendary, Adam Jatowt
Abstract: In the fields of Optical Character Recognition (OCR) and Natural Language Processing (NLP), integrating multilingual capabilities remains a critical challenge, especially when considering languages with complex scripts such as Arabic. This paper introduces the Comprehensive Post-OCR Parsing and Receipt Understanding Dataset (CORU), a novel dataset specifically designed to enhance OCR and information extraction from receipts in multilingual contexts involving Arabic and English. CORU consists of over 20,000 annotated receipts from diverse retail settings, including supermarkets and clothing stores, alongside 30,000 annotated images for OCR that were utilized to recognize each detected line, and 10,000 items annotated for detailed information extraction. These annotations capture essential details such as merchant names, item descriptions, total prices, receipt numbers, and dates. They are structured to support three primary computational tasks: object detection, OCR, and information extraction. We establish the baseline performance for a range of models on CORU to evaluate the effectiveness of traditional methods, like Tesseract OCR, and more advanced neural network-based approaches. These baselines are crucial for processing the complex and noisy document layouts typical of real-world receipts and for advancing the state of automated multilingual document processing. Our datasets are publicly accessible (https://github.com/Update-For-Integrated-Business-AI/CORU).
URLs: https://github.com/Update-For-Integrated-Business-AI/CORU).
Authors: Max Zhu, Adri\'an Bazaga, Pietro Li\`o
Abstract: Learning computational fluid dynamics (CFD) traditionally relies on computationally intensive simulations of the Navier-Stokes equations. Recently, large language models (LLMs) have shown remarkable pattern recognition and reasoning abilities in natural language processing (NLP) and computer vision (CV). However, these models struggle with the complex geometries inherent in fluid dynamics. We introduce FLUID-LLM, a novel framework combining pre-trained LLMs with spatiotemporal-aware encoding to predict unsteady fluid dynamics. Our approach leverages the temporal autoregressive abilities of LLMs alongside spatial-aware layers, bridging the gap between previous CFD prediction methods. Evaluations on standard benchmarks reveal significant performance improvements across various fluid datasets. Our results demonstrate that FLUID-LLM effectively integrates spatiotemporal information into pre-trained LLMs, enhancing CFD task performance.
Authors: Xintong Wang, Mingqian Shi, Ye Wang
Abstract: Mispronunciation Detection and Diagnosis (MDD) systems, leveraging Automatic Speech Recognition (ASR), face two main challenges in Mandarin Chinese: 1) The two-stage models create an information gap between the phoneme or tone classification stage and the MDD stage. 2) The scarcity of Mandarin MDD datasets limits model training. In this paper, we introduce a stateless RNN-T model for Mandarin MDD, utilizing HuBERT features with pitch embedding through a Pitch Fusion Block. Our model, trained solely on native speaker data, shows a 3% improvement in Phone Error Rate and a 7% increase in False Acceptance Rate over the state-of-the-art baseline in non-native scenarios
Authors: Enis Berk \c{C}oban, Michael I. Mandel, Johanna Devaney
Abstract: Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities, notably in connecting ideas and adhering to logical rules to solve problems. These models have evolved to accommodate various data modalities, including sound and images, known as multimodal LLMs (MLLMs), which are capable of describing images or sound recordings. Previous work has demonstrated that when the LLM component in MLLMs is frozen, the audio or visual encoder serves to caption the sound or image input facilitating text-based reasoning with the LLM component. We are interested in using the LLM's reasoning capabilities in order to facilitate classification. In this paper, we demonstrate through a captioning/classification experiment that an audio MLLM cannot fully leverage its LLM's text-based reasoning when generating audio captions. We also consider how this may be due to MLLMs separately representing auditory and textual information such that it severs the reasoning pathway from the LLM to the audio encoder.
Authors: Tobias Alt, Andrea Ibisch, Clemens Meiser, Anna Wilhelm, Raphael Zimmer, Christian Berghoff, Christoph Droste, Jens Karschau, Friederike Laus, Rainer Plaga, Carola Plesch, Britta Sennewald, Thomas Thaeren, Kristina Unverricht, Steffen Waurick
Abstract: Generative AI models are capable of performing a wide range of tasks that traditionally require creativity and human understanding. They learn patterns from existing data during training and can subsequently generate new content such as texts, images, and music that follow these patterns. Due to their versatility and generally high-quality results, they, on the one hand, represent an opportunity for digitalization. On the other hand, the use of generative AI models introduces novel IT security risks that need to be considered for a comprehensive analysis of the threat landscape in relation to IT security. In response to this risk potential, companies or authorities using them should conduct an individual risk analysis before integrating generative AI into their workflows. The same applies to developers and operators, as many risks in the context of generative AI have to be taken into account at the time of development or can only be influenced by the operating company. Based on this, existing security measures can be adjusted, and additional measures can be taken.
Authors: Eduard Poesina, Adriana Valentina Costache, Adrian-Gabriel Chifu, Josiane Mothe, Radu Tudor Ionescu
Abstract: Text-to-image generation has recently emerged as a viable alternative to text-to-image retrieval, due to the visually impressive results of generative diffusion models. Although query performance prediction is an active research topic in information retrieval, to the best of our knowledge, there is no prior study that analyzes the difficulty of queries (prompts) in text-to-image generation, based on human judgments. To this end, we introduce the first dataset of prompts which are manually annotated in terms of image generation performance. In order to determine the difficulty of the same prompts in image retrieval, we also collect manual annotations that represent retrieval performance. We thus propose the first benchmark for joint text-to-image prompt and query performance prediction, comprising 10K queries. Our benchmark enables: (i) the comparative assessment of the difficulty of prompts/queries in image generation and image retrieval, and (ii) the evaluation of prompt/query performance predictors addressing both generation and retrieval. We present results with several pre-generation/retrieval and post-generation/retrieval performance predictors, thus providing competitive baselines for future research. Our benchmark and code is publicly available under the CC BY 4.0 license at https://github.com/Eduard6421/PQPP.
Authors: Weizhi Tang, Vaishak Belle
Abstract: Large Language Models (LLMs) have recently shown a promise and emergence of Theory of Mind (ToM) ability and even outperform humans in certain ToM tasks. To evaluate and extend the boundaries of the ToM reasoning ability of LLMs, we propose a novel concept, taxonomy, and framework, the ToM reasoning with Zero, Finite, and Infinite Belief History and develop a multi-round text-based game, called $\textit{Pick the Right Stuff}$, as a benchmark. We have evaluated six LLMs with this game and found their performance on Zero Belief History is consistently better than on Finite Belief History. In addition, we have found two of the models with small parameter sizes outperform all the evaluated models with large parameter sizes. We expect this work to pave the way for future ToM benchmark development and also for the promotion and development of more complex AI agents or systems which are required to be equipped with more complex ToM reasoning ability.
Authors: Emilia Lesiak, Grzegorz Wolny, Bartosz Przyby{\l}, Micha{\l} Szczerbak
Abstract: This article investigates the deployment of a Voice User Interface (VUI)-powered digital assistant in a retail setting and assesses its impact on customer engagement and service efficiency. The study explores how digital assistants can enhance user interactions through advanced conversational capabilities with multilingual support. By integrating a digital assistant into a high-traffic retail environment, we evaluate its effectiveness in improving the quality of customer service and operational efficiency. Data collected during the experiment demonstrate varied impacts on customer interaction, revealing insights into the future optimizations of digital assistant technologies in customer-facing roles. This study contributes to the understanding of digital transformation strategies within the customer relations domain emphasizing the need for service flexibility and user-centric design in modern retail stores.
Authors: Yuxing Long, Wenzhe Cai, Hongcheng Wang, Guanqi Zhan, Hao Dong
Abstract: Enabling robots to navigate following diverse language instructions in unexplored environments is an attractive goal for human-robot interaction. However, this goal is challenging because different navigation tasks require different strategies. The scarcity of instruction navigation data hinders training an instruction navigation model with varied strategies. Therefore, previous methods are all constrained to one specific type of navigation instruction. In this work, we propose InstructNav, a generic instruction navigation system. InstructNav makes the first endeavor to handle various instruction navigation tasks without any navigation training or pre-built maps. To reach this goal, we introduce Dynamic Chain-of-Navigation (DCoN) to unify the planning process for different types of navigation instructions. Furthermore, we propose Multi-sourced Value Maps to model key elements in instruction navigation so that linguistic DCoN planning can be converted into robot actionable trajectories. With InstructNav, we complete the R2R-CE task in a zero-shot way for the first time and outperform many task-training methods. Besides, InstructNav also surpasses the previous SOTA method by 10.48% on the zero-shot Habitat ObjNav and by 86.34% on demand-driven navigation DDN. Real robot experiments on diverse indoor scenes further demonstrate our method's robustness in coping with the environment and instruction variations.
Authors: Abisek Rajakumar Kalarani, Pushpak Bhattacharyya, Sumit Shekhar
Abstract: Metaphors are a common communication tool used in our day-to-day life. The detection and generation of metaphors in textual form have been studied extensively but metaphors in other forms have been under-explored. Recent studies have shown that Vision-Language (VL) models cannot understand visual metaphors in memes and adverts. As of now, no probing studies have been done that involve complex language phenomena like metaphors with videos. Hence, we introduce a new VL task of describing the metaphors present in the videos in our work. To facilitate this novel task, we construct and release a manually created dataset with 705 videos and 2115 human-written captions, along with a new metric called Average Concept Distance (ACD), to automatically evaluate the creativity of the metaphors generated. We also propose a novel low-resource video metaphor captioning system: GIT-LLaVA, which obtains comparable performance to SoTA video language models on the proposed task. We perform a comprehensive analysis of existing video language models on this task and publish our dataset, models, and benchmark results to enable further research.
Authors: Edresson Casanova, Kelly Davis, Eren G\"olge, G\"orkem G\"oknar, Iulian Gulea, Logan Hart, Aya Aljafari, Joshua Meyer, Reuben Morais, Samuel Olayemi, Julian Weber
Abstract: Most Zero-shot Multi-speaker TTS (ZS-TTS) systems support only a single language. Although models like YourTTS, VALL-E X, Mega-TTS 2, and Voicebox explored Multilingual ZS-TTS they are limited to just a few high/medium resource languages, limiting the applications of these models in most of the low/medium resource languages. In this paper, we aim to alleviate this issue by proposing and making publicly available the XTTS system. Our method builds upon the Tortoise model and adds several novel modifications to enable multilingual training, improve voice cloning, and enable faster training and inference. XTTS was trained in 16 languages and achieved state-of-the-art (SOTA) results in most of them.
Authors: Georgios Efstathiadis, Vijay Yadav, Anzar Abbas
Abstract: Speaker diarization is necessary for interpreting conversations transcribed using automated speech recognition (ASR) tools. Despite significant developments in diarization methods, diarization accuracy remains an issue. Here, we investigate the use of large language models (LLMs) for diarization correction as a post-processing step. LLMs were fine-tuned using the Fisher corpus, a large dataset of transcribed conversations. The ability of the models to improve diarization accuracy in a holdout dataset was measured. We report that fine-tuned LLMs can markedly improve diarization accuracy. However, model performance is constrained to transcripts produced using the same ASR tool as the transcripts used for fine-tuning, limiting generalizability. To address this constraint, an ensemble model was developed by combining weights from three separate models, each fine-tuned using transcripts from a different ASR tool. The ensemble model demonstrated better overall performance than each of the ASR-specific models, suggesting that a generalizable and ASR-agnostic approach may be achievable. We hope to make these models accessible through public-facing APIs for use by third-party applications.
Authors: Yani Zhang, Dongming Wu, Wencheng Han, Xingping Dong
Abstract: Referring multi-object tracking (RMOT) aims at detecting and tracking multiple objects following human instruction represented by a natural language expression. Existing RMOT benchmarks are usually formulated through manual annotations, integrated with static regulations. This approach results in a dearth of notable diversity and a constrained scope of implementation. In this work, our key idea is to bootstrap the task of referring multi-object tracking by introducing discriminative language words as much as possible. In specific, we first develop Refer-KITTI into a large-scale dataset, named Refer-KITTI-V2. It starts with 2,719 manual annotations, addressing the issue of class imbalance and introducing more keywords to make it closer to real-world scenarios compared to Refer-KITTI. They are further expanded to a total of 9,758 annotations by prompting large language models, which create 617 different words, surpassing previous RMOT benchmarks. In addition, the end-to-end framework in RMOT is also bootstrapped by a simple yet elegant temporal advancement strategy, which achieves better performance than previous approaches. The source code and dataset is available at https://github.com/zyn213/TempRMOT.
Authors: Francesco Taioli, Stefano Rosa, Alberto Castellini, Lorenzo Natale, Alessio Del Bue, Alessandro Farinelli, Marco Cristani, Yiming Wang
Abstract: In the Vision-and-Language Navigation in Continuous Environments (VLN-CE) task, the human user guides an autonomous agent to reach a target goal via a series of low-level actions following a textual instruction in natural language. However, most existing methods do not address the likely case where users may make mistakes when providing such instruction (e.g. "turn left" instead of "turn right"). In this work, we address a novel task of Interactive VLN in Continuous Environments (IVLN-CE), which allows the agent to interact with the user during the VLN-CE navigation to verify any doubts regarding the instruction errors. We propose an Interactive Instruction Error Detector and Localizer (I2EDL) that triggers the user-agent interaction upon the detection of instruction errors during the navigation. We leverage a pre-trained module to detect instruction errors and pinpoint them in the instruction by cross-referencing the textual input and past observations. In such way, the agent is able to query the user for a timely correction, without demanding the user's cognitive load, as we locate the probable errors to a precise part of the instruction. We evaluate the proposed I2EDL on a dataset of instructions containing errors, and further devise a novel metric, the Success weighted by Interaction Number (SIN), to reflect both the navigation performance and the interaction effectiveness. We show how the proposed method can ask focused requests for corrections to the user, which in turn increases the navigation success, while minimizing the interactions.
Authors: Pompeu Casanovas
Abstract: This is a long paper, an essay, on ambiguity, pragmatics, legal ecosystems, and the expressive function of law. It is divided into two parts and fifteen sections. The first part (Pragmatics) addresses ambiguity from the perspective of linguistic and cognitive pragmatics in the legal field. The second part (Computing) deals with this issue from the point of view of human-centered design and artificial intelligence, specifically focusing on the notion and modelling of rules and what it means to comply with the rules. This is necessary for the scaffolding of smart legal ecosystems (SLE). I will develop this subject with the example of the architecture, information flows, and smart ecosystem of OPTIMAI, an EU project of Industry 4.0 for zero-defect manufacturing (Optimizing Manufacturing Processes through Artificial Intelligence and Virtualization).
Authors: Jianing Yang, Xuweiyi Chen, Nikhil Madaan, Madhavan Iyengar, Shengyi Qian, David F. Fouhey, Joyce Chai
Abstract: The integration of language and 3D perception is crucial for developing embodied agents and robots that comprehend and interact with the physical world. While large language models (LLMs) have demonstrated impressive language understanding and generation capabilities, their adaptation to 3D environments (3D-LLMs) remains in its early stages. A primary challenge is the absence of large-scale datasets that provide dense grounding between language and 3D scenes. In this paper, we introduce 3D-GRAND, a pioneering large-scale dataset comprising 40,087 household scenes paired with 6.2 million densely-grounded scene-language instructions. Our results show that instruction tuning with 3D-GRAND significantly enhances grounding capabilities and reduces hallucinations in 3D-LLMs. As part of our contributions, we propose a comprehensive benchmark 3D-POPE to systematically evaluate hallucination in 3D-LLMs, enabling fair comparisons among future models. Our experiments highlight a scaling effect between dataset size and 3D-LLM performance, emphasizing the critical role of large-scale 3D-text datasets in advancing embodied AI research. Notably, our results demonstrate early signals for effective sim-to-real transfer, indicating that models trained on large synthetic data can perform well on real-world 3D scans. Through 3D-GRAND and 3D-POPE, we aim to equip the embodied AI community with essential resources and insights, setting the stage for more reliable and better-grounded 3D-LLMs. Project website: https://3d-grand.github.io
Authors: Shizhe Diao, Pengcheng Wang, Yong Lin, Tong Zhang
Abstract: The increasing scale of large language models (LLMs) brings emergent abilities to various complex tasks requiring reasoning, such as arithmetic and commonsense reasoning. It is known that the effective design of task-specific prompts is critical for LLMs' ability to produce high-quality answers. In particular, an effective approach for complex question-and-answer tasks is example-based prompting with chain-of-thought (CoT) reasoning, which significantly improves the performance of LLMs. However, current CoT methods rely on a fixed set of human-annotated exemplars, which are not necessarily the most effective examples for different tasks. This paper proposes a new method, Active-Prompt, to adapt LLMs to different tasks with task-specific example prompts (annotated with human-designed CoT reasoning). For this purpose, we propose a solution to the key problem of determining which questions are the most important and helpful ones to annotate from a pool of task-specific queries. By borrowing ideas from the related problem of uncertainty-based active learning, we introduce several metrics to characterize the uncertainty so as to select the most uncertain questions for annotation. Experimental results demonstrate the superiority of our proposed method, achieving state-of-the-art on eight complex reasoning tasks. Further analyses of different uncertainty metrics, pool sizes, zero-shot learning, and accuracy-uncertainty relationship demonstrate the effectiveness of our method. Our code will be available at https://github.com/shizhediao/active-prompt.
Authors: Sidi Lu, Wenbo Zhao, Chenyang Tao, Arpit Gupta, Shanchan Wu, Tagyoung Chung, Nanyun Peng
Abstract: NeurAlly-Decomposed Oracle (NADO) is a powerful approach for controllable generation with large language models. It is designed to avoid catastrophic forgetting while achieving guaranteed convergence to an entropy-maximized closed-form optimal solution with reasonable modeling capacity. Despite the success, several challenges arise when apply NADO to a wide range of scenarios. Vanilla NADO suffers from gradient vanishing for low-probability control signals and is highly reliant on a regularization to satisfy the stochastic version of Bellman equation. In addition, the vanilla implementation of NADO introduces a few additional transformer layers, suffering from a limited capacity especially compared to other finetune-based model adaptation methods like LoRA. In this paper, we propose a improved version of the NADO algorithm, namely DiNADO (norm-Disentangled NeurAlly-Decomposed Oracles), which improves the performance of the NADO algorithm through disentangling the step-wise global norm over the approximated oracle $R$-value for all potential next-tokens, allowing DiNADO to be combined with finetuning methods like LoRA. We discuss in depth how DiNADO achieves better capacity, stability and flexibility with both empirical and theoretical results. Experiments on formality control in machine translation and the lexically constrained generation task CommonGen demonstrates the significance of the improvements.
Authors: Sidi Lu, Hongyi Liu, Asli Celikyilmaz, Tianlu Wang, Nanyun Peng
Abstract: Recent advancements in open-domain text generation, driven by the power of large pre-trained language models (LLMs), have demonstrated remarkable performance. However, assessing these models' generation quality remains a challenge. In this paper, we introduce a novel method for evaluating open-domain text generation called Contrastive Distribution Methods (CDM). Leveraging the connection between increasing model parameters and enhanced LLM performance, CDM creates a mapping from the _contrast_ of two probabilistic distributions -- one known to be superior to the other -- to quality measures. We investigate CDM for open-domain text generation evaluation under two paradigms: 1) _Generative_ CDM, which harnesses the contrast of two language models' distributions to generate synthetic examples for training discriminator-based metrics; 2) _Discriminative_ CDM, which directly uses distribution disparities between two language models for evaluation. Our experiments on coherence evaluation for multi-turn dialogue and commonsense evaluation for controllable generation demonstrate CDM's superior correlate with human judgment than existing automatic evaluation metrics, highlighting the strong performance and generalizability of our approach.
Authors: Zhuosheng Zhang, Aston Zhang
Abstract: Autonomous graphical user interface (GUI) agents aim to facilitate task automation by interacting with the user interface without manual intervention. Recent studies have investigated eliciting the capabilities of large language models (LLMs) for effective engagement in diverse environments. To align with the input-output requirement of LLMs, most existing approaches are developed under a sandbox setting where they rely on external tools and application-specific APIs to parse the environment into textual elements and interpret the predicted actions. Consequently, those approaches often grapple with inference inefficiency and error propagation risks. To mitigate the challenges, we introduce Auto-GUI, a multimodal solution that directly interacts with the interface, bypassing the need for environment parsing or reliance on application-dependent APIs. Moreover, we propose a chain-of-action technique -- leveraging a series of intermediate previous action histories and future action plans -- to help the agent decide what action to execute. We evaluate our approach on a new device-control benchmark AITW with 30$K$ unique instructions, spanning multi-step tasks such as application operation, web searching, and web shopping. Experimental results show that Auto-GUI achieves state-of-the-art performance with an action type prediction accuracy of 90\% and an overall action success rate of 74\%. Code is publicly available at https://github.com/cooelf/Auto-GUI.
Authors: Shih-Cheng Huang, Pin-Zu Li, Yu-Chi Hsu, Kuang-Ming Chen, Yu Tung Lin, Shih-Kai Hsiao, Richard Tzong-Han Tsai, Hung-yi Lee
Abstract: Recently, the development of open-source large language models (LLMs) has advanced rapidly. Nevertheless, due to data constraints, the capabilities of most open-source LLMs are primarily focused on English. To address this issue, we introduce the concept of $\textit{chat vector}$ to equip pre-trained language models with instruction following and human value alignment via simple model arithmetic. The chat vector is derived by subtracting the weights of a pre-trained base model (e.g. LLaMA2) from those of its corresponding chat model (e.g. LLaMA2-chat). By simply adding the chat vector to a continual pre-trained model's weights, we can endow the model with chat capabilities in new languages without the need for further training. Our empirical studies demonstrate the superior efficacy of the chat vector from three different aspects: instruction following, toxicity mitigation, and multi-turn dialogue. Moreover, to showcase the adaptability of our approach, we extend our experiments to encompass various languages, base models, and chat vectors. The results underscore the chat vector's simplicity, effectiveness, and wide applicability, making it a compelling solution for efficiently enabling conversational capabilities in pre-trained language models. Our code is available at https://github.com/aqweteddy/ChatVector.
Authors: Guanting Dong, Hongyi Yuan, Keming Lu, Chengpeng Li, Mingfeng Xue, Dayiheng Liu, Wei Wang, Zheng Yuan, Chang Zhou, Jingren Zhou
Abstract: Large language models (LLMs) with enormous pre-training tokens and parameters emerge diverse abilities, including math reasoning, code generation, and instruction following. These abilities are further enhanced by supervised fine-tuning (SFT). While the open-source community has explored ad-hoc SFT for enhancing individual capabilities, proprietary LLMs exhibit versatility across various skills. Therefore, understanding the facilitation of multiple abilities via SFT is paramount. In this study, we specifically focuses on the interplay of data composition between mathematical reasoning, code generation, and general human-aligning abilities during SFT. We propose four intriguing research questions to explore the association between model performance and various factors including data amount, composition ratio, model size and SFT strategies. Our experiments reveal that distinct capabilities scale differently and larger models generally show superior performance with same amount of data. Mathematical reasoning and code generation consistently improve with increasing data amount, whereas general abilities plateau after roughly a thousand samples. Moreover, we observe data composition appears to enhance various abilities under limited data conditions, yet can lead to performance conflicts when data is plentiful. Our findings also suggest the amount of composition data influences performance more than the composition ratio. In analysis of SFT strategies, we find that sequentially learning multiple skills risks catastrophic forgetting. Our proposed Dual-stage Mixed Fine-tuning (DMT) strategy offers a promising solution to learn multiple abilities with different scaling patterns.
Authors: Swarnadeep Saha, Omer Levy, Asli Celikyilmaz, Mohit Bansal, Jason Weston, Xian Li
Abstract: Large Language Models (LLMs) are frequently used for multi-faceted language generation and evaluation tasks that involve satisfying intricate user constraints or taking into account multiple aspects and criteria. However, their performance can fall short, due to the model's lack of coherence and inability to plan and decompose the problem. We propose Branch-Solve-Merge (BSM), a Large Language Model program (Schlag et al., 2023) for tackling such challenging natural language tasks. It consists of branch, solve, and merge modules that are parameterized with specific prompts to the base LLM. These three modules plan a decomposition of the task into multiple parallel sub-tasks, independently solve them, and fuse the solutions to the sub-tasks. We apply our method to the tasks of LLM response evaluation and constrained text generation and evaluate its effectiveness with multiple LLMs, including Vicuna, LLaMA-2-chat, and GPT-4. BSM improves the evaluation correctness and consistency for each LLM by enhancing human-LLM agreement by up to 26%, reducing length and pairwise position biases by up to 50%, and allowing LLaMA2-chat to match or outperform GPT-4 on most domains. On a constraint story generation task, BSM improves the coherence of stories while also improving constraint satisfaction by 12%.
Authors: Xintao Wang, Yunze Xiao, Jen-tse Huang, Siyu Yuan, Rui Xu, Haoran Guo, Quan Tu, Yaying Fei, Ziang Leng, Wei Wang, Jiangjie Chen, Cheng Li, Yanghua Xiao
Abstract: Role-playing agents (RPAs), powered by large language models, have emerged as a flourishing field of applications. However, a key challenge lies in assessing whether RPAs accurately reproduce the personas of target characters, namely their character fidelity. Existing methods mainly focus on the knowledge and linguistic patterns of characters. This paper, instead, introduces a novel perspective to evaluate the personality fidelity of RPAs with psychological scales. Overcoming drawbacks of previous self-report assessments on RPAs, we propose InCharacter, namely Interviewing Character agents for personality tests. Experiments include various types of RPAs and LLMs, covering 32 distinct characters on 14 widely used psychological scales. The results validate the effectiveness of InCharacter in measuring RPA personalities. Then, with InCharacter, we show that state-of-the-art RPAs exhibit personalities highly aligned with the human-perceived personalities of the characters, achieving an accuracy up to 80.7%.
Authors: Hieu Tran, Zhichao Yang, Zonghai Yao, Hong Yu
Abstract: To enhance the performance of large language models (LLMs) in biomedical natural language processing (BioNLP) by introducing a domain-specific instruction dataset and examining its impact when combined with multi-task learning principles. We created the BioInstruct, comprising 25,005 instructions to instruction-tune LLMs(LLaMA 1 & 2, 7B & 13B version). The instructions were created by prompting the GPT-4 language model with three-seed samples randomly drawn from an 80 human curated instructions. We employed Low-Rank Adaptation(LoRA) for parameter-efficient fine-tuning. We then evaluated these instruction-tuned LLMs on several BioNLP tasks, which can be grouped into three major categories: question answering(QA), information extraction(IE), and text generation(GEN). We also examined whether categories(e.g., QA, IE, and generation) of instructions impact model performance. Comparing with LLMs without instruction-tuned, our instruction-tuned LLMs demonstrated marked performance gains: 17.3% in QA, 5.7% in IE, and 96% in Generation tasks. Our 7B-parameter instruction-tuned LLaMA 1 model was competitive or even surpassed other LLMs in the biomedical domain that were also fine-tuned from LLaMA 1 with vast domain-specific data or a variety of tasks. Our results also show that the performance gain is significantly higher when instruction fine-tuning is conducted with closely related tasks. Our findings align with the observations of multi-task learning, suggesting the synergies between two tasks. The BioInstruct dataset serves as a valuable resource and instruction tuned LLMs lead to the best performing BioNLP applications.
Authors: Nafis Irtiza Tripto, Saranya Venkatraman, Dominik Macko, Robert Moro, Ivan Srba, Adaku Uchendu, Thai Le, Dongwon Lee
Abstract: In the realm of text manipulation and linguistic transformation, the question of authorship has been a subject of fascination and philosophical inquiry. Much like the Ship of Theseus paradox, which ponders whether a ship remains the same when each of its original planks is replaced, our research delves into an intriguing question: Does a text retain its original authorship when it undergoes numerous paraphrasing iterations? Specifically, since Large Language Models (LLMs) have demonstrated remarkable proficiency in both the generation of original content and the modification of human-authored texts, a pivotal question emerges concerning the determination of authorship in instances where LLMs or similar paraphrasing tools are employed to rephrase the text--i.e., whether authorship should be attributed to the original human author or the AI-powered tool. Therefore, we embark on a philosophical voyage through the seas of language and authorship to unravel this intricate puzzle. Using a computational approach, we discover that the diminishing performance in text classification models, with each successive paraphrasing iteration, is closely associated with the extent of deviation from the original author's style, thus provoking a reconsideration of the current notion of authorship.
Authors: Weixiang Yan, Haitian Liu, Yunkun Wang, Yunzhe Li, Qian Chen, Wen Wang, Tingyu Lin, Weishan Zhao, Li Zhu, Hari Sundaram, Shuiguang Deng
Abstract: Large Language Models (LLMs) have demonstrated remarkable performance on assisting humans in programming and facilitating programming automation. However, existing benchmarks for evaluating the code understanding and generation capacities of LLMs suffer from severe limitations. First, most benchmarks are insufficient as they focus on a narrow range of popular programming languages and specific tasks, whereas real-world software development scenarios show a critical need to implement systems with multilingual and multitask programming environments to satisfy diverse requirements. Second, most benchmarks fail to consider the actual executability and the consistency of execution results of the generated code. To bridge these gaps between existing benchmarks and expectations from practical applications, we introduce CodeScope, an execution-based, multilingual, multitask, multidimensional evaluation benchmark for comprehensively measuring LLM capabilities on coding tasks. CodeScope covers 43 programming languages and eight coding tasks. It evaluates the coding performance of LLMs from three dimensions (perspectives): length, difficulty, and efficiency. To facilitate execution-based evaluations of code generation, we develop MultiCodeEngine, an automated code execution engine that supports 14 programming languages. Finally, we systematically evaluate and analyze eight mainstream LLMs and demonstrate the superior breadth and challenges of CodeScope for evaluating LLMs on code understanding and generation tasks compared to other benchmarks. The CodeScope benchmark and code are publicly available at https://github.com/WeixiangYAN/CodeScope.
Authors: Lei Shu, Nevan Wichers, Liangchen Luo, Yun Zhu, Yinxiao Liu, Jindong Chen, Lei Meng
Abstract: Evaluating natural language systems poses significant challenges, particularly in the realms of natural language understanding and high-level reasoning. In this paper, we introduce 'Fusion-Eval', an innovative approach that leverages Large Language Models (LLMs) to integrate insights from various assistant evaluators. The LLM is given the example to evaluate along with scores from the assistant evaluators. Each of these evaluators specializes in assessing distinct aspects of responses. Fusion-Eval achieves a 0.962 system-level Kendall-Tau correlation with humans on SummEval and a 0.744 turn-level Spearman correlation on TopicalChat, which is significantly higher than baseline methods. These results highlight Fusion-Eval's significant potential in the realm of natural language system evaluation.
Authors: Hanning Zhang, Shizhe Diao, Yong Lin, Yi R. Fung, Qing Lian, Xingyao Wang, Yangyi Chen, Heng Ji, Tong Zhang
Abstract: Large language models (LLMs) have revolutionized numerous domains with their impressive performance but still face their challenges. A predominant issue is the propensity for these models to generate non-existent facts, a concern termed hallucination. Our research is motivated by the observation that previous instruction tuning methods force the model to complete a sentence no matter whether the model knows the knowledge or not. When the question is out of the parametric knowledge, it will try to make up something and fail to indicate when it lacks knowledge. In this paper, we present a new approach called Refusal-Aware Instruction Tuning (R-Tuning). This approach is formalized by first identifying the disparity in knowledge encompassed by pre-trained parameters compared to that of instruction tuning data. Then, we construct the refusal-aware data based on the knowledge intersection, to tune LLMs to refrain from responding to questions beyond its parametric knowledge. Experimental results demonstrate R-Tuning effectively improves a model's ability to answer known questions and refrain from answering unknown questions. Furthermore, when tested on out-of-domain datasets, the refusal ability was found to be a meta-skill that could be generalized to other tasks. Further analysis surprisingly finds that learning the uncertainty results in better calibration and an improved ability to estimate the uncertainty than uncertainty-based testing. Our code is available at https://github.com/shizhediao/R-Tuning.
Authors: Yiqi Liu, Nafise Sadat Moosavi, Chenghua Lin
Abstract: Automatic evaluation of generated textual content presents an ongoing challenge within the field of NLP. Given the impressive capabilities of modern language models (LMs) across diverse NLP tasks, there is a growing trend to employ these models in creating innovative evaluation metrics for automated assessment of generation tasks. This paper investigates a pivotal question: Do language model-driven evaluation metrics inherently exhibit bias favoring texts generated by the same underlying language model? Specifically, we assess whether prominent LM-based evaluation metrics (e.g. BARTScore, T5Score, and GPTScore) demonstrate a favorable bias toward their respective underlying LMs in the context of summarization tasks. Our findings unveil a latent bias, particularly pronounced when such evaluation metrics are used in a reference-free manner without leveraging gold summaries. These results underscore that assessments provided by generative evaluation models can be influenced by factors beyond the inherent text quality, highlighting the necessity of developing more reliable evaluation protocols in the future.
Authors: Yinya Huang, Ruixin Hong, Hongming Zhang, Wei Shao, Zhicheng Yang, Dong Yu, Changshui Zhang, Xiaodan Liang, Linqi Song
Abstract: In this study, we delve into the realm of counterfactual reasoning capabilities of large language models (LLMs). Our primary objective is to cultivate the counterfactual thought processes within LLMs and rigorously assess these processes for their validity. Specifically, we introduce a novel task, Counterfactual Logical Modification (CLOMO), and a high-quality human-annotated benchmark. In this task, LLMs must adeptly alter a given argumentative text to uphold a predetermined logical relationship. To effectively evaluate a generation model's counterfactual capabilities, we propose an innovative evaluation metric, the decomposed Self-Evaluation Score (SES) to directly evaluate the natural language output of LLMs instead of modeling the task as a multiple-choice problem. Analysis shows that the proposed automatic metric aligns well with human preference. Our experimental results show that while LLMs demonstrate a notable capacity for logical counterfactual thinking, there remains a discernible gap between their current abilities and human performance. Code and data are available at https://github.com/Eleanor-H/CLOMO.
Authors: Yuxiang Wei, Zhe Wang, Jiawei Liu, Yifeng Ding, Lingming Zhang
Abstract: We introduce Magicoder, a series of fully open-source (code, weights, and data) Large Language Models (LLMs) for code that significantly closes the gap with top code models while having no more than 7B parameters. Magicoder models are trained on 75K synthetic instruction data using OSS-Instruct, a novel approach to enlightening LLMs with open-source code snippets to generate diverse instruction data for code. Our main motivation is to mitigate the inherent bias of the synthetic data generated by LLMs through the wealth of open-source references for the production of more realistic and controllable data. The orthogonality of OSS-Instruct and other data generation methods like Evol-Instruct further enables us to build an enhanced MagicoderS. Both Magicoder and MagicoderS substantially outperform state-of-the-art code models with similar or even larger sizes on a wide range of coding benchmarks. Notably, MagicoderS-CL-7B based on CodeLlama even surpasses the prominent ChatGPT on HumanEval+ (66.5 vs. 65.9 in pass@1 ). Overall, OSS-Instruct opens a new direction for crafting diverse synthetic instruction data for code using abundant open-source references.
Authors: Yu Fu, Yufei Li, Wen Xiao, Cong Liu, Yue Dong
Abstract: Recent developments in balancing the usefulness and safety of Large Language Models (LLMs) have raised a critical question: Are mainstream NLP tasks adequately aligned with safety consideration? Our study, focusing on safety-sensitive documents obtained through adversarial attacks, reveals significant disparities in the safety alignment of various NLP tasks. For instance, LLMs can effectively summarize malicious long documents but often refuse to translate them. This discrepancy highlights a previously unidentified vulnerability: attacks exploiting tasks with weaker safety alignment, like summarization, can potentially compromise the integrity of tasks traditionally deemed more robust, such as translation and question-answering (QA). Moreover, the concurrent use of multiple NLP tasks with lesser safety alignment increases the risk of LLMs inadvertently processing harmful content. We demonstrate these vulnerabilities in various safety-aligned LLMs, particularly Llama2 models, Gemini and GPT-4, indicating an urgent need for strengthening safety alignments across a broad spectrum of NLP tasks.
Authors: Zhaojian Yu, Xin Zhang, Ning Shang, Yangyu Huang, Can Xu, Yishujie Zhao, Wenxiang Hu, Qiufeng Yin
Abstract: Recent work demonstrates that, after instruction tuning, Code Large Language Models (Code LLMs) can obtain impressive capabilities to address a wide range of code-related tasks. However, current instruction tuning methods for Code LLMs mainly focus on the traditional code generation task, resulting in poor performance in complex multi-task scenarios. In this paper, we concentrate on multiple code-related tasks and present WaveCoder, a series of Code LLMs trained with Widespread And Versatile Enhanced instruction data. To enable the models to tackle complex code-related tasks, we propose a method to stably generate diverse, high-quality instruction data from open source code dataset in multi-task scenarios and obtain CodeSeaXDataset, a dataset comprising 19,915 instruction instances across 4 code-related tasks, which is aimed at improving the generalization ability of Code LLM. Our experiments demonstrate that WaveCoder models significantly outperform other open-source models in terms of the generalization ability across different code-related tasks. Moreover, WaveCoder-Ultra-6.7B presents the state-of-the-art generalization abilities on a wide range of code-related tasks.
Authors: Wenqi Zhang, Yongliang Shen, Linjuan Wu, Qiuying Peng, Jun Wang, Yueting Zhuang, Weiming Lu
Abstract: The reflection capacity of Large Language Model (LLM) has garnered extensive attention. A post-hoc prompting strategy, e.g., reflexion and self-refine, refines LLM's response based on self-evaluated or external feedback. However, recent research indicates without external feedback, LLM's intrinsic reflection is unstable. Our investigation unveils that the key bottleneck is the quality of the self-evaluated feedback. We find LLMs often exhibit overconfidence or high randomness when self-evaluate, offering stubborn or inconsistent feedback, which causes poor reflection. To remedy this, we advocate Self-Contrast: It adaptively explores diverse solving perspectives tailored to the request, contrasts the differences, and summarizes these discrepancies into a checklist which could be used to re-examine and eliminate discrepancies. Our method endows LLM with diverse perspectives to alleviate stubborn biases. Moreover, their discrepancies indicate potential errors or inherent uncertainties that LLM often overlooks. Reflecting upon these can catalyze more accurate and stable reflection. Experiments conducted on a series of reasoning and translation tasks with different LLMs serve to underscore the effectiveness and generality of our strategy.
Authors: Asma Ghandeharioun, Avi Caciularu, Adam Pearce, Lucas Dixon, Mor Geva
Abstract: Understanding the internal representations of large language models (LLMs) can help explain models' behavior and verify their alignment with human values. Given the capabilities of LLMs in generating human-understandable text, we propose leveraging the model itself to explain its internal representations in natural language. We introduce a framework called Patchscopes and show how it can be used to answer a wide range of questions about an LLM's computation. We show that many prior interpretability methods based on projecting representations into the vocabulary space and intervening on the LLM computation can be viewed as instances of this framework. Moreover, several of their shortcomings such as failure in inspecting early layers or lack of expressivity can be mitigated by Patchscopes. Beyond unifying prior inspection techniques, Patchscopes also opens up new possibilities such as using a more capable model to explain the representations of a smaller model, and multihop reasoning error correction.
Authors: Anton Voronov, Lena Wolf, Max Ryabinin
Abstract: Large language models demonstrate a remarkable capability for learning to solve new tasks from a few examples. The prompt template, or the way the input examples are formatted to obtain the prompt, is an important yet often overlooked aspect of in-context learning. In this work, we conduct a comprehensive study of the template format's influence on the in-context learning performance. We evaluate the impact of the prompt template across 21 models (from 770M to 70B parameters) and 4 standard classification datasets. We show that a poor choice of the template can reduce the performance of the strongest models and inference methods to a random guess level. More importantly, the best templates do not transfer between different setups and even between models of the same family. Our findings show that the currently prevalent approach to evaluation, which ignores template selection, may give misleading results due to different templates in different works. As a first step towards mitigating this issue, we propose Template Ensembles that aggregate model predictions across several templates. This simple test-time augmentation boosts average performance while being robust to the choice of random set of templates.
Authors: Hongliang He, Wenlin Yao, Kaixin Ma, Wenhao Yu, Yong Dai, Hongming Zhang, Zhenzhong Lan, Dong Yu
Abstract: The rapid advancement of large language models (LLMs) has led to a new era marked by the development of autonomous applications in real-world scenarios, which drives innovation in creating advanced web agents. Existing web agents typically only handle one input modality and are evaluated only in simplified web simulators or static web snapshots, greatly limiting their applicability in real-world scenarios. To bridge this gap, we introduce WebVoyager, an innovative Large Multimodal Model (LMM) powered web agent that can complete user instructions end-to-end by interacting with real-world websites. Moreover, we establish a new benchmark by compiling real-world tasks from 15 popular websites and introduce an automatic evaluation protocol leveraging multimodal understanding abilities of GPT-4V to evaluate open-ended web agents. We show that WebVoyager achieves a 59.1% task success rate on our benchmark, significantly surpassing the performance of both GPT-4 (All Tools) and the WebVoyager (text-only) setups, underscoring the exceptional capability of WebVoyager. The proposed automatic evaluation metric achieves 85.3% agreement with human judgment, indicating its effectiveness in providing reliable and accurate assessments of web agents.
Authors: Luca Soldaini, Rodney Kinney, Akshita Bhagia, Dustin Schwenk, David Atkinson, Russell Authur, Ben Bogin, Khyathi Chandu, Jennifer Dumas, Yanai Elazar, Valentin Hofmann, Ananya Harsh Jha, Sachin Kumar, Li Lucy, Xinxi Lyu, Nathan Lambert, Ian Magnusson, Jacob Morrison, Niklas Muennighoff, Aakanksha Naik, Crystal Nam, Matthew E. Peters, Abhilasha Ravichander, Kyle Richardson, Zejiang Shen, Emma Strubell, Nishant Subramani, Oyvind Tafjord, Pete Walsh, Luke Zettlemoyer, Noah A. Smith, Hannaneh Hajishirzi, Iz Beltagy, Dirk Groeneveld, Jesse Dodge, Kyle Lo
Abstract: Information about pretraining corpora used to train the current best-performing language models is seldom discussed: commercial models rarely detail their data, and even open models are often released without accompanying training data or recipes to reproduce them. As a result, it is challenging to conduct and advance scientific research on language modeling, such as understanding how training data impacts model capabilities and limitations. To facilitate scientific research on language model pretraining, we curate and release Dolma, a three-trillion-token English corpus, built from a diverse mixture of web content, scientific papers, code, public-domain books, social media, and encyclopedic materials. We extensively document Dolma, including its design principles, details about its construction, and a summary of its contents. We present analyses and experimental results on intermediate states of Dolma to share what we have learned about important data curation practices. Finally, we open-source our data curation toolkit to enable reproduction of our work as well as support further research in large-scale data curation.
Authors: Xingyao Wang, Yangyi Chen, Lifan Yuan, Yizhe Zhang, Yunzhu Li, Hao Peng, Heng Ji
Abstract: Large Language Model (LLM) agents, capable of performing a broad range of actions, such as invoking tools and controlling robots, show great potential in tackling real-world challenges. LLM agents are typically prompted to produce actions by generating JSON or text in a pre-defined format, which is usually limited by constrained action space (e.g., the scope of pre-defined tools) and restricted flexibility (e.g., inability to compose multiple tools). This work proposes to use executable Python code to consolidate LLM agents' actions into a unified action space (CodeAct). Integrated with a Python interpreter, CodeAct can execute code actions and dynamically revise prior actions or emit new actions upon new observations through multi-turn interactions. Our extensive analysis of 17 LLMs on API-Bank and a newly curated benchmark shows that CodeAct outperforms widely used alternatives (up to 20% higher success rate). The encouraging performance of CodeAct motivates us to build an open-source LLM agent that interacts with environments by executing interpretable code and collaborates with users using natural language. To this end, we collect an instruction-tuning dataset CodeActInstruct that consists of 7k multi-turn interactions using CodeAct. We show that it can be used with existing data to improve models in agent-oriented tasks without compromising their general capability. CodeActAgent, finetuned from Llama2 and Mistral, is integrated with Python interpreter and uniquely tailored to perform sophisticated tasks (e.g., model training) using existing libraries and autonomously self-debug.
Authors: Lijun Li, Bowen Dong, Ruohui Wang, Xuhao Hu, Wangmeng Zuo, Dahua Lin, Yu Qiao, Jing Shao
Abstract: In the rapidly evolving landscape of Large Language Models (LLMs), ensuring robust safety measures is paramount. To meet this crucial need, we propose \emph{SALAD-Bench}, a safety benchmark specifically designed for evaluating LLMs, attack, and defense methods. Distinguished by its breadth, SALAD-Bench transcends conventional benchmarks through its large scale, rich diversity, intricate taxonomy spanning three levels, and versatile functionalities.SALAD-Bench is crafted with a meticulous array of questions, from standard queries to complex ones enriched with attack, defense modifications and multiple-choice. To effectively manage the inherent complexity, we introduce an innovative evaluators: the LLM-based MD-Judge for QA pairs with a particular focus on attack-enhanced queries, ensuring a seamless, and reliable evaluation. Above components extend SALAD-Bench from standard LLM safety evaluation to both LLM attack and defense methods evaluation, ensuring the joint-purpose utility. Our extensive experiments shed light on the resilience of LLMs against emerging threats and the efficacy of contemporary defense tactics. Data and evaluator are released under https://github.com/OpenSafetyLab/SALAD-BENCH.
Authors: Cheng-Han Chiang, Hung-yi Lee
Abstract: Long-form generations from large language models (LLMs) contain a mix of factual and non-factual claims, making evaluating factuality difficult. Prior works evaluate the factuality of a long paragraph by decomposing it into multiple facts, verifying those facts independently, and aggregating the results. Such methods assume that combining factual claims forms a factual paragraph. The above assumption can be violated: we show that strong open-source models like Llama-chat can generate paragraphs that contain verifiable facts, but the facts are combined into a non-factual paragraph due to entity ambiguity. We further reveal that existing factuality metrics, including FActScore and citation recall, cannot properly evaluate these non-factual paragraphs and overestimate their factuality. To address this, we introduce an enhanced metric, D-FActScore, specifically designed for content with ambiguous entities. We evaluate the D-FActScores of people biographies generated by retrieval-augmented LLMs. We show that D-FActScore can better assess the factuality of paragraphs with entity ambiguity than FActScore. We also find that four widely used open-source LLMs tend to mix information of distinct entities to form non-factual paragraphs, making their D-FActScore much lower than FActScore by over 10%.
Authors: Yinhong Liu, Yimai Fang, David Vandyke, Nigel Collier
Abstract: In light of recent advances in large language models (LLMs), the expectations for the next generation of virtual assistants include enhanced naturalness and adaptability across diverse usage scenarios. However, the creation of high-quality annotated data for Task-Oriented Dialog (TOD) is recognized to be slow and costly. To address these challenges, we introduce Task-Oriented Automatic Dialogs (TOAD), a novel and scalable TOD dataset along with its automatic generation pipeline. The TOAD dataset simulates realistic app context interaction and provide a variety of system response style options. Two aspects of system response styles are considered, verbosity level and users' expression mirroring. We benchmark TOAD on two response generation tasks, and the results show that modeling more verbose responses or responses without user expression mirroring is more challenging.
Authors: Jiaheng Wei, Yuanshun Yao, Jean-Francois Ton, Hongyi Guo, Andrew Estornell, Yang Liu
Abstract: LLM hallucination, i.e. generating factually incorrect yet seemingly convincing answers, is currently a major threat to the trustworthiness and reliability of LLMs. The first step towards solving this complicated problem is to measure it. However, existing hallucination metrics require having a benchmark dataset with gold-standard answers, i.e. "best" or "correct" answers written by humans. Such requirements make hallucination measurement costly and prone to human errors. In this work, we propose Factualness Evaluations via Weighting LLMs (FEWL), an innovative hallucination metric that is specifically designed for the scenario when gold-standard answers are absent. FEWL leverages the answers from off-the-shelf LLMs that serve as a proxy of gold-standard answers. The key challenge is how to quantify the expertise of reference LLMs resourcefully. We show FEWL has certain theoretical guarantees and demonstrate empirically it gives more accurate hallucination measures than naively using reference LLMs. We also show how to leverage FEWL to reduce hallucination through both in-context learning and supervised fine-tuning. Extensive experiment results on Truthful-QA, CHALE, and HaluEval datasets demonstrate the effectiveness of FEWL.
Authors: Zae Myung Kim, Kwang Hee Lee, Preston Zhu, Vipul Raheja, Dongyeop Kang
Abstract: With the advent of large language models (LLM), the line between human-crafted and machine-generated texts has become increasingly blurred. This paper delves into the inquiry of identifying discernible and unique linguistic properties in texts that were written by humans, particularly uncovering the underlying discourse structures of texts beyond their surface structures. Introducing a novel methodology, we leverage hierarchical parse trees and recursive hypergraphs to unveil distinctive discourse patterns in texts produced by both LLMs and humans. Empirical findings demonstrate that, although both LLMs and humans generate distinct discourse patterns influenced by specific domains, human-written texts exhibit more structural variability, reflecting the nuanced nature of human writing in different domains. Notably, incorporating hierarchical discourse features enhances binary classifiers' overall performance in distinguishing between human-written and machine-generated texts, even on out-of-distribution and paraphrased samples. This underscores the significance of incorporating hierarchical discourse features in the analysis of text patterns. The code and dataset are available at https://github.com/minnesotanlp/threads-of-subtlety.
Authors: Yuanzhen Xie, Xinzhou Jin, Tao Xie, MingXiong Lin, Liang Chen, Chenyun Yu, Lei Cheng, ChengXiang Zhuo, Bo Hu, Zang Li
Abstract: In-context learning of large-language models (LLMs) has achieved remarkable success in the field of natural language processing, while extensive case studies reveal that the single-step chain-of-thought prompting approach faces challenges such as attention diffusion and inadequate performance in complex tasks like text-to-SQL. To improve the contextual learning capabilities of LLMs in text-to-SQL, a workflow paradigm method is proposed, aiming to enhance the attention and problem-solving scope of LLMs through decomposition. Specifically, the information determination module for eliminating redundant information and the brand-new prompt structure based on problem classification greatly enhance the model's attention. Additionally, the inclusion of self-correction and active learning modules greatly expands the problem-solving scope of LLMs, hence improving the upper limit of LLM-based approaches. Extensive experiments conducted on three datasets demonstrate that our approach outperforms other methods by a significant margin. About 2-3 percentage point improvements compared to the existing baseline on the Spider Dev, Spider-Realistic, and Bird Dev datasets and new SOTA results on the Spider Test dataset are achieved. Our code is available on GitHub: \url{https://github.com/FlyingFeather/DEA-SQL}.
Authors: Yejun Yoon, Seunghyun Yoon, Kunwoo Park
Abstract: This paper addresses the critical challenge of assessing the representativeness of news thumbnail images, which often serve as the first visual engagement for readers when an article is disseminated on social media. We focus on whether a news image represents the actors discussed in the news text. To serve the challenge, we introduce NewsTT, a manually annotated dataset of 1000 news thumbnail images and text pairs. We found that the pretrained vision and language models, such as BLIP-2, struggle with this task. Since news subjects frequently involve named entities or proper nouns, the pretrained models could have a limited capability to match news actors' visual and textual appearances. We hypothesize that learning to contrast news text with its counterfactual, of which named entities are replaced, can enhance the cross-modal matching ability of vision and language models. We propose CFT-CLIP, a contrastive learning framework that updates vision and language bi-encoders according to the hypothesis. We found that our simple method can boost the performance for assessing news thumbnail representativeness, supporting our assumption. Code and data can be accessed at https://github.com/ssu-humane/news-images-acl24.
Authors: Francesco Ortu, Zhijing Jin, Diego Doimo, Mrinmaya Sachan, Alberto Cazzaniga, Bernhard Sch\"olkopf
Abstract: Interpretability research aims to bridge the gap between empirical success and our scientific understanding of the inner workings of large language models (LLMs). However, most existing research focuses on analyzing a single mechanism, such as how models copy or recall factual knowledge. In this work, we propose a formulation of competition of mechanisms, which focuses on the interplay of multiple mechanisms instead of individual mechanisms and traces how one of them becomes dominant in the final prediction. We uncover how and where mechanisms compete within LLMs using two interpretability methods: logit inspection and attention modification. Our findings show traces of the mechanisms and their competition across various model components and reveal attention positions that effectively control the strength of certain mechanisms. Code: https://github.com/francescortu/comp-mech. Data: https://huggingface.co/datasets/francescortu/comp-mech.
URLs: https://github.com/francescortu/comp-mech., https://huggingface.co/datasets/francescortu/comp-mech.
Authors: Shuzhou Yuan, Ercong Nie, Michael F\"arber, Helmut Schmid, Hinrich Sch\"utze
Abstract: Large Language Models (LLMs) exhibit strong In-Context Learning (ICL) capabilities when prompts with demonstrations are used. However, fine-tuning still remains crucial to further enhance their adaptability. Prompt-based fine-tuning proves to be an effective fine-tuning method in low-data scenarios, but high demands on computing resources limit its practicality. We address this issue by introducing a prompt-based parameter-efficient fine-tuning (PEFT) approach. GNNavi leverages insights into ICL's information flow dynamics, which indicates that label words act in prompts as anchors for information propagation. GNNavi employs a Graph Neural Network (GNN) layer to precisely guide the aggregation and distribution of information flow during the processing of prompts by hardwiring the desired information flow into the GNN. Our experiments on text classification tasks with GPT-2 and Llama2 show GNNavi surpasses standard prompt-based fine-tuning methods in few-shot settings by updating just 0.2% to 0.5% of parameters. We compare GNNavi with prevalent PEFT approaches, such as prefix tuning, LoRA and Adapter in terms of performance and efficiency. Our analysis reveals that GNNavi enhances information flow and ensures a clear aggregation process.
Authors: Fengqing Jiang, Zhangchen Xu, Luyao Niu, Zhen Xiang, Bhaskar Ramasubramanian, Bo Li, Radha Poovendran
Abstract: Safety is critical to the usage of large language models (LLMs). Multiple techniques such as data filtering and supervised fine-tuning have been developed to strengthen LLM safety. However, currently known techniques presume that corpora used for safety alignment of LLMs are solely interpreted by semantics. This assumption, however, does not hold in real-world applications, which leads to severe vulnerabilities in LLMs. For example, users of forums often use ASCII art, a form of text-based art, to convey image information. In this paper, we propose a novel ASCII art-based jailbreak attack and introduce a comprehensive benchmark Vision-in-Text Challenge (ViTC) to evaluate the capabilities of LLMs in recognizing prompts that cannot be solely interpreted by semantics. We show that five SOTA LLMs (GPT-3.5, GPT-4, Gemini, Claude, and Llama2) struggle to recognize prompts provided in the form of ASCII art. Based on this observation, we develop the jailbreak attack ArtPrompt, which leverages the poor performance of LLMs in recognizing ASCII art to bypass safety measures and elicit undesired behaviors from LLMs. ArtPrompt only requires black-box access to the victim LLMs, making it a practical attack. We evaluate ArtPrompt on five SOTA LLMs, and show that ArtPrompt can effectively and efficiently induce undesired behaviors from all five LLMs. Our code is available at https://github.com/uw-nsl/ArtPrompt.
Authors: Verena Blaschke, Christoph Purschke, Hinrich Sch\"utze, Barbara Plank
Abstract: Natural language processing (NLP) has largely focused on modelling standardized languages. More recently, attention has increasingly shifted to local, non-standardized languages and dialects. However, the relevant speaker populations' needs and wishes with respect to NLP tools are largely unknown. In this paper, we focus on dialects and regional languages related to German -- a group of varieties that is heterogeneous in terms of prestige and standardization. We survey speakers of these varieties (N=327) and present their opinions on hypothetical language technologies for their dialects. Although attitudes vary among subgroups of our respondents, we find that respondents are especially in favour of potential NLP tools that work with dialectal input (especially audio input) such as virtual assistants, and less so for applications that produce dialectal output such as machine translation or spellcheckers.
Authors: Sahand Sabour, Siyang Liu, Zheyuan Zhang, June M. Liu, Jinfeng Zhou, Alvionna S. Sunaryo, Juanzi Li, Tatia M. C. Lee, Rada Mihalcea, Minlie Huang
Abstract: Recent advances in Large Language Models (LLMs) have highlighted the need for robust, comprehensive, and challenging benchmarks. Yet, research on evaluating their Emotional Intelligence (EI) is considerably limited. Existing benchmarks have two major shortcomings: first, they mainly focus on emotion recognition, neglecting essential EI capabilities such as emotion regulation and thought facilitation through emotion understanding; second, they are primarily constructed from existing datasets, which include frequent patterns, explicit information, and annotation errors, leading to unreliable evaluation. We propose EmoBench, a benchmark that draws upon established psychological theories and proposes a comprehensive definition for machine EI, including Emotional Understanding and Emotional Application. EmoBench includes a set of 400 hand-crafted questions in English and Chinese, which are meticulously designed to require thorough reasoning and understanding. Our findings reveal a considerable gap between the EI of existing LLMs and the average human, highlighting a promising direction for future research. Our code and data are publicly available at https://github.com/Sahandfer/EmoBench.
Authors: Miaoran Zhang, Vagrant Gautam, Mingyang Wang, Jesujoba O. Alabi, Xiaoyu Shen, Dietrich Klakow, Marius Mosbach
Abstract: In-context learning is a popular inference strategy where large language models solve a task using only a few labeled demonstrations without needing any parameter updates. Although there have been extensive studies on English in-context learning, multilingual in-context learning remains under-explored, and we lack an in-depth understanding of the role of demonstrations in this context. To address this gap, we conduct a multidimensional analysis of multilingual in-context learning, experimenting with 5 models from different model families, 9 datasets covering classification and generation tasks, and 56 typologically diverse languages. Our results reveal that the effectiveness of demonstrations varies significantly across models, tasks, and languages. We also find that strong instruction-following models including Llama 2-Chat, GPT-3.5, and GPT-4 are largely insensitive to the quality of demonstrations. Instead, a carefully crafted template often eliminates the benefits of demonstrations for some tasks and languages altogether. These findings show that the importance of demonstrations might be overestimated. Our work highlights the need for granular evaluation across multiple axes towards a better understanding of in-context learning.
Authors: Yuwei Xia, Ding Wang, Qiang Liu, Liang Wang, Shu Wu, Xiaoyu Zhang
Abstract: Temporal Knowledge Graph (TKG) forecasting aims to predict future facts based on given histories. Most recent graph-based models excel at capturing structural information within TKGs but lack semantic comprehension abilities. Nowadays, with the surge of LLMs, the LLM-based TKG prediction model has emerged. However, the existing LLM-based model exhibits three shortcomings: (1) It only focuses on the first-order history for prediction while ignoring high-order historical information, resulting in the provided information for LLMs being extremely limited. (2) LLMs struggle with optimal reasoning performance under heavy historical information loads. (3) For TKG prediction, the temporal reasoning capability of LLM alone is limited. To address the first two challenges, we propose Chain-of-History (CoH) reasoning which explores high-order histories step-by-step, achieving effective utilization of high-order historical information for LLMs on TKG prediction. To address the third issue, we design CoH as a plug-and-play module to enhance the performance of graph-based models for TKG prediction. Extensive experiments on three datasets and backbones demonstrate the effectiveness of CoH.
Authors: Kaijie Zhu, Jindong Wang, Qinlin Zhao, Ruochen Xu, Xing Xie
Abstract: Evaluation of large language models (LLMs) has raised great concerns in the community due to the issue of data contamination. Existing work designed evaluation protocols using well-defined algorithms for specific tasks, which cannot be easily extended to diverse scenarios. Moreover, current evaluation benchmarks can only provide the overall benchmark results and cannot support a fine-grained and multifaceted analysis of LLMs' abilities. In this paper, we propose meta probing agents (MPA), a general dynamic evaluation protocol inspired by psychometrics to evaluate LLMs. MPA is the key component of DyVal 2, which naturally extends the previous DyVal~\citep{zhu2023dyval}. MPA designs the probing and judging agents to automatically transform an original evaluation problem into a new one following psychometric theory on three basic cognitive abilities: language understanding, problem solving, and domain knowledge. These basic abilities are also dynamically configurable, allowing multifaceted analysis. We conducted extensive evaluations using MPA and found that most LLMs achieve poorer performance, indicating room for improvement. Our multifaceted analysis demonstrated the strong correlation between the basic abilities and an implicit Matthew effect on model size, i.e., larger models possess stronger correlations of the abilities. MPA can also be used as a data augmentation approach to enhance LLMs. Code is available at: https://github.com/microsoft/promptbench.
Authors: Michael J. Ryan, William Held, Diyi Yang
Abstract: Before being deployed for user-facing applications, developers align Large Language Models (LLMs) to user preferences through a variety of procedures, such as Reinforcement Learning From Human Feedback (RLHF) and Direct Preference Optimization (DPO). Current evaluations of these procedures focus on benchmarks of instruction following, reasoning, and truthfulness. However, human preferences are not universal, and aligning to specific preference sets may have unintended effects. We explore how alignment impacts performance along three axes of global representation: English dialects, multilingualism, and opinions from and about countries worldwide. Our results show that current alignment procedures create disparities between English dialects and global opinions. We find alignment improves capabilities in several languages. We conclude by discussing design decisions that led to these unintended impacts and recommendations for more equitable preference tuning. We make our code and data publicly available on Github.
Authors: Yihan Wang, Zhouxing Shi, Andrew Bai, Cho-Jui Hsieh
Abstract: Although many large language models (LLMs) have been trained to refuse harmful requests, they are still vulnerable to jailbreaking attacks which rewrite the original prompt to conceal its harmful intent. In this paper, we propose a new method for defending LLMs against jailbreaking attacks by ``backtranslation''. Specifically, given an initial response generated by the target LLM from an input prompt, our backtranslation prompts a language model to infer an input prompt that can lead to the response. The inferred prompt is called the backtranslated prompt which tends to reveal the actual intent of the original prompt, since it is generated based on the LLM's response and not directly manipulated by the attacker. We then run the target LLM again on the backtranslated prompt, and we refuse the original prompt if the model refuses the backtranslated prompt. We explain that the proposed defense provides several benefits on its effectiveness and efficiency. We empirically demonstrate that our defense significantly outperforms the baselines, in the cases that are hard for the baselines, and our defense also has little impact on the generation quality for benign input prompts. Our implementation is based on our library for LLM jailbreaking defense algorithms at \url{https://github.com/YihanWang617/llm-jailbreaking-defense}, and the code for reproducing our experiments is available at \url{https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation}.
URLs: https://github.com/YihanWang617/llm-jailbreaking-defense, https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation
Authors: Pratiksha Thaker, Yash Maurya, Shengyuan Hu, Virginia Smith, Zhiwei Steven Wu
Abstract: Recent work has demonstrated that finetuning is a promising approach to 'unlearn' concepts from large language models. However, finetuning can be expensive, as it requires both generating a set of examples and running iterations of finetuning to update the model. In this work, we show that simple guardrail-based approaches such as prompting and filtering can achieve unlearning results comparable to finetuning. We recommend that researchers investigate these lightweight baselines when evaluating the performance of more computationally intensive finetuning methods. While we do not claim that methods such as prompting or filtering are universal solutions to the problem of unlearning, our work suggests the need for evaluation metrics that can better separate the power of guardrails vs. finetuning, and highlights scenarios where guardrails expose possible unintended behavior in existing metrics and benchmarks.
Authors: Ekaterina Fadeeva, Aleksandr Rubashevskii, Artem Shelmanov, Sergey Petrakov, Haonan Li, Hamdy Mubarak, Evgenii Tsymbalov, Gleb Kuzmin, Alexander Panchenko, Timothy Baldwin, Preslav Nakov, Maxim Panov
Abstract: Large language models (LLMs) are notorious for hallucinating, i.e., producing erroneous claims in their output. Such hallucinations can be dangerous, as occasional factual inaccuracies in the generated text might be obscured by the rest of the output being generally factually correct, making it extremely hard for the users to spot them. Current services that leverage LLMs usually do not provide any means for detecting unreliable generations. Here, we aim to bridge this gap. In particular, we propose a novel fact-checking and hallucination detection pipeline based on token-level uncertainty quantification. Uncertainty scores leverage information encapsulated in the output of a neural network or its layers to detect unreliable predictions, and we show that they can be used to fact-check the atomic claims in the LLM output. Moreover, we present a novel token-level uncertainty quantification method that removes the impact of uncertainty about what claim to generate on the current step and what surface form to use. Our method Claim Conditioned Probability (CCP) measures only the uncertainty of a particular claim value expressed by the model. Experiments on the task of biography generation demonstrate strong improvements for CCP compared to the baselines for seven LLMs and four languages. Human evaluation reveals that the fact-checking pipeline based on uncertainty quantification is competitive with a fact-checking tool that leverages external knowledge.
Authors: Ming Dong, Yujing Chen, Miao Zhang, Hao Sun, Tingting He
Abstract: Chinese Spell Checking (CSC) is a widely used technology, which plays a vital role in speech to text (STT) and optical character recognition (OCR). Most of the existing CSC approaches relying on BERT architecture achieve excellent performance. However, limited by the scale of the foundation model, BERT-based method does not work well in few-shot scenarios, showing certain limitations in practical applications. In this paper, we explore using an in-context learning method named RS-LLM (Rich Semantic based LLMs) to introduce large language models (LLMs) as the foundation model. Besides, we study the impact of introducing various Chinese rich semantic information in our framework. We found that by introducing a small number of specific Chinese rich semantic structures, LLMs achieve better performance than the BERT-based model on few-shot CSC task. Furthermore, we conduct experiments on multiple datasets, and the experimental results verified the superiority of our proposed framework.
Authors: Young Hyun Yoo, Jii Cha, Changhyeon Kim, Taeuk Kim
Abstract: While the introduction of contrastive learning frameworks in sentence representation learning has significantly contributed to advancements in the field, it still remains unclear whether state-of-the-art sentence embeddings can capture the fine-grained semantics of sentences, particularly when conditioned on specific perspectives. In this paper, we introduce Hyper-CL, an efficient methodology that integrates hypernetworks with contrastive learning to compute conditioned sentence representations. In our proposed approach, the hypernetwork is responsible for transforming pre-computed condition embeddings into corresponding projection layers. This enables the same sentence embeddings to be projected differently according to various conditions. Evaluation on two representative conditioning benchmarks, namely conditional semantic text similarity and knowledge graph completion, demonstrates that Hyper-CL is effective in flexibly conditioning sentence representations, showcasing its computational efficiency at the same time. We also provide a comprehensive analysis of the inner workings of our approach, leading to a better interpretation of its mechanisms.
Authors: Simon A. Lee, Timothy Lindsey
Abstract: Large Language Models (LLMs) have become a pivotal research area, potentially making beneficial contributions in fields like healthcare where they can streamline automated billing and decision support. However, the frequent use of specialized coded languages like ICD-10, which are regularly updated and deviate from natural language formats, presents potential challenges for LLMs in creating accurate and meaningful latent representations. This raises concerns among healthcare professionals about potential inaccuracies or ``hallucinations" that could result in the direct impact of a patient. Therefore, this study evaluates whether large language models (LLMs) are aware of medical code ontologies and can accurately generate names from these codes. We assess the capabilities and limitations of both general and biomedical-specific generative models, such as GPT, LLaMA-2, and Meditron, focusing on their proficiency with domain-specific terminologies. While the results indicate that LLMs struggle with coded language, we offer insights on how to adapt these models to reason more effectively.
Authors: Yongqi Tong, Dawei Li, Sizhe Wang, Yujia Wang, Fei Teng, Jingbo Shang
Abstract: Recent works have shown the benefits to LLMs from fine-tuning golden-standard Chain-of-Thought (CoT) rationales or using them as correct examples in few-shot prompting. While humans can indeed imitate correct examples, learning from our mistakes is another vital aspect of human cognition. Hence, a question naturally arises: \textit{can LLMs learn and benefit from their mistakes, especially for their reasoning? } This study investigates this problem from both the prompting and model-tuning perspectives. We begin by introducing \textsc{CoTErrorSet}, a new benchmark with 609,432 questions, each designed with both correct and error references, and demonstrating the types and reasons for making such mistakes. To explore the effectiveness of those mistakes, we design two methods: (1) \textbf{Self-rethinking} prompting guides LLMs to rethink whether they have made similar previous mistakes; and (2) \textbf{Mistake tuning} involves finetuning models in both correct and incorrect reasoning domains, rather than only tuning models to learn ground truth in traditional methodology. We conduct a series of experiments to prove LLMs can obtain benefits from mistakes in both directions. Our two methods offer potentially cost-effective strategies by leveraging errors to enhance reasoning capabilities, which costs significantly less than creating meticulously hand-crafted golden references. We ultimately make a thorough analysis of the reasons behind LLMs' errors, which provides directions that future research needs to overcome. \textsc{CoTErrorSet} will be published soon on \texttt{\url{https://github.com/YookiTong/Learn-from-Mistakes-CotErrorSet}}.
URLs: https://github.com/YookiTong/Learn-from-Mistakes-CotErrorSet
Authors: Miaoran Zhang, Mingyang Wang, Jesujoba O. Alabi, Dietrich Klakow
Abstract: This paper presents our system developed for the SemEval-2024 Task 1: Semantic Textual Relatedness for African and Asian Languages. The shared task aims at measuring the semantic textual relatedness between pairs of sentences, with a focus on a range of under-represented languages. In this work, we propose using machine translation for data augmentation to address the low-resource challenge of limited training data. Moreover, we apply task-adaptive pre-training on unlabeled task data to bridge the gap between pre-training and task adaptation. For model training, we investigate both full fine-tuning and adapter-based tuning, and adopt the adapter framework for effective zero-shot cross-lingual transfer. We achieve competitive results in the shared task: our system performs the best among all ranked teams in both subtask A (supervised learning) and subtask C (cross-lingual transfer).
Authors: Noah Y. Siegel, Oana-Maria Camburu, Nicolas Heess, Maria Perez-Ortiz
Abstract: In order to oversee advanced AI systems, it is important to understand their underlying decision-making process. When prompted, large language models (LLMs) can provide natural language explanations or reasoning traces that sound plausible and receive high ratings from human annotators. However, it is unclear to what extent these explanations are faithful, i.e., truly capture the factors responsible for the model's predictions. In this work, we introduce Correlational Explanatory Faithfulness (CEF), a metric that can be used in faithfulness tests based on input interventions. Previous metrics used in such tests take into account only binary changes in the predictions. Our metric accounts for the total shift in the model's predicted label distribution, more accurately reflecting the explanations' faithfulness. We then introduce the Correlational Counterfactual Test (CCT) by instantiating CEF on the Counterfactual Test (CT) from Atanasova et al. (2023). We evaluate the faithfulness of free-text explanations generated by few-shot-prompted LLMs from the Llama2 family on three NLP tasks. We find that our metric measures aspects of faithfulness which the CT misses.
Authors: Yifeng Ding, Jiawei Liu, Yuxiang Wei, Terry Yue Zhuo, Lingming Zhang
Abstract: We introduce XFT, a simple yet powerful training scheme, by simply merging upcycled Mixture-of-Experts (MoE) to unleash the performance limit of instruction-tuned code Large Language Models (LLMs). While vanilla sparse upcycling fails to improve instruction tuning, XFT introduces a shared expert mechanism with a novel routing weight normalization strategy into sparse upcycling, which significantly boosts instruction tuning. After fine-tuning the upcycled MoE model, XFT introduces a learnable model merging mechanism to compile the upcycled MoE model back to a dense model, achieving upcycled MoE-level performance with only dense-model compute. By applying XFT to a 1.3B model, we create a new state-of-the-art tiny code LLM (<3B) with 67.1 and 64.6 pass@1 on HumanEval and HumanEval+ respectively. With the same data and model architecture, XFT improves supervised fine-tuning (SFT) by 13% on HumanEval+, along with consistent improvements from 2% to 13% on MBPP+, MultiPL-E, and DS-1000, demonstrating its generalizability. XFT is fully orthogonal to existing techniques such as Evol-Instruct and OSS-Instruct, opening a new dimension for improving code instruction tuning. Codes are available at https://github.com/ise-uiuc/xft.
Authors: Het Patel, Umair Rehman, Farkhund Iqbal
Abstract: Phishing, a prevalent cybercrime tactic for decades, remains a significant threat in today's digital world. By leveraging clever social engineering elements and modern technology, cybercrime targets many individuals, businesses, and organizations to exploit trust and security. These cyber-attackers are often disguised in many trustworthy forms to appear as legitimate sources. By cleverly using psychological elements like urgency, fear, social proof, and other manipulative strategies, phishers can lure individuals into revealing sensitive and personalized information. Building on this pervasive issue within modern technology, this paper aims to analyze the effectiveness of 15 Large Language Models (LLMs) in detecting phishing attempts, specifically focusing on a randomized set of "419 Scam" emails. The objective is to determine which LLMs can accurately detect phishing emails by analyzing a text file containing email metadata based on predefined criteria. The experiment concluded that the following models, ChatGPT 3.5, GPT-3.5-Turbo-Instruct, and ChatGPT, were the most effective in detecting phishing emails.
Authors: Davis Wertheimer, Joshua Rosenkranz, Thomas Parnell, Sahil Suneja, Pavithra Ranganathan, Raghu Ganti, Mudhakar Srivatsa
Abstract: This technical report describes the design and training of novel speculative decoding draft models, for accelerating the inference speeds of large language models in a production environment. By conditioning draft predictions on both context vectors and sampled tokens, we can train our speculators to efficiently predict high-quality n-grams, which the base model then accepts or rejects. This allows us to effectively predict multiple tokens per inference forward pass, accelerating wall-clock inference speeds of highly optimized base model implementations by a factor of 2-3x. We explore these initial results and describe next steps for further improvements.
Authors: Farima Fatahi Bayat, Xin Liu, H. V. Jagadish, Lu Wang
Abstract: Large language models (LLMs) can generate long-form and coherent text, yet they often hallucinate facts, which undermines their reliability. To mitigate this issue, inference-time methods steer LLM representations toward the "truthful directions" previously learned for truth elicitation. However, applying these truthful directions with the same intensity fails to generalize across different query contexts. We propose LITO, a Learnable Intervention method for Truthfulness Optimization that automatically identifies the optimal intervention intensity tailored to each specific context. LITO explores a sequence of model generations based on increasing levels of intervention intensities. It selects the most accurate response or refuses to answer when the predictions are highly uncertain. Experiments on multiple LLMs and question-answering datasets demonstrate that LITO improves truthfulness while preserving task accuracy. The adaptive nature of LITO counters the limitations of one-size-fits-all intervention methods, maximizing truthfulness by reflecting the model's internal knowledge only when it is confident. Our code is available at https://github.com/launchnlp/LITO.
Authors: Hanyin Wang, Chufan Gao, Bolun Liu, Qiping Xu, Guleid Hussein, Mohamad El Labban, Kingsley Iheasirim, Hariprasad Korsapati, Chuck Outcalt, Jimeng Sun
Abstract: Proprietary Large Language Models (LLMs) such as GPT-4 and Gemini have demonstrated promising capabilities in clinical text summarization tasks. However, due to patient data privacy concerns and computational costs, many healthcare providers prefer using small, locally-hosted models over external generic LLMs. This study presents a comprehensive domain- and task-specific adaptation process for the open-source LLaMA-2 13 billion parameter model, enabling it to generate high-quality clinical notes from outpatient patient-doctor dialogues. Our process incorporates continued pre-training, supervised fine-tuning, and reinforcement learning from both AI and human feedback. We introduced a new approach, DistillDirect, for performing on-policy reinforcement learning with Gemini 1.0 Pro as the teacher model. Our resulting model, LLaMA-Clinic, can generate clinical notes comparable in quality to those authored by physicians. In a blinded physician reader study, the majority (90.4%) of individual evaluations rated the notes generated by LLaMA-Clinic as "acceptable" or higher across all three criteria: real-world readiness, completeness, and accuracy. In the more challenging "Assessment and Plan" section, LLaMA-Clinic scored higher (4.2/5) in real-world readiness than physician-authored notes (4.1/5). Our cost analysis for inference shows that our LLaMA-Clinic model achieves a 4.375-fold cost reduction compared to an external generic LLM service. Additionally, we highlight key considerations for future clinical note-generation tasks, emphasizing the importance of pre-defining a best-practice note format, rather than relying on LLMs to determine this for clinical practice. We have made our newly created synthetic clinic dialogue-note dataset and the physician feedback dataset publicly available to foster future research.
Authors: Amr Keleg, Walid Magdy, Sharon Goldwater
Abstract: On annotating multi-dialect Arabic datasets, it is common to randomly assign the samples across a pool of native Arabic speakers. Recent analyses recommended routing dialectal samples to native speakers of their respective dialects to build higher-quality datasets. However, automatically identifying the dialect of samples is hard. Moreover, the pool of annotators who are native speakers of specific Arabic dialects might be scarce. Arabic Level of Dialectness (ALDi) was recently introduced as a quantitative variable that measures how sentences diverge from Standard Arabic. On randomly assigning samples to annotators, we hypothesize that samples of higher ALDi scores are harder to label especially if they are written in dialects that the annotators do not speak. We test this by analyzing the relation between ALDi scores and the annotators' agreement, on 15 public datasets having raw individual sample annotations for various sentence-classification tasks. We find strong evidence supporting our hypothesis for 11 of them. Consequently, we recommend prioritizing routing samples of high ALDi scores to native speakers of each sample's dialect, for which the dialect could be automatically identified at higher accuracies.
Authors: Wei Zhu, Aaron Xuxiang Tian, Congrui Yin, Yuan Ni, Xiaoling Wang, Guotong Xie
Abstract: Soft prompt tuning is a widely studied parameter-efficient fine-tuning method. However, it has a clear drawback: many soft tokens must be inserted into the input sequences to guarantee downstream performance. As a result, soft prompt tuning is less considered than Low-rank adaptation (LoRA) in the large language modeling (LLM) era. In this work, we propose a novel prompt tuning method, Instruction-Aware Prompt Tuning (IAPT), that requires only four soft tokens. First, we install a parameter-efficient soft prompt generator at each Transformer layer to generate idiosyncratic soft prompts for each input instruction. The generated soft prompts can be seen as a semantic summary of the input instructions and can effectively guide the output generation. Second, the soft prompt generators are modules with a bottleneck architecture consisting of a self-attention pooling operation, two linear projections, and an activation function. Pilot experiments show that prompt generators at different Transformer layers require different activation functions. Thus, we propose to learn the idiosyncratic activation functions for prompt generators automatically with the help of rational functions. We have conducted experiments on various tasks, and the experimental results demonstrate that (a) our IAPT method can outperform the recent baselines with comparable tunable parameters. (b) Our IAPT method is more efficient than LoRA under the single-backbone multi-tenant setting.
Authors: Langlin Huang, Yang Feng
Abstract: Subword tokenization is a common method for vocabulary building in Neural Machine Translation (NMT) models. However, increasingly complex tasks have revealed its disadvantages. First, a vocabulary cannot be modified once it is learned, making it hard to adapt to new words. Second, in multilingual translation, the imbalance in data volumes across different languages spreads to the vocabulary, exacerbating translations involving low-resource languages. While byte-based tokenization addresses these issues, byte-based models struggle with the low information density inherent in UTF-8 byte sequences. Previous works enhance token semantics through local contextualization but fail to select an appropriate contextualizing scope based on the input. Consequently, we propose the Multi-Scale Contextualization (MSC) method, which learns contextualized information of varying scales across different hidden state dimensions. It then leverages the attention module to dynamically integrate the multi-scale contextualized information. Experiments show that MSC significantly outperforms subword-based and other byte-based methods in both multilingual and out-of-domain scenarios. Code can be found in https://github.com/ictnlp/Multiscale-Contextualization.
URLs: https://github.com/ictnlp/Multiscale-Contextualization.
Authors: Lirong Gao, Ru Peng, Yiming Zhang, Junbo Zhao
Abstract: Prompt recovery in large language models (LLMs) is crucial for understanding how LLMs work and addressing concerns regarding privacy, copyright, etc. The trend towards inference-only APIs complicates this task by restricting access to essential outputs for recovery. To tackle this challenge, we extract prompt-related information from limited outputs and identify a strong(negative) correlation between output probability-based uncertainty and the success of prompt recovery. This finding led to the development of Deliberative PrOmpt RecoverY (DORY), our novel approach that leverages uncertainty to recover prompts accurately. DORY involves reconstructing drafts from outputs, refining these with hints, and filtering out noise based on uncertainty. Our evaluation across diverse LLMs and prompt benchmarks shows that DORY outperforms existing baselines, improving performance by approximately 10.82% and establishing a new state-of-the-art record in prompt recovery tasks. Significantly, DORY operates using a single LLM without any external resources or model, offering a cost-effective, user-friendly prompt recovery solution.
Authors: Ke Ji, Junying Chen, Anningzhe Gao, Wenya Xie, Xiang Wan, Benyou Wang
Abstract: In the quest for super-human performance, Large Language Models (LLMs) have traditionally been tethered to human-annotated datasets and predefined training objectives-a process that is both labor-intensive and inherently limited. This paper presents a transformative approach: Autonomous Learning for LLMs, a self-sufficient learning paradigm that frees models from the constraints of human supervision. This method endows LLMs with the ability to self-educate through direct interaction with text, akin to a human reading and comprehending literature. Our approach eliminates the reliance on annotated data, fostering an Autonomous Learning environment where the model independently identifies and reinforces its knowledge gaps. Empirical results from our comprehensive experiments, which utilized a diverse array of learning materials and were evaluated against standard public quizzes, reveal that Autonomous Learning outstrips the performance of both Pre-training and Supervised Fine-Tuning (SFT), as well as retrieval-augmented methods. These findings underscore the potential of Autonomous Learning to not only enhance the efficiency and effectiveness of LLM training but also to pave the way for the development of more advanced, self-reliant AI systems.
Authors: Dong Chen, Shaoxin Lin, Muhan Zeng, Daoguang Zan, Jian-Gang Wang, Anton Cheshkov, Jun Sun, Hao Yu, Guoliang Dong, Artem Aliev, Jie Wang, Xiao Cheng, Guangtai Liang, Yuchi Ma, Pan Bian, Tao Xie, Qianxiang Wang
Abstract: GitHub issue resolving recently has attracted significant attention from academia and industry. SWE-bench is proposed to measure the performance in resolving issues. In this paper, we propose CodeR, which adopts a multi-agent framework and pre-defined task graphs to Repair & Resolve reported bugs and add new features within code Repository. On SWE-bench lite, CodeR is able to solve 29.00% of issues, when submitting only once for each issue. We examine the performance impact of each design of CodeR and offer insights to advance this research direction.
Authors: Maciej Besta, Lorenzo Paleari, Ales Kubicek, Piotr Nyczyk, Robert Gerstenberger, Patrick Iff, Tomasz Lehmann, Hubert Niewiadomski, Torsten Hoefler
Abstract: Large Language Models (LLMs) are revolutionizing various domains, yet verifying their answers remains a significant challenge, especially for intricate open-ended tasks such as consolidation, summarization, and extraction of knowledge. In this work, we propose CheckEmbed: an accurate, scalable, and simple LLM verification approach. CheckEmbed is driven by a straightforward yet powerful idea: in order to compare LLM solutions to one another or to the ground-truth, compare their corresponding answer-level embeddings obtained with a model such as GPT Text Embedding Large. This reduces a complex textual answer to a single embedding, facilitating straightforward, fast, and meaningful verification. We develop a comprehensive verification pipeline implementing the CheckEmbed methodology. The CheckEmbed pipeline also comes with metrics for assessing the truthfulness of the LLM answers, such as embedding heatmaps and their summaries. We show how to use these metrics for deploying practical engines that decide whether an LLM answer is satisfactory or not. We apply the pipeline to real-world document analysis tasks, including term extraction and document summarization, showcasing significant improvements in accuracy, cost-effectiveness, and runtime performance compared to existing token-, sentence-, and fact-level schemes such as BERTScore or SelfCheckGPT.
Authors: Bingfeng Chen, Qihan Ouyang, Yongqi Luo, Boyan Xu, Ruichu Cai, Zhifeng Hao
Abstract: Previous graph-based approaches in Aspect based Sentiment Analysis(ABSA) have demonstrated impressive performance by utilizing graph neural networks and attention mechanisms to learn structures of static dependency trees and dynamic latent trees. However, incorporating both semantic and syntactic information simultaneously within complex global structures can introduce irrelevant contexts and syntactic dependencies during the process of graph structure learning, potentially resulting in inaccurate predictions. In order to address the issues above, we propose S$^2$GSL, incorporating Segment to Syntactic enhanced Graph Structure Learning for ABSA. Specifically,S$^2$GSL is featured with a segment-aware semantic graph learning and a syntax-based latent graph learning enabling the removal of irrelevant contexts and dependencies, respectively. We further propose a self-adaptive aggregation network that facilitates the fusion of two graph learning branches, thereby achieving complementarity across diverse structures. Experimental results on four benchmarks demonstrate the effectiveness of our framework.
Authors: Faisal Tareque Shohan, Mir Tafseer Nayeem, Samsul Islam, Abu Ubaida Akash, Shafiq Joty
Abstract: Millions of news articles published online daily can overwhelm readers. Headlines and entity (topic) tags are essential for guiding readers to decide if the content is worth their time. While headline generation has been extensively studied, tag generation remains largely unexplored, yet it offers readers better access to topics of interest. The need for conciseness in capturing readers' attention necessitates improved content selection strategies for identifying salient and relevant segments within lengthy articles, thereby guiding language models effectively. To address this, we propose to leverage auxiliary information such as images and captions embedded in the articles to retrieve relevant sentences and utilize instruction tuning with variations to generate both headlines and tags for news articles in a multilingual context. To make use of the auxiliary information, we have compiled a dataset named XL-HeadTags, which includes 20 languages across 6 diverse language families. Through extensive evaluation, we demonstrate the effectiveness of our plug-and-play multimodal-multilingual retrievers for both tasks. Additionally, we have developed a suite of tools for processing and evaluating multilingual texts, significantly contributing to the research community by enabling more accurate and efficient analysis across languages.
Authors: Huaiyuan Ying, Zijian Wu, Yihan Geng, Jiayu Wang, Dahua Lin, Kai Chen
Abstract: Large language models have demonstrated impressive capabilities across various natural language processing tasks, especially in solving mathematical problems. However, large language models are not good at math theorem proving using formal languages like Lean. A significant challenge in this area is the scarcity of training data available in these formal languages. To address this issue, we propose a novel pipeline that iteratively generates and filters synthetic data to translate natural language mathematical problems into Lean 4 statements, and vice versa. Our results indicate that the synthetic data pipeline can provide useful training data and improve the performance of LLMs in translating and understanding complex mathematical problems and proofs. Our final dataset contains about 57K formal-informal question pairs along with searched proof from the math contest forum and 21 new IMO questions. We open-source our code at https://github.com/InternLM/InternLM-Math and our data at https://huggingface.co/datasets/InternLM/Lean-Workbook.
URLs: https://github.com/InternLM/InternLM-Math, https://huggingface.co/datasets/InternLM/Lean-Workbook.
Authors: Aryo Pradipta Gema, Joshua Ong Jun Leang, Giwon Hong, Alessio Devoto, Alberto Carlo Maria Mancino, Rohit Saxena, Xuanli He, Yu Zhao, Xiaotang Du, Mohammad Reza Ghasemi Madani, Claire Barale, Robert McHardy, Joshua Harris, Jean Kaddour, Emile van Krieken, Pasquale Minervini
Abstract: Maybe not. We identify and analyse errors in the popular Massive Multitask Language Understanding (MMLU) benchmark. Even though MMLU is widely adopted, our analysis demonstrates numerous ground truth errors that obscure the true capabilities of LLMs. For example, we find that 57% of the analysed questions in the Virology subset contain errors. To address this issue, we introduce a comprehensive framework for identifying dataset errors using a novel error taxonomy. Then, we create MMLU-Redux, which is a subset of 3,000 manually re-annotated questions across 30 MMLU subjects. Using MMLU-Redux, we demonstrate significant discrepancies with the model performance metrics that were originally reported. Our results strongly advocate for revising MMLU's error-ridden questions to enhance its future utility and reliability as a benchmark. Therefore, we open up MMLU-Redux for additional annotation https://huggingface.co/datasets/edinburgh-dawg/mmlu-redux.
URLs: https://huggingface.co/datasets/edinburgh-dawg/mmlu-redux.
Authors: Shaina Raza, Mizanur Rahman, Michael R. Zhang
Abstract: Recent improvements in large language models (LLMs) have significantly enhanced natural language processing (NLP) applications. However, these models can also inherit and perpetuate biases from their training data. Addressing this issue is crucial, yet many existing datasets do not offer evaluation across diverse NLP tasks. To tackle this, we introduce the Bias Evaluations Across Domains (BEADs) dataset, designed to support a wide range of NLP tasks, including text classification, bias entity recognition, bias quantification, and benign language generation. BEADs uses AI-driven annotation combined with experts' verification to provide reliable labels. This method overcomes the limitations of existing datasets that typically depend on crowd-sourcing, expert-only annotations with limited bias evaluations, or unverified AI labeling. Our empirical analysis shows that BEADs is effective in detecting and reducing biases across different language models, with smaller models fine-tuned on BEADs often outperforming LLMs in bias classification tasks. However, these models may still exhibit biases towards certain demographics. Fine-tuning LLMs with our benign language data also reduces biases while preserving the models' knowledge. Our findings highlight the importance of comprehensive bias evaluation and the potential of targeted fine-tuning for reducing the bias of LLMs. We are making BEADs publicly available at https://huggingface.co/datasets/shainar/BEAD Warning: This paper contains examples that may be considered offensive.
Authors: Nadav Borenstein, Anej Svete, Robin Chan, Josef Valvoda, Franz Nowak, Isabelle Augenstein, Eleanor Chodroff, Ryan Cotterell
Abstract: What can large language models learn? By definition, language models (LM) are distributions over strings. Therefore, an intuitive way of addressing the above question is to formalize it as a matter of learnability of classes of distributions over strings. While prior work in this direction focused on assessing the theoretical limits, in contrast, we seek to understand the empirical learnability. Unlike prior empirical work, we evaluate neural LMs on their home turf-learning probabilistic languages-rather than as classifiers of formal languages. In particular, we investigate the learnability of regular LMs (RLMs) by RNN and Transformer LMs. We empirically test the learnability of RLMs as a function of various complexity parameters of the RLM and the hidden state size of the neural LM. We find that the RLM rank, which corresponds to the size of linear space spanned by the logits of its conditional distributions, and the expected length of sampled strings are strong and significant predictors of learnability for both RNNs and Transformers. Several other predictors also reach significance, but with differing patterns between RNNs and Transformers.
Authors: Sreyan Ghosh, Chandra Kiran Reddy Evuru, Sonal Kumar, Utkarsh Tyagi, Sakshi Singh, Sanjoy Chowdhury, Dinesh Manocha
Abstract: Neural image classifiers can often learn to make predictions by overly relying on non-predictive features that are spuriously correlated with the class labels in the training data. This leads to poor performance in real-world atypical scenarios where such features are absent. This paper presents ASPIRE (Language-guided Data Augmentation for SPurIous correlation REmoval), a simple yet effective solution for supplementing the training dataset with images without spurious features, for robust learning against spurious correlations via better generalization. ASPIRE, guided by language at various steps, can generate non-spurious images without requiring any group labeling or existing non-spurious images in the training set. Precisely, we employ LLMs to first extract foreground and background features from textual descriptions of an image, followed by advanced language-guided image editing to discover the features that are spuriously correlated with the class label. Finally, we personalize a text-to-image generation model using the edited images to generate diverse in-domain images without spurious features. ASPIRE is complementary to all prior robust training methods in literature, and we demonstrate its effectiveness across 4 datasets and 9 baselines and show that ASPIRE improves the worst-group classification accuracy of prior methods by 1% - 38%. We also contribute a novel test set for the challenging Hard ImageNet dataset.
Authors: Xiaoxuan Liu, Lanxiang Hu, Peter Bailis, Ion Stoica, Zhijie Deng, Alvin Cheung, Hao Zhang
Abstract: Speculative decoding is a pivotal technique to accelerate the inference of large language models (LLMs) by employing a smaller draft model to predict the target model's outputs. However, its efficacy can be limited due to the low predictive accuracy of the draft model, particularly when faced with diverse text inputs and a significant capability gap between the draft and target models. We introduce online speculative decoding to address this challenge. The main idea is to continuously update the (multiple) draft model(s) on observed user query data. Adapting to query distribution mitigates the shifts between the training distribution of the draft model and the query distribution, enabling the draft model to more accurately predict the target model's outputs. We develop a prototype of online speculative decoding based on knowledge distillation and evaluate it using both synthetic and real query data. The results show a substantial increase in the token acceptance rate by 0.1 to 0.65, bringing 1.42x to 2.17x latency reduction. Our code is available at https://github.com/LiuXiaoxuanPKU/OSD.
Authors: Aaron Lou, Chenlin Meng, Stefano Ermon
Abstract: Despite their groundbreaking performance for many generative modeling tasks, diffusion models have fallen short on discrete data domains such as natural language. Crucially, standard diffusion models rely on the well-established theory of score matching, but efforts to generalize this to discrete structures have not yielded the same empirical gains. In this work, we bridge this gap by proposing score entropy, a novel loss that naturally extends score matching to discrete spaces, integrates seamlessly to build discrete diffusion models, and significantly boosts performance. Experimentally, we test our Score Entropy Discrete Diffusion models (SEDD) on standard language modeling tasks. For comparable model sizes, SEDD beats existing language diffusion paradigms (reducing perplexity by $25$-$75$\%) and is competitive with autoregressive models, in particular outperforming GPT-2. Furthermore, compared to autoregressive mdoels, SEDD generates faithful text without requiring distribution annealing techniques like temperature scaling (around $6$-$8\times$ better generative perplexity than un-annealed GPT-2), can trade compute and quality (similar quality with $32\times$ fewer network evaluations), and enables controllable infilling (matching nucleus sampling quality while enabling other strategies besides left to right prompting).
Authors: Qinlin Zhao, Jindong Wang, Yixuan Zhang, Yiqiao Jin, Kaijie Zhu, Hao Chen, Xing Xie
Abstract: Large language models (LLMs) have been widely used as agents to complete different tasks, such as personal assistance or event planning. While most of the work has focused on cooperation and collaboration between agents, little work explores competition, another important mechanism that promotes the development of society and economy. In this paper, we seek to examine the competition dynamics in LLM-based agents. We first propose a general framework for studying the competition between agents. Then, we implement a practical competitive environment using GPT-4 to simulate a virtual town with two types of agents, restaurant agents and customer agents. Specifically, the restaurant agents compete with each other to attract more customers, where competition encourages them to transform, such as cultivating new operating strategies. Simulation experiments reveal several interesting findings at the micro and macro levels, which align well with existing market and sociological theories. We hope that the framework and environment can be a promising testbed to study competition that fosters understanding of society. Code is available at: https://github.com/microsoft/competeai.
Authors: Cheng Li, Jindong Wang, Yixuan Zhang, Kaijie Zhu, Xinyi Wang, Wenxin Hou, Jianxun Lian, Fang Luo, Qiang Yang, Xing Xie
Abstract: Emotion significantly impacts our daily behaviors and interactions. While recent generative AI models, such as large language models, have shown impressive performance in various tasks, it remains unclear whether they truly comprehend emotions. This paper aims to address this gap by incorporating psychological theories to gain a holistic understanding of emotions in generative AI models. Specifically, we propose three approaches: 1) EmotionPrompt to enhance AI model performance, 2) EmotionAttack to impair AI model performance, and 3) EmotionDecode to explain the effects of emotional stimuli, both benign and malignant. Through extensive experiments involving language and multi-modal models on semantic understanding, logical reasoning, and generation tasks, we demonstrate that both textual and visual EmotionPrompt can boost the performance of AI models while EmotionAttack can hinder it. Additionally, EmotionDecode reveals that AI models can comprehend emotional stimuli akin to the mechanism of dopamine in the human brain. Our work heralds a novel avenue for exploring psychology to enhance our understanding of generative AI models.
Authors: Dong Huang, Yuhao Qing, Weiyi Shang, Heming Cui, Jie M. Zhang
Abstract: Code generation models have increasingly become integral to aiding software development. Although current research has thoroughly examined the correctness of the code produced by code generation models, a vital aspect that plays a pivotal role in green computing and sustainability efforts has often been neglected. This paper presents EffiBench, a benchmark with 1,000 efficiency-critical coding problems to assess the efficiency of code generated by code generation models. EffiBench contains a diverse set of LeetCode coding problems. Each problem is paired with an executable human-written canonical solution, which obtains the SOTA efficiency on the LeetCode solution leaderboard. With EffiBench, we empirically examine the ability of 42 large language models (35 open-source and 7 closed-source) to generate efficient code. Our evaluation results demonstrate that the efficiency of the code generated by LLMs is generally worse than the efficiency of human-written canonical solutions. For example, GPT-4 generated code has an average \textbf{3.12} times execution time that of the human-written canonical solutions. In the most extreme cases, the execution time and total memory usage of GPT-4 generated code are \textbf{13.89} and \textbf{43.92} times that of the canonical solutions. The source code of EffiBench is released on https://github.com/huangd1999/EffiBench. We also provide the LeaderBoard at https://huggingface.co/spaces/EffiBench/effibench-leaderboard.
URLs: https://github.com/huangd1999/EffiBench., https://huggingface.co/spaces/EffiBench/effibench-leaderboard.
Authors: Alexander Pan, Erik Jones, Meena Jagadeesan, Jacob Steinhardt
Abstract: Language models influence the external world: they query APIs that read and write to web pages, generate content that shapes human behavior, and run system commands as autonomous agents. These interactions form feedback loops: LLM outputs affect the world, which in turn affect subsequent LLM outputs. In this work, we show that feedback loops can cause in-context reward hacking (ICRH), where the LLM at test-time optimizes a (potentially implicit) objective but creates negative side effects in the process. For example, consider an LLM agent deployed to increase Twitter engagement; the LLM may retrieve its previous tweets into the context window and make them more controversial, increasing engagement but also toxicity. We identify and study two processes that lead to ICRH: output-refinement and policy-refinement. For these processes, evaluations on static datasets are insufficient -- they miss the feedback effects and thus cannot capture the most harmful behavior. In response, we provide three recommendations for evaluation to capture more instances of ICRH. As AI development accelerates, the effects of feedback loops will proliferate, increasing the need to understand their role in shaping LLM behavior.
Authors: Xingang Guo, Fangxu Yu, Huan Zhang, Lianhui Qin, Bin Hu
Abstract: Jailbreaks on large language models (LLMs) have recently received increasing attention. For a comprehensive assessment of LLM safety, it is essential to consider jailbreaks with diverse attributes, such as contextual coherence and sentiment/stylistic variations, and hence it is beneficial to study controllable jailbreaking, i.e. how to enforce control on LLM attacks. In this paper, we formally formulate the controllable attack generation problem, and build a novel connection between this problem and controllable text generation, a well-explored topic of natural language processing. Based on this connection, we adapt the Energy-based Constrained Decoding with Langevin Dynamics (COLD), a state-of-the-art, highly efficient algorithm in controllable text generation, and introduce the COLD-Attack framework which unifies and automates the search of adversarial LLM attacks under a variety of control requirements such as fluency, stealthiness, sentiment, and left-right-coherence. The controllability enabled by COLD-Attack leads to diverse new jailbreak scenarios which not only cover the standard setting of generating fluent (suffix) attack with continuation constraint, but also allow us to address new controllable attack settings such as revising a user query adversarially with paraphrasing constraint, and inserting stealthy attacks in context with position constraint. Our extensive experiments on various LLMs (Llama-2, Mistral, Vicuna, Guanaco, GPT-3.5, and GPT-4) show COLD-Attack's broad applicability, strong controllability, high success rate, and attack transferability. Our code is available at https://github.com/Yu-Fangxu/COLD-Attack.
Authors: Zhangchen Xu, Fengqing Jiang, Luyao Niu, Jinyuan Jia, Bill Yuchen Lin, Radha Poovendran
Abstract: As large language models (LLMs) become increasingly integrated into real-world applications such as code generation and chatbot assistance, extensive efforts have been made to align LLM behavior with human values, including safety. Jailbreak attacks, aiming to provoke unintended and unsafe behaviors from LLMs, remain a significant/leading LLM safety threat. In this paper, we aim to defend LLMs against jailbreak attacks by introducing SafeDecoding, a safety-aware decoding strategy for LLMs to generate helpful and harmless responses to user queries. Our insight in developing SafeDecoding is based on the observation that, even though probabilities of tokens representing harmful contents outweigh those representing harmless responses, safety disclaimers still appear among the top tokens after sorting tokens by probability in descending order. This allows us to mitigate jailbreak attacks by identifying safety disclaimers and amplifying their token probabilities, while simultaneously attenuating the probabilities of token sequences that are aligned with the objectives of jailbreak attacks. We perform extensive experiments on five LLMs using six state-of-the-art jailbreak attacks and four benchmark datasets. Our results show that SafeDecoding significantly reduces the attack success rate and harmfulness of jailbreak attacks without compromising the helpfulness of responses to benign user queries. SafeDecoding outperforms six defense methods.
Authors: Wenqi Zhang, Ke Tang, Hai Wu, Mengna Wang, Yongliang Shen, Guiyang Hou, Zeqi Tan, Peng Li, Yueting Zhuang, Weiming Lu
Abstract: Large Language Models (LLMs) exhibit robust problem-solving capabilities for diverse tasks. However, most LLM-based agents are designed as specific task solvers with sophisticated prompt engineering, rather than agents capable of learning and evolving through interactions. These task solvers necessitate manually crafted prompts to inform task rules and regulate LLM behaviors, inherently incapacitating to address complex dynamic scenarios e.g., large interactive games. In light of this, we propose Agent-Pro: an LLM-based Agent with Policy-level Reflection and Optimization that can learn a wealth of expertise from interactive experiences and progressively elevate its behavioral policy. Specifically, it involves a dynamic belief generation and reflection process for policy evolution. Rather than action-level reflection, Agent-Pro iteratively reflects on past trajectories and beliefs, fine-tuning its irrational beliefs for a better policy. Moreover, a depth-first search is employed for policy optimization, ensuring continual enhancement in policy payoffs. Agent-Pro is evaluated across two games: Blackjack and Texas Hold'em, outperforming vanilla LLM and specialized models. Our results show Agent-Pro can learn and evolve in complex and dynamic scenes, which also benefits numerous LLM-based applications.
Authors: Xinyu Ma, Xu Chu, Zhibang Yang, Yang Lin, Xin Gao, Junfeng Zhao
Abstract: With the increasingly powerful performances and enormous scales of pretrained models, promoting parameter efficiency in fine-tuning has become a crucial need for effective and efficient adaptation to various downstream tasks. One representative line of fine-tuning methods is Orthogonal Fine-tuning (OFT), which rigorously preserves the angular distances within the parameter space to preserve the pretrained knowledge. Despite the empirical effectiveness, OFT still suffers low parameter efficiency at $\mathcal{O}(d^2)$ and limited capability of downstream adaptation. Inspired by Givens rotation, in this paper, we proposed quasi-Givens Orthogonal Fine-Tuning (qGOFT) to address the problems. We first use $\mathcal{O}(d)$ Givens rotations to accomplish arbitrary orthogonal transformation in $SO(d)$ with provable equivalence, reducing parameter complexity from $\mathcal{O}(d^2)$ to $\mathcal{O}(d)$. Then we introduce flexible norm and relative angular adjustments under soft orthogonality regularization to enhance the adaptation capability of downstream semantic deviations. Extensive experiments on various tasks and pretrained models validate the effectiveness of our methods.
Authors: Suyuan Zhao, Jiahuan Zhang, Yushuai Wu, Yizhen Luo, Zaiqing Nie
Abstract: Cell identity encompasses various semantic aspects of a cell, including cell type, pathway information, disease information, and more, which are essential for biologists to gain insights into its biological characteristics. Understanding cell identity from the transcriptomic data, such as annotating cell types, has become an important task in bioinformatics. As these semantic aspects are determined by human experts, it is impossible for AI models to effectively carry out cell identity understanding tasks without the supervision signals provided by single-cell and label pairs. The single-cell pre-trained language models (PLMs) currently used for this task are trained only on a single modality, transcriptomics data, lack an understanding of cell identity knowledge. As a result, they have to be fine-tuned for downstream tasks and struggle when lacking labeled data with the desired semantic labels. To address this issue, we propose an innovative solution by constructing a unified representation of single-cell data and natural language during the pre-training phase, allowing the model to directly incorporate insights related to cell identity. More specifically, we introduce LangCell, the first Language-Cell pre-training framework. LangCell utilizes texts enriched with cell identity information to gain a profound comprehension of cross-modal knowledge. Results from experiments conducted on different benchmarks show that LangCell is the only single-cell PLM that can work effectively in zero-shot cell identity understanding scenarios, and also significantly outperforms existing models in few-shot and fine-tuning cell identity understanding scenarios.
Authors: David Xu
Abstract: Multi-modal learning in the audio-language domain has seen significant advancements in recent years. However, audio-language learning faces challenges due to limited and lower-quality data compared to image-language tasks. Existing audio-language datasets are notably smaller, and manual labeling is hindered by the need to listen to entire audio clips for accurate labeling. Our method systematically generates audio-caption pairs by augmenting audio clips with natural language labels and corresponding audio signal processing operations. Leveraging a Large Language Model, we generate descriptions of augmented audio clips with a prompt template. This scalable method produces AudioSetMix, a high-quality training dataset for text-and-audio related models. Integration of our dataset improves models performance on benchmarks by providing diversified and better-aligned examples. Notably, our dataset addresses the absence of modifiers (adjectives and adverbs) in existing datasets. By enabling models to learn these concepts, and generating hard negative examples during training, we achieve state-of-the-art performance on multiple benchmarks.
Authors: Abrar Fahim, Alex Murphy, Alona Fyshe
Abstract: Multi-modal contrastive models such as CLIP achieve state-of-the-art performance in zero-shot classification by embedding input images and texts on a joint representational space. Recently, a modality gap has been reported in two-encoder contrastive models like CLIP, meaning that the image and text embeddings reside in disjoint areas of the latent space. Previous studies suggest that this gap exists due to 1) the cone effect, 2) mismatched pairs in the dataset, and 3) insufficient training. We show that, even when accounting for all these factors, and even when using the same modality, the contrastive loss actually creates a gap during training. As a result, We propose that the modality gap is inherent to the two-encoder contrastive loss and rename it the contrastive gap. We present evidence that attributes this contrastive gap to low uniformity in CLIP space, resulting in embeddings that occupy only a small portion of the latent space. To close the gap, we adapt the uniformity and alignment properties of unimodal contrastive loss to the multi-modal setting and show that simply adding these terms to the CLIP loss distributes the embeddings more uniformly in the representational space, closing the gap. In our experiments, we show that the modified representational space achieves better performance than default CLIP loss in downstream tasks such as zero-shot image classification and multi-modal arithmetic.
Authors: Wenbo Hu, Zi-Yi Dou, Liunian Harold Li, Amita Kamath, Nanyun Peng, Kai-Wei Chang
Abstract: Large Vision-Language Models (LVLMs) typically encode an image into a fixed number of visual tokens (e.g., 576) and process these tokens with a language model. Despite their strong performance, LVLMs face challenges in adapting to varying computational constraints. This raises the question: can we achieve flexibility in the number of visual tokens to suit different tasks and computational resources? We answer this with an emphatic yes. Inspired by Matryoshka Representation Learning, we introduce the Matryoshka Query Transformer (MQT), capable of encoding an image into m visual tokens during inference, where m can be any number up to a predefined maximum. This is achieved by employing a query transformer with M latent query tokens to compress the visual embeddings. During each training step, we randomly select m <= M latent query tokens and train the model using only these first m tokens, discarding the rest. Combining MQT with LLaVA, we train a single model once, and flexibly and drastically reduce the number of inference-time visual tokens while maintaining similar or better performance compared to training independent models for each number of tokens. Our model, MQT-LLAVA, matches LLaVA-1.5 performance across 11 benchmarks using a maximum of 256 tokens instead of LLaVA's fixed 576. Reducing to 16 tokens (8x less TFLOPs) only sacrifices the performance by 2.4 points on MMBench. On certain tasks such as ScienceQA and MMMU, we can even go down to only 2 visual tokens with performance drops of just 3% and 6% each. Our exploration of the trade-off between the accuracy and computational cost brought about by the number of visual tokens facilitates future research to achieve the best of both worlds.
Authors: Anke Tang, Li Shen, Yong Luo, Han Hu, Bo Do, Dacheng Tao
Abstract: Deep model fusion is an emerging technique that unifies the predictions or parameters of several deep neural networks into a single model in a cost-effective and data-efficient manner. This enables the unified model to take advantage of the original models' strengths, potentially exceeding their performance. Although a variety of deep model fusion techniques have been introduced, their evaluations tend to be inconsistent and often inadequate to validate their effectiveness and robustness against distribution shifts. To address this issue, we introduce FusionBench, which is the first comprehensive benchmark dedicated to deep model fusion. FusionBench covers a wide range of tasks, including open-vocabulary image classification, text classification, and text-to-text generation. Each category includes up to eight tasks with corresponding task-specific models, featuring both full fine-tuning and LoRA fine-tuning, as well as models of different sizes, to ensure fair and balanced comparisons of various multi-task model fusion techniques across different tasks, model scales, and fine-tuning strategies. We implement and evaluate a broad spectrum of deep model fusion techniques. These techniques range from model ensemble methods, which combine the predictions to improve the overall performance, to model merging, which integrates different models into a single one, and model mixing methods, which upscale or recombine the components of the original models. FusionBench now contains 26 distinct tasks, 74 fine-tuned models, and 16 fusion techniques, and we are committed to consistently expanding the benchmark with more tasks, models, and fusion techniques. In addition, we offer a well-documented set of resources and guidelines to aid researchers in understanding and replicating the benchmark results. Homepage https://github.com/tanganke/fusion_bench
Authors: Max Mueller-Eberstein, Dianna Yee, Karren Yang, Gautam Varma Mantena, Colin Lea
Abstract: Parameter-efficient fine-tuning (PEFT) for personalizing automatic speech recognition (ASR) has recently shown promise for adapting general population models to atypical speech. However, these approaches assume a priori knowledge of the atypical speech disorder being adapted for -- the diagnosis of which requires expert knowledge that is not always available. Even given this knowledge, data scarcity and high inter/intra-speaker variability further limit the effectiveness of traditional fine-tuning. To circumvent these challenges, we first identify the minimal set of model parameters required for ASR adaptation. Our analysis of each individual parameter's effect on adaptation performance allows us to reduce Word Error Rate (WER) by half while adapting 0.03% of all weights. Alleviating the need for cohort-specific models, we next propose the novel use of a meta-learned hypernetwork to generate highly individualized, utterance-level adaptations on-the-fly for a diverse set of atypical speech characteristics. Evaluating adaptation at the global, cohort and individual-level, we show that hypernetworks generalize better to out-of-distribution speakers, while maintaining an overall relative WER reduction of 75.2% using 0.1% of the full parameter budget.