id stringlengths 33 34 | doc_url stringlengths 33 34 | title stringlengths 10 211 | publication_date timestamp[s] | update_date timestamp[s] | authors stringlengths 5 20.1k | category_primary stringclasses 78
values | category_all stringlengths 5 138 | abstract stringlengths 123 3.09k | embeddings listlengths 768 768 |
|---|---|---|---|---|---|---|---|---|---|
http://arxiv.org/abs/2406.11354v2 | http://arxiv.org/pdf/2406.11354v2 | Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression | 2024-06-17T09:17:40 | 2024-06-19T11:36:30 | Zilun Zhang, Yutao Sun, Tiancheng Zhao, Leigang Sha, Ruochen Xu, Kyusong Lee, Jianwei Yin | cs.CL | cs.CL, cs.AI, cs.CV | Humans can retain old knowledge while learning new information, but Large
Language Models (LLMs) often suffer from catastrophic forgetting when
post-pretrained or supervised fine-tuned (SFT) on domain-specific data.
Moreover, for Multimodal Large Language Models (MLLMs) which are composed of
the LLM base and visual pro... | [
-0.1846190095,
-0.7702288032,
0.3645899296,
-0.1153070703,
-0.216497466,
0.2802288234,
-0.3093538582,
-0.1158535704,
-0.2082952112,
0.5012716651,
-0.5831431746,
-0.5810500979,
-0.2941257358,
-0.0972457975,
-0.3493289948,
0.9399265051,
0.0088101933,
-0.303558439,
-0.2454229742,
... |
http://arxiv.org/abs/2406.11353v1 | http://arxiv.org/pdf/2406.11353v1 | $\texttt{MoE-RBench}$: Towards Building Reliable Language Models with Sparse Mixture-of-Experts | 2024-06-17T09:17:05 | 2024-06-17T09:17:05 | Guanjie Chen, Xinyu Zhao, Tianlong Chen, Yu Cheng | cs.LG | cs.LG, cs.CL | Mixture-of-Experts (MoE) has gained increasing popularity as a promising
framework for scaling up large language models (LLMs). However, the reliability
assessment of MoE lags behind its surging applications. Moreover, when
transferred to new domains such as in fine-tuning MoE models sometimes
underperform their dense ... | [
-0.6176921725,
-0.8514074087,
0.3047843874,
0.2501880825,
-0.0928577334,
-0.2110058218,
-0.3954578042,
-0.3627986312,
0.2448377758,
0.3047888875,
-0.3168923259,
-0.0906323195,
-0.8971933722,
0.1314515471,
-0.3735282421,
0.89718014,
0.3032341599,
-0.0791065916,
-0.2751869857,
-0... |
http://arxiv.org/abs/2406.15484v1 | http://arxiv.org/pdf/2406.15484v1 | JobFair: A Framework for Benchmarking Gender Hiring Bias in Large Language Models | 2024-06-17T09:15:57 | 2024-06-17T09:15:57 | Ze Wang, Zekun Wu, Xin Guan, Michael Thaler, Adriano Koshiyama, Skylar Lu, Sachin Beepath, Ediz Ertekin Jr., Maria Perez-Ortiz | cs.CL | cs.CL, cs.AI, cs.CY | This paper presents a novel framework for benchmarking hierarchical gender
hiring bias in Large Language Models (LLMs) for resume scoring, revealing
significant issues of reverse bias and overdebiasing. Our contributions are
fourfold: First, we introduce a framework using a real, anonymized resume
dataset from the Heal... | [
0.0325679481,
-0.4813630879,
0.5208905935,
0.2748689353,
0.4979777336,
0.1148079857,
0.2744686604,
-0.2820324898,
0.0661438406,
0.4577411711,
-0.7438774705,
-0.4980036616,
-0.5526089668,
-0.0672055036,
-0.5402633548,
0.8554881215,
0.2365952134,
0.3391891718,
-0.5415896773,
0.02... |
http://arxiv.org/abs/2406.11345v1 | http://arxiv.org/pdf/2406.11345v1 | Full-ECE: A Metric For Token-level Calibration on Large Language Models | 2024-06-17T09:07:58 | 2024-06-17T09:07:58 | Han Liu, Yupeng Zhang, Bingning Wang, Weipeng Chen, Xiaolin Hu | cs.CL | cs.CL, cs.AI | Deep Neural Networks (DNNs) excel in various domains but face challenges in
providing accurate uncertainty estimates, which are crucial for high-stakes
applications. Large Language Models (LLMs) have recently emerged as powerful
tools, demonstrating exceptional performance in language tasks. However,
traditional calibr... | [
-0.3196446598,
-0.4378741682,
0.6189894676,
0.0237736721,
-0.177022025,
0.1953769624,
-0.7569636106,
-0.0075604226,
0.2102389932,
0.3128041625,
-0.339886874,
-0.8098566532,
-0.7452117801,
-0.1000066549,
-0.3952412009,
0.6971084476,
-0.0176097155,
0.1110671386,
-0.0311231297,
-0... |
http://arxiv.org/abs/2406.11341v1 | http://arxiv.org/pdf/2406.11341v1 | A Systematic Analysis of Large Language Models as Soft Reasoners: The Case of Syllogistic Inferences | 2024-06-17T08:59:04 | 2024-06-17T08:59:04 | Leonardo Bertolazzi, Albert Gatt, Raffaella Bernardi | cs.CL | cs.CL | The reasoning abilities of Large Language Models (LLMs) are becoming a
central focus of study in NLP. In this paper, we consider the case of
syllogistic reasoning, an area of deductive reasoning studied extensively in
logic and cognitive psychology. Previous research has shown that pre-trained
LLMs exhibit reasoning bi... | [
-0.2184423804,
-0.9098820686,
0.8410255909,
-0.1905307025,
-0.2584424913,
-0.2992892265,
-0.4434317946,
-0.2929137945,
-0.1869326383,
0.6134648323,
-0.4212142825,
-0.3854779303,
-0.7192646265,
0.2239360511,
-0.330555141,
1.0029587746,
0.1609761566,
0.1320094317,
-0.1540304571,
... |
http://arxiv.org/abs/2406.11338v1 | http://arxiv.org/pdf/2406.11338v1 | Fine-grained Controllable Text Generation through In-context Learning with Feedback | 2024-06-17T08:55:48 | 2024-06-17T08:55:48 | Sarubi Thillainathan, Alexander Koller | cs.CL | cs.CL | We present a method for rewriting an input sentence to match specific values
of nontrivial linguistic features, such as dependency depth. In contrast to
earlier work, our method uses in-context learning rather than finetuning,
making it applicable in use cases where data is sparse. We show that our model
performs accur... | [
-0.2174963355,
-1.2700181007,
0.5558526516,
0.0528291836,
-0.0060943151,
-0.2460798174,
-0.2592506707,
-0.3194390833,
0.2129455805,
0.7378769517,
-0.7022182345,
-0.4584128857,
-0.2247378826,
-0.0167731699,
-0.4631778002,
1.2819805145,
-0.3840815127,
-0.2653155625,
-0.360457778,
... |
http://arxiv.org/abs/2406.11328v1 | http://arxiv.org/pdf/2406.11328v1 | Are Large Language Models True Healthcare Jacks-of-All-Trades? Benchmarking Across Health Professions Beyond Physician Exams | 2024-06-17T08:40:36 | 2024-06-17T08:40:36 | Zheheng Luo, Chenhan Yuan, Qianqian Xie, Sophia Ananiadou | cs.CL | cs.CL | Recent advancements in Large Language Models (LLMs) have demonstrated their
potential in delivering accurate answers to questions about world knowledge.
Despite this, existing benchmarks for evaluating LLMs in healthcare
predominantly focus on medical doctors, leaving other critical healthcare
professions underrepresen... | [
0.1221671402,
-0.5513182282,
0.9422490001,
-0.0376939029,
-0.0137967598,
-0.2035546303,
-0.1793486923,
-0.7365011573,
-0.0785392374,
0.4886451662,
-0.2407741249,
-0.432990551,
-0.5851826668,
0.247628957,
-0.0837332532,
0.4397221506,
0.0690368041,
0.4411430955,
-0.111847274,
-0.... |
http://arxiv.org/abs/2406.11925v2 | http://arxiv.org/pdf/2406.11925v2 | DocCGen: Document-based Controlled Code Generation | 2024-06-17T08:34:57 | 2024-07-03T09:16:27 | Sameer Pimparkhede, Mehant Kammakomati, Srikanth Tamilselvam, Prince Kumar, Ashok Pon Kumar, Pushpak Bhattacharyya | cs.SE | cs.SE, cs.AI, cs.CL | Recent developments show that Large Language Models (LLMs) produce
state-of-the-art performance on natural language (NL) to code generation for
resource-rich general-purpose languages like C++, Java, and Python. However,
their practical usage for structured domain-specific languages (DSLs) such as
YAML, JSON is limited... | [
-0.3352304697,
-0.6841657162,
0.540427804,
0.162819773,
-0.0903388485,
0.1650247276,
-0.4739351273,
-0.2275744081,
-0.2462607622,
0.8357560635,
-0.4928077161,
-0.8909671903,
-0.3787554801,
0.3026487231,
-0.4014577568,
0.9003685713,
-0.1039980054,
0.0384986959,
-0.3701395392,
0.... |
http://arxiv.org/abs/2406.11317v1 | http://arxiv.org/pdf/2406.11317v1 | GUICourse: From General Vision Language Models to Versatile GUI Agents | 2024-06-17T08:30:55 | 2024-06-17T08:30:55 | Wentong Chen, Junbo Cui, Jinyi Hu, Yujia Qin, Junjie Fang, Yue Zhao, Chongyi Wang, Jun Liu, Guirong Chen, Yupeng Huo, Yuan Yao, Yankai Lin, Zhiyuan Liu, Maosong Sun | cs.AI | cs.AI, cs.CL, cs.CV, cs.HC | Utilizing Graphic User Interface (GUI) for human-computer interaction is
essential for accessing a wide range of digital tools. Recent advancements in
Vision Language Models (VLMs) highlight the compelling potential to develop
versatile agents to help humans finish GUI navigation tasks. However, current
VLMs are challe... | [
-0.5003546476,
-0.6908404231,
0.3838460743,
-0.5320090055,
0.0687146261,
-0.0060830978,
-0.0602362379,
-0.476595372,
-0.2454515696,
0.9937617779,
-0.2442816496,
-0.7322865129,
-0.0304776244,
-0.2222711891,
-0.270899564,
0.9934287667,
-0.1216298044,
-0.1285363436,
-0.4574580789,
... |
http://arxiv.org/abs/2406.11307v1 | http://arxiv.org/pdf/2406.11307v1 | An Empirical Investigation of Matrix Factorization Methods for Pre-trained Transformers | 2024-06-17T08:14:23 | 2024-06-17T08:14:23 | Ashim Gupta, Sina Mahdipour Saravani, P. Sadayappan, Vivek Srikumar | cs.CL | cs.CL | The increasing size of transformer-based models in NLP makes the question of
compressing them important. In this work, we present a comprehensive analysis
of factorization based model compression techniques. Specifically, we focus on
comparing straightforward low-rank factorization against the recently
introduced Monar... | [
-0.7806764841,
-0.54532516,
0.1508667916,
0.3955614865,
-0.2124977708,
0.226355955,
-0.2191699445,
-0.124251999,
0.2619798481,
0.3327849805,
-0.4568063021,
-0.0564390458,
-0.7020364404,
0.3840777576,
-0.3886747062,
0.7081990838,
0.196180433,
0.2129375041,
-0.3475687504,
0.14980... |
http://arxiv.org/abs/2406.11303v1 | http://arxiv.org/pdf/2406.11303v1 | VideoVista: A Versatile Benchmark for Video Understanding and Reasoning | 2024-06-17T08:09:00 | 2024-06-17T08:09:00 | Yunxin Li, Xinyu Chen, Baotian Hu, Longyue Wang, Haoyuan Shi, Min Zhang | cs.CV | cs.CV, cs.AI, cs.CL | Despite significant breakthroughs in video analysis driven by the rapid
development of large multimodal models (LMMs), there remains a lack of a
versatile evaluation benchmark to comprehensively assess these models'
performance in video understanding and reasoning. To address this, we present
VideoVista, a video QA ben... | [
-0.4875843525,
-0.8442133665,
0.7962428927,
0.1313372254,
-0.3100166321,
0.1287917495,
0.1052114964,
-0.0872011855,
-0.1250322908,
0.3192950487,
-0.5887498856,
-0.2822966576,
-0.5679146647,
-0.0177418217,
-0.4281231463,
0.7018978,
-0.4007052481,
-0.000110719,
-0.3918897212,
-0.... |
http://arxiv.org/abs/2406.11301v1 | http://arxiv.org/pdf/2406.11301v1 | Optimizing and Testing Instruction-Following: Analyzing the Impact of Fine-Grained Instruction Variants on instruction-tuned LLMs | 2024-06-17T08:08:11 | 2024-06-17T08:08:11 | Jiuding Yang, Weidong Guo, Kaitong Yang, Xiangyang Li, Zhuwei Rao, Yu Xu, Di Niu | cs.AI | cs.AI, cs.CL, cs.LG | The effective alignment of Large Language Models (LLMs) with precise
instructions is essential for their application in diverse real-world
scenarios. Current methods focus on enhancing the diversity and complexity of
training and evaluation samples, yet they fall short in accurately assessing
LLMs' ability to follow si... | [
-0.2835016549,
-0.7766416669,
0.1374371797,
-0.0858886391,
-0.1637322754,
0.0877482891,
-0.4660558403,
-0.3510484099,
-0.1962009668,
0.5910698175,
-0.7463867664,
-0.6119781137,
-0.4722058475,
0.0986696854,
-0.230629921,
0.7123804688,
-0.2610374689,
0.1748802811,
-0.2323251665,
... |
http://arxiv.org/abs/2406.11924v1 | http://arxiv.org/pdf/2406.11924v1 | Explainable assessment of financial experts' credibility by classifying social media forecasts and checking the predictions with actual market data | 2024-06-17T08:08:03 | 2024-06-17T08:08:03 | Silvia García-Méndez, Francisco de Arriba-Pérez, Jaime González-Gonzáleza, Francisco J. González-Castaño | cs.SI | cs.SI, cs.AI, cs.CL, cs.LG | Social media include diverse interaction metrics related to user popularity,
the most evident example being the number of user followers. The latter has
raised concerns about the credibility of the posts by the most popular
creators. However, most existing approaches to assess credibility in social
media strictly consi... | [
-0.3413769603,
-0.9265115261,
0.4930124283,
0.5305688381,
-0.3016545177,
0.5472733378,
-0.1823462248,
-0.6617336869,
0.2752859294,
0.3433044553,
-0.3401076794,
-0.6393294334,
-0.8447538614,
-0.1295553893,
-0.6252467036,
0.9460728765,
0.1024823338,
-0.0759941265,
-0.2567360997,
... |
http://arxiv.org/abs/2406.11290v1 | http://arxiv.org/pdf/2406.11290v1 | Iterative Utility Judgment Framework via LLMs Inspired by Relevance in Philosophy | 2024-06-17T07:52:42 | 2024-06-17T07:52:42 | Hengran Zhang, Keping Bi, Jiafeng Guo, Xueqi Cheng | cs.IR | cs.IR, cs.AI, cs.CL, cs.LG | Utility and topical relevance are critical measures in information retrieval
(IR), reflecting system and user perspectives, respectively. While topical
relevance has long been emphasized, utility is a higher standard of relevance
and is more useful for facilitating downstream tasks, e.g., in
Retrieval-Augmented Generat... | [
-0.2460183799,
-0.3960668445,
0.9074727893,
-0.0654981956,
-0.3742806613,
0.0500731207,
0.0194968022,
-0.0858459324,
0.3057219386,
0.804993391,
-0.258000195,
-0.5059928298,
-0.2780376673,
-0.1508444101,
-0.3106775284,
1.054800868,
0.1663595587,
-0.1514378935,
-0.4076305032,
-0.... |
http://arxiv.org/abs/2406.11289v1 | http://arxiv.org/pdf/2406.11289v1 | A Systematic Survey of Text Summarization: From Statistical Methods to Large Language Models | 2024-06-17T07:52:32 | 2024-06-17T07:52:32 | Haopeng Zhang, Philip S. Yu, Jiawei Zhang | cs.CL | cs.CL | Text summarization research has undergone several significant transformations
with the advent of deep neural networks, pre-trained language models (PLMs),
and recent large language models (LLMs). This survey thus provides a
comprehensive review of the research progress and evolution in text
summarization through the le... | [
-0.3124839365,
-0.700886488,
0.4807543159,
0.0730794817,
-0.2339822352,
0.2580465078,
-0.3199242651,
-0.1683631688,
0.4455010593,
0.5852796435,
-0.3923329115,
-0.5338870287,
-0.6548560858,
0.2856366336,
-0.6145885587,
1.3168549538,
0.0006116691,
0.247290805,
-0.1681466252,
-0.0... |
http://arxiv.org/abs/2406.11288v1 | http://arxiv.org/pdf/2406.11288v1 | MFC-Bench: Benchmarking Multimodal Fact-Checking with Large Vision-Language Models | 2024-06-17T07:51:44 | 2024-06-17T07:51:44 | Shengkang Wang, Hongzhan Lin, Ziyang Luo, Zhen Ye, Guang Chen, Jing Ma | cs.CL | cs.CL, cs.CV | Large vision-language models (LVLMs) have significantly improved multimodal
reasoning tasks, such as visual question answering and image captioning. These
models embed multimodal facts within their parameters, rather than relying on
external knowledge bases to store factual information explicitly. However, the
content ... | [
-0.4336991012,
-0.677852869,
0.7572839856,
0.0621208847,
-0.0312953927,
0.088075079,
-0.234447673,
-0.4640329778,
-0.3574169576,
0.400972724,
-0.5660224557,
-0.4599485099,
-0.4224151969,
0.0668351352,
-0.5096951723,
0.5530930161,
-0.235240072,
0.2395678908,
-0.7788724303,
-0.47... |
http://arxiv.org/abs/2406.11285v1 | http://arxiv.org/pdf/2406.11285v1 | Self and Cross-Model Distillation for LLMs: Effective Methods for Refusal Pattern Alignment | 2024-06-17T07:46:45 | 2024-06-17T07:46:45 | Jie Li, Yi Liu, Chongyang Liu, Xiaoning Ren, Ling Shi, Weisong Sun, Yinxing Xue | cs.CR | cs.CR, cs.CL | Large Language Models (LLMs) like OpenAI's GPT series, Anthropic's Claude,
and Meta's LLaMa have shown remarkable capabilities in text generation.
However, their susceptibility to toxic prompts presents significant security
challenges. This paper investigates alignment techniques, including Supervised
Fine-Tuning (SFT)... | [
-0.0368759707,
-0.8229523897,
0.8399309516,
0.1078759432,
-0.2468550652,
0.2874962986,
0.0873056054,
-0.3256058097,
-0.2738947272,
0.7267509103,
-0.8877512217,
-0.5051232576,
-0.5272347331,
0.3881394267,
-0.3151962459,
1.2310495377,
-0.183669135,
0.1146631241,
-0.0171422884,
-0... |
http://arxiv.org/abs/2406.11278v1 | http://arxiv.org/pdf/2406.11278v1 | Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs | 2024-06-17T07:30:40 | 2024-06-17T07:30:40 | Duygu Nur Yaldiz, Yavuz Faruk Bakman, Baturalp Buyukates, Chenyang Tao, Anil Ramakrishna, Dimitrios Dimitriadis, Salman Avestimehr | cs.CL | cs.CL | In this work, we introduce the Learnable Response Scoring Function (LARS) for
Uncertainty Estimation (UE) in generative Large Language Models (LLMs). Current
scoring functions for probability-based UE, such as length-normalized scoring
and semantic contribution-based weighting, are designed to solve specific
aspects of... | [
-0.4179138839,
-0.6524320841,
0.6525469422,
0.2639382482,
-0.3320170343,
0.021705348,
-0.2815877199,
-0.2406140417,
0.1356334537,
0.122479327,
-0.5755642653,
-0.708344996,
-0.5774453878,
-0.1466776431,
-0.1332042217,
0.7541136146,
-0.0502355471,
0.2095987052,
-0.3245388865,
0.1... |
http://arxiv.org/abs/2406.11277v1 | http://arxiv.org/pdf/2406.11277v1 | Small Agent Can Also Rock! Empowering Small Language Models as Hallucination Detector | 2024-06-17T07:30:05 | 2024-06-17T07:30:05 | Xiaoxue Cheng, Junyi Li, Wayne Xin Zhao, Hongzhi Zhang, Fuzheng Zhang, Di Zhang, Kun Gai, Ji-Rong Wen | cs.CL | cs.CL | Hallucination detection is a challenging task for large language models
(LLMs), and existing studies heavily rely on powerful closed-source LLMs such
as GPT-4. In this paper, we propose an autonomous LLM-based agent framework,
called HaluAgent, which enables relatively smaller LLMs (e.g. Baichuan2-Chat
7B) to actively ... | [
-0.4114121497,
-0.8540773392,
0.7427718639,
0.2272192091,
-0.1738254875,
-0.2257107645,
-0.5215839744,
-0.5639721155,
0.3902721405,
0.5120628476,
-0.4813485444,
-0.5284742117,
-0.3424887061,
0.281198591,
-0.2668715715,
0.716573894,
0.1368624419,
-0.0334576704,
0.2080079615,
-0.... |
http://arxiv.org/abs/2406.11275v1 | http://arxiv.org/pdf/2406.11275v1 | Self-training Large Language Models through Knowledge Detection | 2024-06-17T07:25:09 | 2024-06-17T07:25:09 | Wei Jie Yeo, Teddy Ferdinan, Przemyslaw Kazienko, Ranjan Satapathy, Erik Cambria | cs.CL | cs.CL | Large language models (LLMs) often necessitate extensive labeled datasets and
training compute to achieve impressive performance across downstream tasks.
This paper explores a self-training paradigm, where the LLM autonomously
curates its own labels and selectively trains on unknown data samples
identified through a re... | [
-0.2877644002,
-0.2005983442,
0.6842004061,
0.0529127046,
0.0629758164,
-0.135008961,
-0.3066110313,
-0.3335695267,
0.1339263171,
0.8345610499,
-0.5194101334,
-0.3988105059,
-0.5992196798,
0.0364060365,
-0.4588057101,
0.7219036818,
-0.050709229,
-0.0713593811,
-0.3011923134,
-0... |
http://arxiv.org/abs/2406.11274v1 | http://arxiv.org/pdf/2406.11274v1 | Skip-Layer Attention: Bridging Abstract and Detailed Dependencies in Transformers | 2024-06-17T07:24:38 | 2024-06-17T07:24:38 | Qian Chen, Wen Wang, Qinglin Zhang, Siqi Zheng, Shiliang Zhang, Chong Deng, Hai Yu, Jiaqing Liu, Yukun Ma, Chong Zhang | cs.CL | cs.CL | The Transformer architecture has significantly advanced deep learning,
particularly in natural language processing, by effectively managing long-range
dependencies. However, as the demand for understanding complex relationships
grows, refining the Transformer's architecture becomes critical. This paper
introduces Skip-... | [
-0.6270032525,
-0.5902492404,
0.5700951219,
0.6724511385,
0.214530468,
0.2238691002,
0.2133309394,
-0.7898471951,
0.0432387367,
0.6812784076,
-0.6525278091,
-0.0451748595,
-0.6210983992,
-0.2445515394,
-0.3928731084,
0.9113976955,
0.0594495758,
-0.559715867,
-0.2682649493,
0.10... |
http://arxiv.org/abs/2406.11267v1 | http://arxiv.org/pdf/2406.11267v1 | Mitigating Large Language Model Hallucination with Faithful Finetuning | 2024-06-17T07:16:07 | 2024-06-17T07:16:07 | Minda Hu, Bowei He, Yufei Wang, Liangyou Li, Chen Ma, Irwin King | cs.CL | cs.CL | Large language models (LLMs) have demonstrated remarkable performance on
various natural language processing tasks. However, they are prone to
generating fluent yet untruthful responses, known as "hallucinations".
Hallucinations can lead to the spread of misinformation and cause harm in
critical applications. Mitigatin... | [
-0.3626333773,
-1.2215777636,
0.566919744,
0.0400409698,
-0.2852466702,
-0.2766800523,
-0.3885478079,
-0.297421217,
-0.0386569202,
0.8976831436,
-0.6481493711,
-0.259241581,
-0.3788618147,
0.0057979929,
-0.3332770765,
0.9465870261,
-0.0562209673,
0.3280563354,
-0.2844006121,
0.... |
http://arxiv.org/abs/2406.11263v1 | http://arxiv.org/pdf/2406.11263v1 | The Fall of ROME: Understanding the Collapse of LLMs in Model Editing | 2024-06-17T07:08:29 | 2024-06-17T07:08:29 | Wanli Yang, Fei Sun, Jiajun Tan, Xinyu Ma, Du Su, Dawei Yin, Huawei Shen | cs.CL | cs.CL, cs.AI | Despite significant progress in model editing methods, their application in
real-world scenarios remains challenging as they often cause large language
models (LLMs) to collapse. Among them, ROME is particularly concerning, as it
could disrupt LLMs with only a single edit. In this paper, we study the root
causes of suc... | [
-0.5714841485,
-0.76957798,
0.7815372944,
-0.0552243255,
-0.1361416727,
0.1490954608,
-0.3033258617,
-0.0389121249,
-0.2473690808,
0.4254187346,
-0.5701767206,
-0.3764844835,
-0.7445943952,
0.0929840878,
-0.7543628216,
1.1068460941,
-0.3822778463,
-0.0551220067,
-0.0470702685,
... |
http://arxiv.org/abs/2406.11260v1 | http://arxiv.org/pdf/2406.11260v1 | Adversarial Style Augmentation via Large Language Model for Robust Fake News Detection | 2024-06-17T07:00:41 | 2024-06-17T07:00:41 | Sungwon Park, Sungwon Han, Meeyoung Cha | cs.CL | cs.CL, cs.AI | The spread of fake news negatively impacts individuals and is regarded as a
significant social challenge that needs to be addressed. A number of
algorithmic and insightful features have been identified for detecting fake
news. However, with the recent LLMs and their advanced generation capabilities,
many of the detecta... | [
-0.4289849997,
-1.2196599245,
0.3726121783,
0.1604976356,
-0.3810634017,
0.2196333855,
-0.283249557,
-0.5046780705,
0.3439538777,
0.4170049131,
-0.5977064967,
-0.3000277579,
-0.7900104523,
0.1024328619,
-0.626745224,
0.97732234,
-0.0622566603,
0.1296053827,
0.1277979314,
-0.087... |
http://arxiv.org/abs/2406.11258v1 | http://arxiv.org/pdf/2406.11258v1 | Enhancing Biomedical Knowledge Retrieval-Augmented Generation with Self-Rewarding Tree Search and Proximal Policy Optimization | 2024-06-17T06:48:31 | 2024-06-17T06:48:31 | Minda Hu, Licheng Zong, Hongru Wang, Jingyan Zhou, Jingjing Li, Yichen Gao, Kam-Fai Wong, Yu Li, Irwin King | cs.CL | cs.CL | Large Language Models (LLMs) have shown great potential in the biomedical
domain with the advancement of retrieval-augmented generation (RAG). However,
existing retrieval-augmented approaches face challenges in addressing diverse
queries and documents, particularly for medical knowledge queries, resulting in
sub-optima... | [
0.0265582986,
-0.5587247014,
0.9316715002,
-0.1061686203,
-0.3766529858,
0.3333426714,
0.1539015919,
-0.2926465273,
0.4407300353,
0.6578935385,
-0.4431968331,
-0.3876212239,
-0.3787325323,
0.2083852887,
0.1817737371,
1.0602952242,
0.1661850065,
0.0303343534,
-0.416757822,
0.065... |
http://arxiv.org/abs/2406.11256v1 | http://arxiv.org/pdf/2406.11256v1 | Dynamic Data Mixing Maximizes Instruction Tuning for Mixture-of-Experts | 2024-06-17T06:47:03 | 2024-06-17T06:47:03 | Tong Zhu, Daize Dong, Xiaoye Qu, Jiacheng Ruan, Wenliang Chen, Yu Cheng | cs.CL | cs.CL | Mixture-of-Experts (MoE) models have shown remarkable capability in
instruction tuning, especially when the number of tasks scales. However,
previous methods simply merge all training tasks (e.g. creative writing,
coding, and mathematics) and apply fixed sampling weights, without considering
the importance of different... | [
-0.5047762394,
-0.8065623045,
0.1313487738,
0.2063978016,
-0.2664469779,
0.4207633138,
-0.2529189587,
-0.2672690749,
0.3114634454,
0.418640852,
-0.6951223016,
-0.2747269571,
-0.610947907,
-0.2056509703,
-0.4764548838,
0.7300223708,
0.0268727336,
0.1045291424,
-0.3662165999,
-0.... |
http://arxiv.org/abs/2406.15483v1 | http://arxiv.org/pdf/2406.15483v1 | Duplicate Detection with GenAI | 2024-06-17T06:42:13 | 2024-06-17T06:42:13 | Ian Ormesher | cs.CL | cs.CL, cs.DB, cs.LG | Customer data is often stored as records in Customer Relations Management
systems (CRMs). Data which is manually entered into such systems by one of more
users over time leads to data replication, partial duplication or fuzzy
duplication. This in turn means that there no longer a single source of truth
for customers, c... | [
-0.3102625906,
-0.6432864666,
0.2499811053,
0.0566887632,
-0.136642471,
0.2608412802,
0.2035635114,
-0.5722439885,
0.0376683958,
0.9118508697,
-0.5211260915,
-0.5682019591,
-0.1776176691,
0.2569759786,
-0.5467290878,
0.8556151986,
-0.0420879871,
0.1074958965,
-0.4411004186,
-0.... |
http://arxiv.org/abs/2406.11250v1 | http://arxiv.org/pdf/2406.11250v1 | Can Machines Resonate with Humans? Evaluating the Emotional and Empathic Comprehension of LMs | 2024-06-17T06:22:20 | 2024-06-17T06:22:20 | Muhammad Arslan Manzoor, Yuxia Wang, Minghan Wang, Preslav Nakov | cs.CL | cs.CL | Empathy plays a pivotal role in fostering prosocial behavior, often triggered
by the sharing of personal experiences through narratives. However, modeling
empathy using NLP approaches remains challenging due to its deep
interconnection with human interaction dynamics. Previous approaches, which
involve fine-tuning lang... | [
-0.6207950711,
-0.6616697907,
0.5411852598,
0.1631899625,
-0.2874249518,
0.1818553954,
-0.3046787083,
-0.7405416369,
0.295781523,
0.5638245344,
-0.5990374088,
-0.4434160292,
-0.8438914418,
0.5486633778,
0.1139606237,
0.6528704166,
-0.1016539857,
0.0315972976,
0.0439730212,
-0.5... |
http://arxiv.org/abs/2406.11243v1 | http://arxiv.org/pdf/2406.11243v1 | FamiCom: Further Demystifying Prompts for Language Models with Task-Agnostic Performance Estimation | 2024-06-17T06:14:55 | 2024-06-17T06:14:55 | Bangzheng Li, Ben Zhou, Xingyu Fu, Fei Wang, Dan Roth, Muhao Chen | cs.CL | cs.CL, cs.AI | Language models have shown impressive in-context-learning capabilities, which
allow them to benefit from input prompts and perform better on downstream end
tasks. Existing works investigate the mechanisms behind this observation, and
propose label-agnostic prompt metrics that can better estimate end-task
performances. ... | [
-0.5756632686,
-0.7095085382,
0.8399582505,
0.0149345733,
-0.0793487281,
0.0746445358,
-0.0826915056,
-0.3042564392,
-0.043914225,
0.2081090063,
-0.7351500988,
-0.444221288,
-0.9194718003,
0.2539520264,
-0.3650006652,
0.9126999974,
-0.1525334716,
-0.4719265103,
-0.6930203438,
-... |
http://arxiv.org/abs/2406.15481v1 | http://arxiv.org/pdf/2406.15481v1 | CSRT: Evaluation and Analysis of LLMs using Code-Switching Red-Teaming Dataset | 2024-06-17T06:08:18 | 2024-06-17T06:08:18 | Haneul Yoo, Yongjin Yang, Hwaran Lee | cs.AI | cs.AI, cs.CL | Recent studies in large language models (LLMs) shed light on their
multilingual ability and safety, beyond conventional tasks in language
modeling. Still, current benchmarks reveal their inability to comprehensively
evaluate them and are excessively dependent on manual annotations. In this
paper, we introduce code-swit... | [
-0.2883505821,
-0.9533351064,
-0.070771493,
0.4327894747,
-0.1418052763,
0.5910336971,
-0.5934318304,
-0.5864853263,
-0.1516908556,
0.454911232,
-0.3288187981,
-0.4364506304,
-0.5018764734,
0.2808563411,
-0.6659287214,
1.1449890137,
-0.1512731761,
0.229772836,
-0.0164713655,
-0... |
http://arxiv.org/abs/2406.11239v1 | http://arxiv.org/pdf/2406.11239v1 | Evading AI-Generated Content Detectors using Homoglyphs | 2024-06-17T06:07:32 | 2024-06-17T06:07:32 | Aldan Creo, Shushanta Pudasaini | cs.CL | cs.CL, cs.AI | The generation of text that is increasingly human-like has been enabled by
the advent of large language models (LLMs). As the detection of AI-generated
content holds significant importance in the fight against issues such as
misinformation and academic cheating, numerous studies have been conducted to
develop reliable ... | [
-0.4062648416,
-1.1926567554,
0.4136474431,
-0.4225699306,
-0.3829169273,
0.260215342,
-0.0947968215,
-0.8648331761,
0.3027124405,
0.2839211822,
-0.2058126926,
-0.4302277863,
-0.8271021247,
0.3449953198,
-0.726326406,
0.7217587233,
-0.1539295316,
-0.1035090089,
0.1100939289,
0.... |
http://arxiv.org/abs/2406.11238v1 | http://arxiv.org/pdf/2406.11238v1 | What Kinds of Tokens Benefit from Distant Text? An Analysis on Long Context Language Modeling | 2024-06-17T06:07:29 | 2024-06-17T06:07:29 | Yutong Hu, Quzhe Huang, Kangcheng Luo, Yansong Feng | cs.CL | cs.CL | As the context length that large language models can handle continues to
increase, these models demonstrate an enhanced ability to utilize distant
information for tasks such as language modeling. This capability contrasts with
human reading and writing habits, where it is uncommon to remember and use
particularly dista... | [
-0.0968826711,
-0.8987320662,
0.7759698629,
0.2138908654,
-0.4352093637,
-0.4118277431,
-0.4669578671,
-0.9015724659,
0.1735250801,
0.4572903514,
-0.09609586,
-0.3030605614,
-0.7643964291,
0.0865368173,
-0.5599778295,
1.2079998255,
0.0829205438,
-0.199405551,
-0.3531112969,
0.0... |
http://arxiv.org/abs/2406.11234v1 | http://arxiv.org/pdf/2406.11234v1 | MiniConGTS: A Near Ultimate Minimalist Contrastive Grid Tagging Scheme for Aspect Sentiment Triplet Extraction | 2024-06-17T06:01:11 | 2024-06-17T06:01:11 | Qiao Sun, Liujia Yang, Minghao Ma, Nanyang Ye, Qinying Gu | cs.CL | cs.CL, cs.AI | Aspect Sentiment Triplet Extraction (ASTE) aims to co-extract the sentiment
triplets in a given corpus. Existing approaches within the
pretraining-finetuning paradigm tend to either meticulously craft complex
tagging schemes and classification heads, or incorporate external semantic
augmentation to enhance performance.... | [
-0.8249121904,
-0.1843912303,
0.2761591375,
0.0746853575,
-0.7168347239,
0.2225843072,
-0.3321349025,
-0.4000338316,
0.3109427989,
-0.020818945,
-0.4764662683,
-0.6928393841,
-0.46470204,
0.1896302551,
0.0148402285,
1.3357934952,
-0.1509744525,
-0.0587140843,
0.1173127517,
-0.1... |
http://arxiv.org/abs/2406.11233v1 | http://arxiv.org/pdf/2406.11233v1 | Probing the Decision Boundaries of In-context Learning in Large Language Models | 2024-06-17T06:00:24 | 2024-06-17T06:00:24 | Siyan Zhao, Tung Nguyen, Aditya Grover | cs.LG | cs.LG, cs.AI, cs.CL | In-context learning is a key paradigm in large language models (LLMs) that
enables them to generalize to new tasks and domains by simply prompting these
models with a few exemplars without explicit parameter updates. Many attempts
have been made to understand in-context learning in LLMs as a function of model
scale, pr... | [
-0.6341676712,
-1.1599171162,
0.6386601925,
-0.0421511605,
-0.2317660153,
-0.1375942677,
-0.6512608528,
-0.3001610339,
-0.2830420136,
0.3228175938,
-0.6205151677,
-0.4922565818,
-0.3863255084,
-0.0283522718,
-0.1524386108,
0.8923463821,
-0.1510846019,
0.1386418492,
-0.4542533159,... |
http://arxiv.org/abs/2406.11231v1 | http://arxiv.org/pdf/2406.11231v1 | Enabling robots to follow abstract instructions and complete complex dynamic tasks | 2024-06-17T05:55:35 | 2024-06-17T05:55:35 | Ruaridh Mon-Williams, Gen Li, Ran Long, Wenqian Du, Chris Lucas | cs.RO | cs.RO, cs.AI, cs.CL, cs.LG | Completing complex tasks in unpredictable settings like home kitchens
challenges robotic systems. These challenges include interpreting high-level
human commands, such as "make me a hot beverage" and performing actions like
pouring a precise amount of water into a moving mug. To address these
challenges, we present a n... | [
-0.0893544033,
-0.7637306452,
0.7835434079,
0.0732757077,
0.2533385158,
0.3297221363,
-0.2668717802,
-0.2999096513,
0.0793072656,
0.5777004957,
-0.921979785,
-0.5035629272,
-0.6790095568,
-0.0063487501,
-0.3171806335,
0.7168501616,
-0.3822456896,
-0.0573875532,
-0.1373971701,
0... |
http://arxiv.org/abs/2406.11230v1 | http://arxiv.org/pdf/2406.11230v1 | Multimodal Needle in a Haystack: Benchmarking Long-Context Capability of Multimodal Large Language Models | 2024-06-17T05:54:06 | 2024-06-17T05:54:06 | Hengyi Wang, Haizhou Shi, Shiwei Tan, Weiyi Qin, Wenyuan Wang, Tunyu Zhang, Akshay Nambi, Tanuja Ganu, Hao Wang | cs.LG | cs.LG, cs.AI, cs.CL, cs.CV | Multimodal Large Language Models (MLLMs) have shown significant promise in
various applications, leading to broad interest from researchers and
practitioners alike. However, a comprehensive evaluation of their long-context
capabilities remains underexplored. To address these gaps, we introduce the
MultiModal Needle-in-... | [
-0.415227294,
-0.8735310435,
0.7735009193,
0.163352713,
-0.3059830666,
-0.1359036267,
-0.3559978902,
-0.7263293266,
0.0354255959,
0.2460308224,
-0.4269370735,
-0.5644794703,
-0.4927661717,
0.2318221033,
-0.4647601843,
0.6874763966,
-0.1015317142,
-0.0898928568,
-0.4220932722,
-... |
http://arxiv.org/abs/2406.11228v1 | http://arxiv.org/pdf/2406.11228v1 | ComperDial: Commonsense Persona-grounded Dialogue Dataset and Benchmark | 2024-06-17T05:51:04 | 2024-06-17T05:51:04 | Hiromi Wakaki, Yuki Mitsufuji, Yoshinori Maeda, Yukiko Nishimura, Silin Gao, Mengjie Zhao, Keiichi Yamada, Antoine Bosselut | cs.CL | cs.CL | We propose a new benchmark, ComperDial, which facilitates the training and
evaluation of evaluation metrics for open-domain dialogue systems. ComperDial
consists of human-scored responses for 10,395 dialogue turns in 1,485
conversations collected from 99 dialogue agents submitted to the Commonsense
Persona-grounded Dia... | [
-0.5230044127,
-0.4290450215,
0.3475067019,
0.1398609579,
-0.1047949642,
0.2396782935,
-0.3126096725,
-0.0119011467,
-0.1199850664,
0.5370056033,
-0.7922329307,
-0.5074831843,
-0.3642392755,
0.0960414633,
-0.0060233958,
0.7785192728,
0.4712794721,
0.3121390045,
-0.551641047,
-0... |
http://arxiv.org/abs/2406.11218v1 | http://arxiv.org/pdf/2406.11218v1 | Building another Spanish dictionary, this time with GPT-4 | 2024-06-17T05:25:56 | 2024-06-17T05:25:56 | Miguel Ortega-Martín, Óscar García-Sierra, Alfonso Ardoiz, Juan Carlos Armenteros, Ignacio Garrido, Jorge Álvarez, Camilo Torrón, Iñigo Galdeano, Ignacio Arranz, Oleg Vorontsov, Adrián Alonso | cs.CL | cs.CL, cs.AI | We present the "Spanish Built Factual Freectianary 2.0" (Spanish-BFF-2) as
the second iteration of an AI-generated Spanish dictionary. Previously, we
developed the inaugural version of this unique free dictionary employing GPT-3.
In this study, we aim to improve the dictionary by using GPT-4-turbo instead.
Furthermore,... | [
-0.3805819154,
-0.6232151389,
0.488175869,
0.4588619471,
-0.3644262254,
0.3929308653,
-0.0218457617,
-0.8323905468,
0.4500986934,
0.2275986969,
-0.6326114535,
-0.1832365841,
-0.6855467558,
0.2993045449,
-0.2613011003,
1.2726955414,
-0.1015990078,
-0.0497794114,
0.0998854712,
-0... |
http://arxiv.org/abs/2406.11217v2 | http://arxiv.org/pdf/2406.11217v2 | WeatherQA: Can Multimodal Language Models Reason about Severe Weather? | 2024-06-17T05:23:18 | 2024-06-24T03:55:30 | Chengqian Ma, Zhanxiang Hua, Alexandra Anderson-Frey, Vikram Iyer, Xin Liu, Lianhui Qin | cs.AI | cs.AI, cs.CL, cs.CV, physics.ao-ph | Severe convective weather events, such as hail, tornadoes, and thunderstorms,
often occur quickly yet cause significant damage, costing billions of dollars
every year. This highlights the importance of forecasting severe weather
threats hours in advance to better prepare meteorologists and residents in
at-risk areas. C... | [
-0.4931147993,
-0.6567166448,
0.6935318112,
0.4979730248,
-0.1271256804,
-0.0400698781,
-0.0785550252,
-0.6002420187,
-0.2317693084,
0.4766846895,
-0.6973767877,
-0.3570042551,
-0.2967098653,
-0.1444372833,
-0.6131014228,
0.7211982012,
-0.2253720164,
-0.0320484564,
-0.3168614507,... |
http://arxiv.org/abs/2406.11214v1 | http://arxiv.org/pdf/2406.11214v1 | Global Data Constraints: Ethical and Effectiveness Challenges in Large Language Model | 2024-06-17T05:13:25 | 2024-06-17T05:13:25 | Jin Yang, Zhiqiang Wang, Yanbin Lin, Zunduo Zhao | cs.CL | cs.CL | The efficacy and ethical integrity of large language models (LLMs) are
profoundly influenced by the diversity and quality of their training datasets.
However, the global landscape of data accessibility presents significant
challenges, particularly in regions with stringent data privacy laws or limited
open-source infor... | [
-0.0832760558,
-0.6069164872,
0.5693533421,
-0.2116203606,
0.0066144611,
0.0945367739,
-0.33510077,
-0.7111031413,
-0.5167658329,
0.7209991217,
-0.4468012452,
-0.3991658092,
-0.735950768,
0.0823960677,
-0.1817480922,
1.0358353853,
-0.1202839911,
0.2860052288,
-0.3187815249,
-0.... |
http://arxiv.org/abs/2406.11201v2 | http://arxiv.org/pdf/2406.11201v2 | Fine-Tuning or Fine-Failing? Debunking Performance Myths in Large Language Models | 2024-06-17T04:35:17 | 2024-06-30T14:42:52 | Scott Barnett, Zac Brannelly, Stefanus Kurniawan, Sheng Wong | cs.CL | cs.CL, cs.AI | Large Language Models (LLMs) have the unique capability to understand and
generate human-like text from input queries. When fine-tuned, these models show
enhanced performance on domain-specific queries. OpenAI highlights the process
of fine-tuning, stating: "To fine-tune a model, you are required to provide at
least 10... | [
-0.6966617703,
-0.8438193202,
0.5010921955,
-0.0917165279,
-0.3247233629,
-0.0603707545,
-0.5009127855,
-0.6488294601,
-0.2334928513,
0.6551072598,
-0.3956486881,
-0.3996909559,
-0.574549675,
0.1362912804,
-0.3358142078,
1.0555950403,
-0.0104585486,
-0.0066786739,
-0.176996991,
... |
http://arxiv.org/abs/2406.11200v2 | http://arxiv.org/pdf/2406.11200v2 | AvaTaR: Optimizing LLM Agents for Tool-Assisted Knowledge Retrieval | 2024-06-17T04:20:02 | 2024-06-18T01:39:57 | Shirley Wu, Shiyu Zhao, Qian Huang, Kexin Huang, Michihiro Yasunaga, Kaidi Cao, Vassilis N. Ioannidis, Karthik Subbian, Jure Leskovec, James Zou | cs.LG | cs.LG, cs.CL | Large language model (LLM) agents have demonstrated impressive capability in
utilizing external tools and knowledge to boost accuracy and reduce
hallucinations. However, developing the prompting techniques that make LLM
agents able to effectively use external tools and knowledge is a heuristic and
laborious task. Here,... | [
-0.3580720425,
-0.709143877,
0.6816587448,
-0.015876824,
-0.2519994378,
0.1996206492,
-0.1265243292,
-0.2874618769,
0.1335831583,
0.256403327,
-0.2879007757,
-0.620834589,
-0.4914110005,
0.1556918174,
-0.1585085243,
0.3625257909,
0.2656395733,
0.02433924,
-0.6125474572,
-0.2372... |
http://arxiv.org/abs/2406.11194v1 | http://arxiv.org/pdf/2406.11194v1 | In-Context Editing: Learning Knowledge from Self-Induced Distributions | 2024-06-17T04:00:04 | 2024-06-17T04:00:04 | Siyuan Qi, Bangcheng Yang, Kailin Jiang, Xiaobo Wang, Jiaqi Li, Yifan Zhong, Yaodong Yang, Zilong Zheng | cs.CL | cs.CL | The existing fine-tuning paradigm for language models is brittle in knowledge
editing scenarios, where the model must incorporate new information without
extensive retraining. This brittleness often results in overfitting, reduced
performance, and unnatural language generation. To address this, we propose
Consistent In... | [
-0.1427650303,
-0.8046540618,
0.3284109235,
0.1647336483,
-0.0763475522,
-0.0535404421,
-0.6209040284,
-0.193837449,
-0.2065069675,
0.3668805659,
-0.4213014841,
-0.4644270539,
-0.3740204275,
0.0895503387,
-0.3936286569,
1.2182701826,
-0.2346985042,
-0.0965702906,
-0.2825747132,
... |
http://arxiv.org/abs/2406.11193v1 | http://arxiv.org/pdf/2406.11193v1 | MMNeuron: Discovering Neuron-Level Domain-Specific Interpretation in Multimodal Large Language Model | 2024-06-17T03:59:44 | 2024-06-17T03:59:44 | Jiahao Huo, Yibo Yan, Boren Hu, Yutao Yue, Xuming Hu | cs.CL | cs.CL | Projecting visual features into word embedding space has become a significant
fusion strategy adopted by Multimodal Large Language Models (MLLMs). However,
its internal mechanisms have yet to be explored. Inspired by multilingual
research, we identify domain-specific neurons in multimodal large language
models. Specifi... | [
-0.5539456606,
-0.8500353694,
0.5330050588,
0.151421845,
0.0294791386,
-0.230535388,
-0.1190179884,
-0.3914310634,
0.1830598712,
0.3812574148,
-0.6519564986,
-0.4448502064,
-0.5907247066,
-0.2052993029,
-0.3803517818,
0.585131526,
0.1536367983,
0.1384623796,
-0.2919485569,
-0.2... |
http://arxiv.org/abs/2406.11192v1 | http://arxiv.org/pdf/2406.11192v1 | Beyond Boundaries: Learning a Universal Entity Taxonomy across Datasets and Languages for Open Named Entity Recognition | 2024-06-17T03:57:35 | 2024-06-17T03:57:35 | Yuming Yang, Wantong Zhao, Caishuang Huang, Junjie Ye, Xiao Wang, Huiyuan Zheng, Yang Nan, Yuran Wang, Xueying Xu, Kaixin Huang, Yunke Zhang, Tao Gui, Qi Zhang, Xuanjing Huang | cs.CL | cs.CL | Open Named Entity Recognition (NER), which involves identifying arbitrary
types of entities from arbitrary domains, remains challenging for Large
Language Models (LLMs). Recent studies suggest that fine-tuning LLMs on
extensive NER data can boost their performance. However, training directly on
existing datasets faces ... | [
-0.3864244819,
-0.3619967699,
0.0494860075,
-0.0774459392,
-0.1972242743,
-0.0412432626,
-0.6092331409,
-0.5465478301,
-0.2289302945,
0.4845788479,
-0.2249478251,
-0.5578677654,
-0.6006574035,
0.057789214,
-0.0203616749,
1.0302966833,
-0.2283023298,
0.0121837761,
-0.0368228182,
... |
http://arxiv.org/abs/2406.11191v2 | http://arxiv.org/pdf/2406.11191v2 | A Survey on Human Preference Learning for Large Language Models | 2024-06-17T03:52:51 | 2024-06-18T08:18:33 | Ruili Jiang, Kehai Chen, Xuefeng Bai, Zhixuan He, Juntao Li, Muyun Yang, Tiejun Zhao, Liqiang Nie, Min Zhang | cs.CL | cs.CL | The recent surge of versatile large language models (LLMs) largely depends on
aligning increasingly capable foundation models with human intentions by
preference learning, enhancing LLMs with excellent applicability and
effectiveness in a wide range of contexts. Despite the numerous related studies
conducted, a perspec... | [
-0.493811816,
-0.4109647274,
0.7384845018,
-0.0277846102,
-0.0496680886,
-0.0776848719,
-0.8194928765,
-0.4974820018,
0.1234969348,
0.6275916696,
-0.4333171546,
-0.5178453922,
-0.4772428274,
-0.0222096499,
0.0236958768,
0.6677408814,
0.0883444324,
0.4209164977,
-0.1498170793,
-... |
http://arxiv.org/abs/2406.11190v1 | http://arxiv.org/pdf/2406.11190v1 | Aligning Large Language Models from Self-Reference AI Feedback with one General Principle | 2024-06-17T03:51:46 | 2024-06-17T03:51:46 | Rong Bao, Rui Zheng, Shihan Dou, Xiao Wang, Enyu Zhou, Bo Wang, Qi Zhang, Liang Ding, Dacheng Tao | cs.CL | cs.CL, cs.AI | In aligning large language models (LLMs), utilizing feedback from existing
advanced AI rather than humans is an important method to scale supervisory
signals. However, it is highly challenging for AI to understand human
intentions and societal values, and provide accurate preference feedback based
on these. Current AI ... | [
-0.2661989033,
-0.4368439317,
0.6629231572,
0.5394522548,
-0.063100785,
-0.2400034964,
-0.4918340743,
-0.3948251009,
0.2607523799,
0.6366071105,
-0.4955472946,
-0.2640619576,
-0.4357175231,
0.1698506325,
-0.1424182504,
1.0969390869,
0.128490746,
0.0760361627,
-0.2700459063,
-0.... |
http://arxiv.org/abs/2406.11177v1 | http://arxiv.org/pdf/2406.11177v1 | TIFG: Text-Informed Feature Generation with Large Language Models | 2024-06-17T03:29:14 | 2024-06-17T03:29:14 | Xinhao Zhang, Jinghan Zhang, Fengran Mo, Yuzhong Chen, Kunpeng Liu | cs.CL | cs.CL | Textual information of data is of vital importance for data mining and
feature engineering. However, existing methods focus on learning the data
structures and overlook the textual information along with the data.
Consequently, they waste this valuable resource and miss out on the deeper data
relationships embedded wit... | [
-0.557495892,
-0.7790256143,
0.647469461,
-0.0728807971,
-0.2278099507,
0.1662348062,
-0.3295974731,
-0.5027076602,
0.1672736555,
0.7728456855,
-0.6973465085,
-0.6587972045,
-0.4653974473,
0.0973909125,
-0.5322521925,
1.1641910076,
0.0760634542,
-0.3551256061,
-0.1254758835,
0.... |
http://arxiv.org/abs/2406.11176v1 | http://arxiv.org/pdf/2406.11176v1 | Watch Every Step! LLM Agent Learning via Iterative Step-Level Process Refinement | 2024-06-17T03:29:13 | 2024-06-17T03:29:13 | Weimin Xiong, Yifan Song, Xiutian Zhao, Wenhao Wu, Xun Wang, Ke Wang, Cheng Li, Wei Peng, Sujian Li | cs.CL | cs.CL, cs.AI | Large language model agents have exhibited exceptional performance across a
range of complex interactive tasks. Recent approaches have utilized tuning with
expert trajectories to enhance agent performance, yet they primarily
concentrate on outcome rewards, which may lead to errors or suboptimal actions
due to the absen... | [
-0.1410275698,
-0.3630303442,
0.4000888169,
-0.2265773416,
0.0055570668,
0.4737149179,
0.0366148688,
-0.0545839667,
-0.1501632929,
0.6326153278,
-0.7248690128,
-0.6847813129,
-0.6364169717,
-0.0471188724,
-0.1040739492,
1.0193818808,
-0.298597306,
0.1452726722,
-0.0401795357,
0... |
http://arxiv.org/abs/2406.11173v2 | http://arxiv.org/pdf/2406.11173v2 | BSRBF-KAN: A combination of B-splines and Radial Basic Functions in Kolmogorov-Arnold Networks | 2024-06-17T03:26:02 | 2024-06-19T23:01:14 | Hoang-Thang Ta | cs.CL | cs.CL | In this paper, we introduce BSRBF-KAN, a Kolmogorov Arnold Network (KAN) that
combines Bsplines and radial basis functions (RBFs) to fit input vectors in
data training. We perform experiments with BSRBF-KAN, MLP, and other popular
KANs, including EfficientKAN, FastKAN, FasterKAN, and GottliebKAN over the
MNIST and Fash... | [
-0.5003300905,
-0.6438443065,
0.1972799897,
0.7147544026,
-0.5873472691,
0.1219927371,
-0.0613529757,
-0.4184255004,
0.2690443993,
0.3532179296,
-0.5703148842,
-0.4136730731,
-0.3681482971,
0.1304617971,
-0.3513500988,
0.9982305765,
0.0344942734,
0.7732934356,
-0.0941938832,
-0... |
http://arxiv.org/abs/2406.11172v1 | http://arxiv.org/pdf/2406.11172v1 | Enhancing Criminal Case Matching through Diverse Legal Factors | 2024-06-17T03:23:11 | 2024-06-17T03:23:11 | Jie Zhao, Ziyu Guan, Wei Zhao, Yue Jiang | cs.CL | cs.CL | Criminal case matching endeavors to determine the relevance between different
criminal cases. Conventional methods predict the relevance solely based on
instance-level semantic features and neglect the diverse legal factors (LFs),
which are associated with diverse court judgments. Consequently,
comprehensively represen... | [
-0.3782270849,
-0.5283512473,
0.8228861094,
0.1396917403,
-0.571916759,
0.3691032231,
-0.0658480376,
-0.6345968843,
-0.0287538469,
0.5618696809,
-0.4888629317,
-0.4368411899,
-0.3827097714,
-0.036612656,
-0.2698135078,
0.9287292957,
0.2406543493,
0.2649948597,
-0.2735038698,
-0... |
http://arxiv.org/abs/2406.11171v2 | http://arxiv.org/pdf/2406.11171v2 | SUGARCREPE++ Dataset: Vision-Language Model Sensitivity to Semantic and Lexical Alterations | 2024-06-17T03:22:20 | 2024-06-19T00:03:42 | Sri Harsha Dumpala, Aman Jaiswal, Chandramouli Sastry, Evangelos Milios, Sageev Oore, Hassan Sajjad | cs.CV | cs.CV, cs.CL, cs.LG, 68T45, 68T50, I.2.7; I.2.10 | Despite their remarkable successes, state-of-the-art large language models
(LLMs), including vision-and-language models (VLMs) and unimodal language
models (ULMs), fail to understand precise semantics. For example, semantically
equivalent sentences expressed using different lexical compositions elicit
diverging represe... | [
-0.2556089759,
-0.333958447,
0.4648527503,
-0.0357771851,
-0.2142809927,
0.1926158071,
-0.4737638235,
-0.682411015,
0.0661519021,
0.6033177972,
-0.3659724295,
-0.4478356838,
-0.7840670347,
0.1820923537,
-0.5172110796,
0.957539022,
-0.0542466082,
0.5654807091,
-0.5317587256,
-0.... |
http://arxiv.org/abs/2406.15480v1 | http://arxiv.org/pdf/2406.15480v1 | On Giant's Shoulders: Effortless Weak to Strong by Dynamic Logits Fusion | 2024-06-17T03:07:41 | 2024-06-17T03:07:41 | Chenghao Fan, Zhenyi Lu, Wei Wei, Jie Tian, Xiaoye Qu, Dangyang Chen, Yu Cheng | cs.CL | cs.CL, cs.AI, cs.LG | Efficient fine-tuning of large language models for task-specific applications
is imperative, yet the vast number of parameters in these models makes their
training increasingly challenging. Despite numerous proposals for effective
methods, a substantial memory overhead remains for gradient computations during
updates. ... | [
-0.2035634071,
-0.788988471,
0.208410725,
0.1017576307,
-0.3700113893,
0.26312837,
-0.3193153441,
-0.6123626232,
0.1981616765,
0.2531805038,
-0.591602087,
-0.2743020654,
-0.7344314456,
-0.0681717098,
-0.2369976491,
1.0001584291,
0.1334673613,
-0.0815226734,
0.0742561817,
-0.398... |
http://arxiv.org/abs/2406.11162v2 | http://arxiv.org/pdf/2406.11162v2 | How Good are LLMs at Relation Extraction under Low-Resource Scenario? Comprehensive Evaluation | 2024-06-17T03:02:04 | 2024-06-26T01:43:15 | Dawulie Jinensibieke, Mieradilijiang Maimaiti, Wentao Xiao, Yuanhang Zheng, Xiaobo Wang | cs.CL | cs.CL | Relation Extraction (RE) serves as a crucial technology for transforming
unstructured text into structured information, especially within the framework
of Knowledge Graph development. Its importance is emphasized by its essential
role in various downstream tasks. Besides the conventional RE methods which are
based on n... | [
-0.4110281169,
-0.5026958585,
0.5453346372,
-0.0102578355,
-0.5316414237,
0.0844517797,
-0.5250028372,
-0.601759851,
-0.0290434435,
0.4735000432,
-0.5801272392,
-0.5222004652,
-0.4822663367,
0.5566551089,
-0.1173775271,
0.8670271635,
-0.1343618929,
-0.1906008422,
-0.120463796,
... |
http://arxiv.org/abs/2406.15479v1 | http://arxiv.org/pdf/2406.15479v1 | Twin-Merging: Dynamic Integration of Modular Expertise in Model Merging | 2024-06-17T02:31:55 | 2024-06-17T02:31:55 | Zhenyi Lu, Chenghao Fan, Wei Wei, Xiaoye Qu, Dangyang Chen, Yu Cheng | cs.CL | cs.CL, cs.AI, cs.LG | In the era of large language models, model merging is a promising way to
combine multiple task-specific models into a single multitask model without
extra training. However, two challenges remain: (a) interference between
different models and (b) heterogeneous data during testing. Traditional model
merging methods ofte... | [
-0.5238727331,
-0.8980807662,
0.4142070413,
-0.0571113974,
-0.0267340783,
0.3336971104,
-0.3948140144,
-0.6088340282,
-0.0803999156,
0.3498082459,
-0.5789102316,
-0.2573850751,
-0.5591125488,
-0.3687810004,
-0.5361717939,
0.8963500261,
-0.1586076766,
-0.0370076932,
-0.5385499001,... |
http://arxiv.org/abs/2406.11149v1 | http://arxiv.org/pdf/2406.11149v1 | GoldCoin: Grounding Large Language Models in Privacy Laws via Contextual Integrity Theory | 2024-06-17T02:27:32 | 2024-06-17T02:27:32 | Wei Fan, Haoran Li, Zheye Deng, Weiqi Wang, Yangqiu Song | cs.CL | cs.CL, cs.CR | Privacy issues arise prominently during the inappropriate transmission of
information between entities. Existing research primarily studies privacy by
exploring various privacy attacks, defenses, and evaluations within narrowly
predefined patterns, while neglecting that privacy is not an isolated,
context-free concept ... | [
0.0256199501,
-1.1631131172,
0.6313572526,
0.0067576799,
-0.2834094465,
-0.0557002723,
-0.3443689942,
-0.6082386374,
0.0896067992,
1.0498775244,
-0.2020771503,
-0.632617116,
-0.5278881788,
-0.0531308465,
-0.4878275096,
1.3885136843,
0.0637151822,
-0.4476488531,
-0.1617718488,
-... |
http://arxiv.org/abs/2406.11139v1 | http://arxiv.org/pdf/2406.11139v1 | Breaking Boundaries: Investigating the Effects of Model Editing on Cross-linguistic Performance | 2024-06-17T01:54:27 | 2024-06-17T01:54:27 | Somnath Banerjee, Avik Halder, Rajarshi Mandal, Sayan Layek, Ian Soboroff, Rima Hazra, Animesh Mukherjee | cs.CL | cs.CL | The integration of pretrained language models (PLMs) like BERT and GPT has
revolutionized NLP, particularly for English, but it has also created
linguistic imbalances. This paper strategically identifies the need for
linguistic equity by examining several knowledge editing techniques in
multilingual contexts. We evalua... | [
-0.4654183984,
-0.6901239157,
0.285512656,
0.3038771152,
0.1182216555,
0.3748945296,
-0.3465191126,
-0.7415730953,
-0.051545959,
0.4716417193,
-0.3414515555,
-0.3433421254,
-0.8578800559,
0.2409490198,
-0.3395707905,
0.7849949002,
-0.4017789066,
0.2474256307,
-0.2919023633,
-0.... |
http://arxiv.org/abs/2406.11132v1 | http://arxiv.org/pdf/2406.11132v1 | RePrompt: Planning by Automatic Prompt Engineering for Large Language Models Agents | 2024-06-17T01:23:11 | 2024-06-17T01:23:11 | Weizhe Chen, Sven Koenig, Bistra Dilkina | cs.CL | cs.CL, cs.AI, cs.LG | In this past year, large language models (LLMs) have had remarkable success
in domains outside the traditional natural language processing, and people are
starting to explore the usage of LLMs in more general and close to application
domains like code generation, travel planning, and robot controls. Connecting
these LL... | [
-0.6240975857,
-0.5807256103,
0.6326478124,
0.3284931779,
-0.0875198916,
0.1515526325,
-0.3144367039,
-0.0603429973,
0.1417728662,
0.5820298195,
-1.1376523972,
-0.5493443608,
-0.0903354436,
0.2503814101,
0.1033791974,
0.5962558985,
-0.2513068914,
0.1645629704,
-0.0234390758,
0.... |
http://arxiv.org/abs/2406.11131v2 | http://arxiv.org/pdf/2406.11131v2 | Are Large Language Models a Good Replacement of Taxonomies? | 2024-06-17T01:21:50 | 2024-06-20T08:01:14 | Yushi Sun, Hao Xin, Kai Sun, Yifan Ethan Xu, Xiao Yang, Xin Luna Dong, Nan Tang, Lei Chen | cs.CL | cs.CL, cs.AI, cs.DB | Large language models (LLMs) demonstrate an impressive ability to internalize
knowledge and answer natural language questions. Although previous studies
validate that LLMs perform well on general knowledge while presenting poor
performance on long-tail nuanced knowledge, the community is still doubtful
about whether th... | [
-0.3097949922,
-0.6377136707,
0.5380216241,
-0.16648148,
-0.3933236003,
0.3787720799,
-0.3736402392,
-0.6182379127,
0.2841249108,
0.4167536497,
-0.1097565517,
-0.5202380419,
-0.6486040354,
0.192293033,
-0.4239507616,
0.4879246056,
-0.0378942192,
-0.0022056217,
-0.6159538627,
-0... |
http://arxiv.org/abs/2406.11130v1 | http://arxiv.org/pdf/2406.11130v1 | Dynamic Order Template Prediction for Generative Aspect-Based Sentiment Analysis | 2024-06-17T01:21:28 | 2024-06-17T01:21:28 | Yonghyun Jun, Hwanhee Lee | cs.CL | cs.CL | Aspect-based sentiment analysis (ABSA) assesses sentiments towards specific
aspects within texts, resulting in detailed sentiment tuples. Previous ABSA
models often use static templates to predict all of the elements in the tuples,
and these models often fail to accurately capture dependencies between
elements. Multi-v... | [
-0.5499736667,
-0.6795812249,
0.5886235237,
0.2694815099,
-0.6415034533,
0.2216534168,
0.3557231724,
-0.3471335769,
0.563043654,
0.0891123414,
-0.5044990182,
-0.572191298,
-0.458345443,
0.09574496,
-0.2511916757,
0.5414577723,
0.1470342129,
-0.0218426008,
-0.2656424046,
-0.1228... |
http://arxiv.org/abs/2406.11116v1 | http://arxiv.org/pdf/2406.11116v1 | Grammaticality Representation in ChatGPT as Compared to Linguists and Laypeople | 2024-06-17T00:23:16 | 2024-06-17T00:23:16 | Zhuang Qiu, Xufeng Duan, Zhenguang G. Cai | cs.CL | cs.CL | Large language models (LLMs) have demonstrated exceptional performance across
various linguistic tasks. However, it remains uncertain whether LLMs have
developed human-like fine-grained grammatical intuition. This preregistered
study (https://osf.io/t5nes) presents the first large-scale investigation of
ChatGPT's gramm... | [
-0.2648616731,
-0.7067267895,
0.7785112262,
0.5225029588,
-0.247794047,
-0.1477435529,
-0.7435858846,
-0.3475809097,
0.2715749443,
0.0078742029,
-0.3025262356,
-0.4745597839,
-0.7638944387,
0.0910246894,
-0.2540620565,
0.8860140443,
0.4202919006,
0.2293099761,
-0.3367243409,
0.... |
http://arxiv.org/abs/2406.11115v1 | http://arxiv.org/pdf/2406.11115v1 | Text Grafting: Near-Distribution Weak Supervision for Minority Classes in Text Classification | 2024-06-17T00:23:08 | 2024-06-17T00:23:08 | Letian Peng, Yi Gu, Chengyu Dong, Zihan Wang, Jingbo Shang | cs.CL | cs.CL | For extremely weak-supervised text classification, pioneer research generates
pseudo labels by mining texts similar to the class names from the raw corpus,
which may end up with very limited or even no samples for the minority classes.
Recent works have started to generate the relevant texts by prompting LLMs
using the... | [
-0.0490863174,
-0.5882824659,
0.3081720471,
0.1251894236,
-0.347410202,
0.3743441701,
-0.0804014057,
-0.618183434,
0.3345660567,
0.5878233314,
-0.3324762285,
-0.5917807817,
-0.4037582278,
0.1025559232,
-0.6698814034,
1.4034692049,
-0.1018700153,
-0.2072249055,
-0.2030838728,
0.... |
http://arxiv.org/abs/2406.11109v2 | http://arxiv.org/pdf/2406.11109v2 | Investigating Annotator Bias in Large Language Models for Hate Speech Detection | 2024-06-17T00:18:31 | 2024-06-18T06:21:16 | Amit Das, Zheng Zhang, Fatemeh Jamshidi, Vinija Jain, Aman Chadha, Nilanjana Raychawdhary, Mary Sandage, Lauramarie Pope, Gerry Dozier, Cheryl Seals | cs.CL | cs.CL, cs.AI, cs.LG | Data annotation, the practice of assigning descriptive labels to raw data, is
pivotal in optimizing the performance of machine learning models. However, it
is a resource-intensive process susceptible to biases introduced by annotators.
The emergence of sophisticated Large Language Models (LLMs), like ChatGPT
presents a... | [
-0.5684790015,
-0.8388489485,
0.1359138191,
-0.0675861761,
-0.1634345204,
0.1562730074,
-0.2925709784,
-0.6949970722,
0.14402996,
0.5964358449,
-0.3343835771,
-0.3324960172,
-0.9046738744,
0.0032263284,
-0.3398251534,
1.5414181948,
0.0823260248,
0.0913453996,
0.1400823444,
-0.2... |
http://arxiv.org/abs/2406.11107v1 | http://arxiv.org/pdf/2406.11107v1 | Exploring Safety-Utility Trade-Offs in Personalized Language Models | 2024-06-17T00:17:11 | 2024-06-17T00:17:11 | Anvesh Rao Vijjini, Somnath Basu Roy Chowdhury, Snigdha Chaturvedi | cs.CL | cs.CL | As large language models (LLMs) become increasingly integrated into daily
applications, it is essential to ensure they operate fairly across diverse user
demographics. In this work, we show that LLMs suffer from personalization bias,
where their performance is impacted when they are personalized to a user's
identity. W... | [
-0.2061573714,
-0.6045250893,
0.7141955495,
0.0816085711,
0.0628938377,
0.1498121023,
-0.1885012239,
-0.4438951313,
-0.2125659883,
0.750020802,
-0.3118105233,
-0.5732616186,
-0.5827792883,
0.0037582298,
-0.0359124988,
1.0739235878,
0.2387208939,
-0.1675741822,
-0.1893662661,
0.... |
http://arxiv.org/abs/2406.11106v1 | http://arxiv.org/pdf/2406.11106v1 | From Intentions to Techniques: A Comprehensive Taxonomy and Challenges in Text Watermarking for Large Language Models | 2024-06-17T00:09:31 | 2024-06-17T00:09:31 | Harsh Nishant Lalai, Aashish Anantha Ramakrishnan, Raj Sanjay Shah, Dongwon Lee | cs.CL | cs.CL, cs.AI | With the rapid growth of Large Language Models (LLMs), safeguarding textual
content against unauthorized use is crucial. Text watermarking offers a vital
solution, protecting both - LLM-generated and plain text sources. This paper
presents a unified overview of different perspectives behind designing
watermarking techn... | [
-0.0241802596,
-0.7426496744,
0.7868234515,
0.030228829,
-0.4943702519,
0.1622652709,
-0.0117213288,
-0.5226586461,
0.0608669631,
1.0943946838,
-0.3641218245,
-0.3844343722,
-0.9670155048,
0.1772072166,
-0.6465368271,
0.9696795344,
-0.2716116309,
-0.1812199056,
-0.3392719626,
-... |
http://arxiv.org/abs/2406.11102v1 | http://arxiv.org/pdf/2406.11102v1 | Grading Massive Open Online Courses Using Large Language Models | 2024-06-16T23:42:11 | 2024-06-16T23:42:11 | Shahriar Golchin, Nikhil Garuda, Christopher Impey, Matthew Wenger | cs.CL | cs.CL, cs.AI | Massive open online courses (MOOCs) offer free education globally to anyone
with a computer and internet access. Despite this democratization of learning,
the massive enrollment in these courses makes it impractical for one instructor
to assess every student's writing assignment. As a result, peer grading, often
guided... | [
-0.1864946187,
-0.895410955,
0.8903315663,
-0.1547543406,
-0.1555587351,
0.1246795878,
-0.6258291602,
-0.446692884,
0.0624291301,
0.3944585621,
-0.3452810049,
-0.6838048697,
-0.4271476865,
-0.0278366022,
-0.2164437771,
1.509860158,
-0.2743259072,
0.3518278599,
-0.3690691292,
0.... |
http://arxiv.org/abs/2406.15477v1 | http://arxiv.org/pdf/2406.15477v1 | CrisisSense-LLM: Instruction Fine-Tuned Large Language Model for Multi-label Social Media Text Classification in Disaster Informatics | 2024-06-16T23:01:10 | 2024-06-16T23:01:10 | Kai Yin, Chengkai Liu, Ali Mostafavi, Xia Hu | cs.CL | cs.CL, cs.AI | In the field of crisis/disaster informatics, social media is increasingly
being used for improving situational awareness to inform response and relief
efforts. Efficient and accurate text classification tools have been a focal
area of investigation in crisis informatics. However, current methods mostly
rely on single-l... | [
-0.2391689569,
-0.7201589942,
0.3630447686,
0.5491187572,
-0.2329220325,
0.6581006646,
-0.3530855179,
-0.2276452184,
0.2650880814,
0.330788523,
-0.5671025515,
-0.8747933507,
-0.6567043662,
0.0280126929,
-0.4437769055,
1.087443471,
0.005850337,
-0.069843024,
0.3502022326,
0.1112... |
http://arxiv.org/abs/2406.11097v2 | http://arxiv.org/pdf/2406.11097v2 | InstructCMP: Length Control in Sentence Compression through Instruction-based Large Language Models | 2024-06-16T23:00:47 | 2024-06-18T18:35:52 | Juseon-Do, Jingun Kwon, Hidetaka Kamigaito, Manabu Okumura | cs.CL | cs.CL, cs.AI, I.2.7 | Extractive summarization can produce faithful summaries but often requires
additional constraints such as a desired summary length. Traditional sentence
compression models do not typically consider the constraints because of their
restricted model abilities, which require model modifications for coping with
them. To br... | [
-0.1360982507,
-0.625854075,
0.6078840494,
0.1989514828,
-0.4630026221,
0.0119442856,
-0.5567320585,
-0.2312566787,
0.117528148,
0.6530662775,
-0.544940412,
-0.4312418401,
-0.9285263419,
0.1717293262,
-0.2726252675,
0.9967310429,
-0.1078281105,
-0.041980844,
-0.4716852009,
0.34... |
http://arxiv.org/abs/2406.11096v2 | http://arxiv.org/pdf/2406.11096v2 | The Potential and Challenges of Evaluating Attitudes, Opinions, and Values in Large Language Models | 2024-06-16T22:59:18 | 2024-07-01T10:04:09 | Bolei Ma, Xinpeng Wang, Tiancheng Hu, Anna-Carolina Haensch, Michael A. Hedderich, Barbara Plank, Frauke Kreuter | cs.CL | cs.CL | Recent advances in Large Language Models (LLMs) have sparked wide interest in
validating and comprehending the human-like cognitive-behavioral traits LLMs
may have. These cognitive-behavioral traits include typically Attitudes,
Opinions, Values (AOV). However, measuring AOV embedded within LLMs remains
opaque, and diff... | [
-0.3942783177,
-0.7712155581,
0.8691872358,
0.0755765885,
-0.0259840526,
0.2397295982,
-0.5205266476,
-0.4224661291,
-0.1197477281,
0.329760462,
-0.4970727861,
-0.3761198223,
-0.7822669744,
-0.0879188925,
-0.101928845,
0.8568719029,
-0.0650395527,
0.3083543181,
-0.4497328699,
-... |
http://arxiv.org/abs/2406.11093v1 | http://arxiv.org/pdf/2406.11093v1 | RAEmoLLM: Retrieval Augmented LLMs for Cross-Domain Misinformation Detection Using In-Context Learning based on Emotional Information | 2024-06-16T22:49:11 | 2024-06-16T22:49:11 | Zhiwei Liu, Kailai Yang, Qianqian Xie, Christine de Kock, Sophia Ananiadou, Eduard Hovy | cs.CL | cs.CL | Misinformation is prevalent in various fields such as education, politics,
health, etc., causing significant harm to society. However, current methods for
cross-domain misinformation detection rely on time and resources consuming
fine-tuning and complex model structures. With the outstanding performance of
LLMs, many s... | [
-0.3247259259,
-0.8317123652,
0.3420359194,
0.3238455951,
-0.1391255707,
0.1821706444,
-0.0745328814,
-0.6905113459,
0.5959322453,
0.1690750569,
-0.495552808,
-0.6642531753,
-0.8015477061,
0.2113454193,
-0.2516948581,
0.7652196884,
-0.1424520612,
-0.0209378842,
-0.4202396572,
-... |
http://arxiv.org/abs/2406.11087v2 | http://arxiv.org/pdf/2406.11087v2 | MemDPT: Differential Privacy for Memory Efficient Language Models | 2024-06-16T22:11:41 | 2024-06-20T05:43:50 | Yanming Liu, Xinyue Peng, Jiannan Cao, Yuwei Zhang, Chen Ma, Songhang Deng, Mengchen Fu, Xuhong Zhang, Sheng Cheng, Xun Wang, Jianwei Yin, Tianyu Du | cs.CR | cs.CR, cs.AI, cs.CL, cs.LG | Large language models have consistently demonstrated remarkable performance
across a wide spectrum of applications. Nonetheless, the deployment of these
models can inadvertently expose user privacy to potential risks. The
substantial memory demands of these models during training represent a
significant resource consum... | [
-0.4388596714,
-0.5826792717,
0.2035086602,
0.0548379533,
-0.2324981093,
0.0614452474,
-0.0620787814,
-0.1963827461,
-0.0827162042,
0.5419732928,
-0.7084451318,
-0.4213849902,
-0.4126570225,
-0.1807582527,
-0.0654341429,
0.5243942142,
0.0082900422,
0.093176648,
0.2696002126,
-0... |
http://arxiv.org/abs/2406.11085v1 | http://arxiv.org/pdf/2406.11085v1 | Multiple Sources are Better Than One: Incorporating External Knowledge in Low-Resource Glossing | 2024-06-16T22:01:15 | 2024-06-16T22:01:15 | Changbing Yang, Garrett Nicolai, Miikka Silfverberg | cs.CL | cs.CL | In this paper, we address the data scarcity problem in automatic data-driven
glossing for low-resource languages by coordinating multiple sources of
linguistic expertise. We supplement models with translations at both the token
and sentence level as well as leverage the extensive linguistic capability of
modern LLMs. O... | [
-0.3775053322,
-0.4170348346,
0.6922374368,
0.1854874194,
-0.0752539486,
0.2620881796,
-0.6509305239,
-0.5729327202,
0.2780018449,
0.3853574991,
-0.259211272,
-0.3936476111,
-0.7568491101,
0.1958700866,
-0.1698152572,
0.8078330159,
-0.1527145803,
0.2208266556,
-0.0696164742,
-0... |
http://arxiv.org/abs/2406.15476v1 | http://arxiv.org/pdf/2406.15476v1 | Self-Regulated Data-Free Knowledge Amalgamation for Text Classification | 2024-06-16T21:13:30 | 2024-06-16T21:13:30 | Prashanth Vijayaraghavan, Hongzhi Wang, Luyao Shi, Tyler Baldwin, David Beymer, Ehsan Degan | cs.CL | cs.CL | Recently, there has been a growing availability of pre-trained text models on
various model repositories. These models greatly reduce the cost of training
new models from scratch as they can be fine-tuned for specific tasks or trained
on large datasets. However, these datasets may not be publicly accessible due
to the ... | [
0.0645011961,
-0.0184809379,
0.4512146413,
-0.0328536592,
-0.2568601668,
0.0078183776,
0.0287320893,
-0.4234845638,
0.1320532113,
0.7645108104,
-0.2844776511,
-0.5546978116,
-0.598317802,
0.0528337881,
-0.8638300896,
1.0487239361,
-0.1308723837,
-0.2405442446,
-0.4693556428,
0.... |
http://arxiv.org/abs/2406.11073v1 | http://arxiv.org/pdf/2406.11073v1 | Exploring the Limitations of Detecting Machine-Generated Text | 2024-06-16T21:02:02 | 2024-06-16T21:02:02 | Jad Doughman, Osama Mohammed Afzal, Hawau Olamide Toyin, Shady Shehata, Preslav Nakov, Zeerak Talat | cs.CL | cs.CL | Recent improvements in the quality of the generations by large language
models have spurred research into identifying machine-generated text. Systems
proposed for the task often achieve high performance. However, humans and
machines can produce text in different styles and in different domains, and it
remains unclear w... | [
-0.3277284205,
-0.7192645073,
0.6610377431,
0.1004896462,
-0.0148455808,
0.1725183576,
-0.3293474317,
-0.9330388904,
-0.3348592222,
0.6485029459,
-0.4112384319,
-0.3901267946,
-0.9795871973,
0.7104824185,
-0.3531577587,
1.0527017117,
-0.2600163519,
-0.3901310563,
-0.2169965357,
... |
http://arxiv.org/abs/2406.11069v1 | http://arxiv.org/pdf/2406.11069v1 | WildVision: Evaluating Vision-Language Models in the Wild with Human Preferences | 2024-06-16T20:53:25 | 2024-06-16T20:53:25 | Yujie Lu, Dongfu Jiang, Wenhu Chen, William Yang Wang, Yejin Choi, Bill Yuchen Lin | cs.CV | cs.CV, cs.AI, cs.CL | Recent breakthroughs in vision-language models (VLMs) emphasize the necessity
of benchmarking human preferences in real-world multimodal interactions. To
address this gap, we launched WildVision-Arena (WV-Arena), an online platform
that collects human preferences to evaluate VLMs. We curated WV-Bench by
selecting 500 h... | [
-0.4689777195,
-0.4557703733,
0.6050419211,
0.244851917,
-0.2967833579,
-0.2012158334,
-0.1797201186,
-0.8574142456,
-0.0445239879,
0.5911564231,
-0.3964504004,
-0.4608893096,
-0.3162148297,
-0.10799402,
-0.3228135407,
0.6212417483,
-0.0022890461,
0.1292525828,
-0.2275558412,
-... |
http://arxiv.org/abs/2406.11065v1 | http://arxiv.org/pdf/2406.11065v1 | Can LLMs Understand the Implication of Emphasized Sentences in Dialogue? | 2024-06-16T20:41:44 | 2024-06-16T20:41:44 | Guan-Ting Lin, Hung-yi Lee | cs.CL | cs.CL | Emphasis is a crucial component in human communication, which indicates the
speaker's intention and implication beyond pure text in dialogue. While Large
Language Models (LLMs) have revolutionized natural language processing, their
ability to understand emphasis in dialogue remains unclear. This paper
introduces Emphas... | [
-0.3796223104,
-0.9975184202,
0.9999467134,
0.3681863248,
-0.508656919,
-0.1762265563,
-0.5543020964,
-0.7991424799,
-0.007312438,
0.4252296686,
-0.4218037426,
-0.4441881776,
-0.7611021996,
0.3322784305,
-0.3276628554,
0.7308626175,
0.0621479973,
0.0066649029,
-0.0518628247,
-0... |
http://arxiv.org/abs/2406.11050v1 | http://arxiv.org/pdf/2406.11050v1 | A Peek into Token Bias: Large Language Models Are Not Yet Genuine Reasoners | 2024-06-16T19:22:53 | 2024-06-16T19:22:53 | Bowen Jiang, Yangxinyu Xie, Zhuoqun Hao, Xiaomeng Wang, Tanwi Mallick, Weijie J. Su, Camillo J. Taylor, Dan Roth | cs.CL | cs.CL, cs.AI | This study introduces a hypothesis-testing framework to assess whether large
language models (LLMs) possess genuine reasoning abilities or primarily depend
on token bias. We go beyond evaluating LLMs on accuracy; rather, we aim to
investigate their token bias in solving logical reasoning tasks. Specifically,
we develop... | [
-0.2189116329,
-0.8572227359,
1.0043822527,
0.18308267,
-0.1310603172,
0.0560876429,
-0.1300462484,
-0.1584148705,
0.1919850707,
0.5241861343,
-0.5823005438,
-0.2188230306,
-0.6717125773,
0.2370414883,
-0.5674058795,
0.7065299153,
0.0794890895,
0.4392490387,
-0.3736979067,
-0.1... |
http://arxiv.org/abs/2406.11049v1 | http://arxiv.org/pdf/2406.11049v1 | Reconsidering Sentence-Level Sign Language Translation | 2024-06-16T19:19:54 | 2024-06-16T19:19:54 | Garrett Tanzer, Maximus Shengelia, Ken Harrenstien, David Uthus | cs.CL | cs.CL | Historically, sign language machine translation has been posed as a
sentence-level task: datasets consisting of continuous narratives are chopped
up and presented to the model as isolated clips. In this work, we explore the
limitations of this task framing. First, we survey a number of linguistic
phenomena in sign lang... | [
0.1263560355,
-0.3493473232,
0.4067889154,
0.1110915914,
-0.3569330573,
-0.0288630631,
0.0864279643,
-0.5110616088,
-0.2716988921,
0.7049850821,
-0.5581213236,
-0.5769225955,
-1.0795475245,
0.4227745533,
-0.4375248253,
1.1103476286,
-0.326167345,
-0.1438663751,
0.0167730246,
-0... |
http://arxiv.org/abs/2406.11044v1 | http://arxiv.org/pdf/2406.11044v1 | Evaluating the Performance of Large Language Models via Debates | 2024-06-16T19:02:31 | 2024-06-16T19:02:31 | Behrad Moniri, Hamed Hassani, Edgar Dobriban | cs.CL | cs.CL, cs.AI, cs.LG | Large Language Models (LLMs) are rapidly evolving and impacting various
fields, necessitating the development of effective methods to evaluate and
compare their performance. Most current approaches for performance evaluation
are either based on fixed, domain-specific questions that lack the flexibility
required in many... | [
-0.5943880081,
-0.9035658836,
0.5185980201,
0.1503967643,
0.0310453363,
0.2518402934,
-0.5878498554,
-0.4755961895,
-0.0360631719,
0.6141065955,
-0.2799179256,
-0.3739552796,
-0.757901907,
0.1650623232,
-0.3670042157,
0.4135926068,
-0.0844330788,
0.1080625728,
-0.5394943953,
-0... |
http://arxiv.org/abs/2406.11036v1 | http://arxiv.org/pdf/2406.11036v1 | garak: A Framework for Security Probing Large Language Models | 2024-06-16T18:18:43 | 2024-06-16T18:18:43 | Leon Derczynski, Erick Galinkin, Jeffrey Martin, Subho Majumdar, Nanna Inie | cs.CL | cs.CL, cs.CR | As Large Language Models (LLMs) are deployed and integrated into thousands of
applications, the need for scalable evaluation of how models respond to
adversarial attacks grows rapidly. However, LLM security is a moving target:
models produce unpredictable output, are constantly updated, and the potential
adversary is h... | [
-0.6145964265,
-1.2226730585,
0.3891807795,
-0.0677163526,
-0.0141691584,
0.4089111388,
-0.110050723,
-0.4433082044,
0.1378272772,
0.7057034969,
-0.5244705677,
-0.5187655091,
-0.7245429754,
-0.2366915643,
-0.5393665433,
1.2842915058,
-0.0279583484,
-0.3640417755,
-0.0227732398,
... |
http://arxiv.org/abs/2406.11035v1 | http://arxiv.org/pdf/2406.11035v1 | Scaling Synthetic Logical Reasoning Datasets with Context-Sensitive Declarative Grammars | 2024-06-16T18:10:49 | 2024-06-16T18:10:49 | Damien Sileo | cs.CL | cs.CL, I.2.7 | Logical reasoning remains a challenge for natural language processing, but it
can be improved by training language models to mimic theorem provers on
procedurally generated problems. Previous work used domain-specific proof
generation algorithms, which biases reasoning toward specific proof traces and
limits auditabili... | [
-0.0959851891,
-0.7009065747,
0.7707433701,
0.0883640721,
-0.1221675202,
-0.2983560562,
-0.3945187628,
-0.3610893786,
-0.2547345757,
0.4269829094,
-0.6068210006,
-0.4518432617,
-0.4798328578,
0.4294717908,
-0.0657846108,
0.5673804879,
0.0237917211,
0.0203611255,
-0.2059227526,
... |
http://arxiv.org/abs/2406.11030v1 | http://arxiv.org/pdf/2406.11030v1 | FoodieQA: A Multimodal Dataset for Fine-Grained Understanding of Chinese Food Culture | 2024-06-16T17:59:32 | 2024-06-16T17:59:32 | Wenyan Li, Xinyu Zhang, Jiaang Li, Qiwei Peng, Raphael Tang, Li Zhou, Weijia Zhang, Guimin Hu, Yifei Yuan, Anders Søgaard, Daniel Hershcovich, Desmond Elliott | cs.CL | cs.CL | Food is a rich and varied dimension of cultural heritage, crucial to both
individuals and social groups. To bridge the gap in the literature on the
often-overlooked regional diversity in this domain, we introduce FoodieQA, a
manually curated, fine-grained image-text dataset capturing the intricate
features of food cult... | [
-0.3632334769,
-0.5674921274,
0.3247353435,
-0.1287643909,
0.1304505765,
0.000447425,
-0.2125263065,
-0.6502767801,
0.0878715068,
0.4928760827,
-0.6175938845,
-0.3590791821,
-0.4962178171,
-0.0629544705,
-0.2130785882,
0.7208910584,
-0.158101663,
0.3609746695,
0.024211375,
-0.4... |
http://arxiv.org/abs/2406.11029v1 | http://arxiv.org/pdf/2406.11029v1 | Curating Stopwords in Marathi: A TF-IDF Approach for Improved Text Analysis and Information Retrieval | 2024-06-16T17:59:05 | 2024-06-16T17:59:05 | Rohan Chavan, Gaurav Patil, Vishal Madle, Raviraj Joshi | cs.CL | cs.CL, cs.LG | Stopwords are commonly used words in a language that are often considered to
be of little value in determining the meaning or significance of a document.
These words occur frequently in most texts and don't provide much useful
information for tasks like sentiment analysis and text classification. English,
which is a hi... | [
-0.5963807106,
-1.0155158043,
0.2468867302,
0.3682014346,
-0.7325467467,
0.3362228274,
-0.2950864434,
-0.4882336855,
0.4129756987,
0.7536655068,
-0.3268662989,
-0.4955720305,
-0.8700171113,
0.385738641,
-0.449770689,
0.8982642889,
-0.1655506045,
-0.24937509,
-0.049603913,
-0.60... |
http://arxiv.org/abs/2406.11028v1 | http://arxiv.org/pdf/2406.11028v1 | Universal Cross-Lingual Text Classification | 2024-06-16T17:58:29 | 2024-06-16T17:58:29 | Riya Savant, Anushka Shelke, Sakshi Todmal, Sanskruti Kanphade, Ananya Joshi, Raviraj Joshi | cs.CL | cs.CL, cs.LG | Text classification, an integral task in natural language processing,
involves the automatic categorization of text into predefined classes. Creating
supervised labeled datasets for low-resource languages poses a considerable
challenge. Unlocking the language potential of low-resource languages requires
robust datasets... | [
-0.2671378553,
-0.3211482167,
0.0099817878,
0.2182938755,
-0.1786604226,
0.1713946909,
-0.2700284123,
-0.8708398938,
-0.06636516,
0.6096051335,
-0.3039418757,
-0.4664341807,
-0.5027256012,
0.2203520685,
-0.2063309401,
1.3534913063,
-0.166786015,
0.1652677506,
0.1130776927,
-0.4... |
http://arxiv.org/abs/2406.11025v1 | http://arxiv.org/pdf/2406.11025v1 | Large Language Models for Dysfluency Detection in Stuttered Speech | 2024-06-16T17:51:22 | 2024-06-16T17:51:22 | Dominik Wagner, Sebastian P. Bayerl, Ilja Baumann, Korbinian Riedhammer, Elmar Nöth, Tobias Bocklet | cs.SD | cs.SD, cs.CL, eess.AS | Accurately detecting dysfluencies in spoken language can help to improve the
performance of automatic speech and language processing components and support
the development of more inclusive speech and language technologies. Inspired by
the recent trend towards the deployment of large language models (LLMs) as
universal... | [
-0.3555168211,
-1.0059794188,
0.5284237266,
0.2409139872,
0.0191010702,
0.2399532199,
-0.1550815254,
-0.6672909856,
0.2145408392,
0.512494266,
-0.270319134,
-0.1091423258,
-0.7793136239,
0.2740668654,
-0.6951096058,
0.742412746,
0.0990942493,
0.212676689,
-0.0092549128,
0.07817... |
http://arxiv.org/abs/2406.11020v1 | http://arxiv.org/pdf/2406.11020v1 | RUPBench: Benchmarking Reasoning Under Perturbations for Robustness Evaluation in Large Language Models | 2024-06-16T17:26:44 | 2024-06-16T17:26:44 | Yuqing Wang, Yun Zhao | cs.CL | cs.CL | With the increasing use of large language models (LLMs), ensuring reliable
performance in diverse, real-world environments is essential. Despite their
remarkable achievements, LLMs often struggle with adversarial inputs,
significantly impacting their effectiveness in practical applications. To
systematically understand... | [
-0.2397341281,
-1.0102756023,
0.6822835207,
0.2241910994,
-0.0185008161,
-0.0636541769,
-0.3100273907,
-0.2006951421,
-0.4072684944,
0.4406892359,
-0.4311669171,
-0.627509892,
-0.7996467352,
0.1229179278,
-0.4404576123,
0.6568223238,
-0.0127156312,
0.2812670171,
-0.3937658668,
... |
http://arxiv.org/abs/2406.11016v1 | http://arxiv.org/pdf/2406.11016v1 | Optimized Speculative Sampling for GPU Hardware Accelerators | 2024-06-16T17:19:23 | 2024-06-16T17:19:23 | Dominik Wagner, Seanie Lee, Ilja Baumann, Philipp Seeberger, Korbinian Riedhammer, Tobias Bocklet | cs.LG | cs.LG, cs.CL | In this work, we optimize speculative sampling for parallel hardware
accelerators to improve sampling speed. We notice that substantial portions of
the intermediate matrices necessary for speculative sampling can be computed
concurrently. This allows us to distribute the workload across multiple GPU
threads, enabling s... | [
-0.6984532475,
-1.0324518681,
0.7315135598,
0.1198373139,
-0.2630893588,
0.3857552409,
-0.2118841708,
-0.3859548569,
0.2840053439,
-0.0719060227,
-0.374065727,
-0.1279894114,
-0.4245093763,
-0.2480850518,
-0.2211512178,
0.7299087644,
-0.0439277403,
-0.2669922113,
-0.1503127664,
... |
http://arxiv.org/abs/2406.11012v3 | http://arxiv.org/pdf/2406.11012v3 | Connecting the Dots: Evaluating Abstract Reasoning Capabilities of LLMs Using the New York Times Connections Word Game | 2024-06-16T17:10:32 | 2024-06-22T15:40:14 | Prisha Samadarshi, Mariam Mustafa, Anushka Kulkarni, Raven Rothkopf, Tuhin Chakrabarty, Smaranda Muresan | cs.CL | cs.CL, cs.AI | The New York Times Connections game has emerged as a popular and challenging
pursuit for word puzzle enthusiasts. We collect 200 Connections games to
evaluate the performance of state-of-the-art large language models (LLMs)
against expert and novice human players. Our results show that even the
best-performing LLM, GPT... | [
-0.3957364559,
-0.9535313845,
1.0459669828,
0.0145747997,
-0.2152581364,
0.4644643068,
-0.0364094116,
-0.7716236711,
0.0427584462,
0.3504739702,
-0.4365972579,
-0.4895687103,
-0.6086264849,
0.4937179089,
-0.3355938494,
0.6229938865,
-0.0533108041,
-0.1995043606,
-0.740868628,
0... |
http://arxiv.org/abs/2406.11011v2 | http://arxiv.org/pdf/2406.11011v2 | Data Shapley in One Training Run | 2024-06-16T17:09:24 | 2024-06-29T23:05:32 | Jiachen T. Wang, Prateek Mittal, Dawn Song, Ruoxi Jia | cs.LG | cs.LG, cs.CL, stat.ML | Data Shapley provides a principled framework for attributing data's
contribution within machine learning contexts. However, existing approaches
require re-training models on different data subsets, which is computationally
intensive, foreclosing their application to large-scale models. Furthermore,
they produce the sam... | [
0.0789565891,
-0.4574116766,
0.3813570142,
0.1536251605,
-0.2095779777,
-0.1027419642,
0.1399288476,
-0.1320853531,
0.1453692615,
0.8305829167,
-0.6032734513,
0.0689253435,
-0.5019671321,
-0.3752999008,
-0.8100118637,
1.0774029493,
-0.102489382,
0.0520289205,
-0.948138237,
-0.0... |
http://arxiv.org/abs/2406.11911v1 | http://arxiv.org/pdf/2406.11911v1 | A Notion of Complexity for Theory of Mind via Discrete World Models | 2024-06-16T16:46:55 | 2024-06-16T16:46:55 | X. Angelo Huang, Emanuele La Malfa, Samuele Marro, Andrea Asperti, Anthony Cohn, Michael Wooldridge | cs.AI | cs.AI, cs.CL, cs.LG | Theory of Mind (ToM) can be used to assess the capabilities of Large Language
Models (LLMs) in complex scenarios where social reasoning is required. While
the research community has proposed many ToM benchmarks, their hardness varies
greatly, and their complexity is not well defined. This work proposes a
framework to m... | [
-0.6688780785,
-0.9313619733,
1.1467983723,
0.227149263,
-0.1986577958,
0.0262198746,
-0.387123704,
-0.4795284569,
-0.1647303402,
0.0064972891,
-0.7680091262,
0.0347476453,
-0.6415761709,
0.1293286234,
-0.0956869647,
0.6072156429,
-0.0786194205,
-0.0550468713,
-0.3464925289,
-0... |
http://arxiv.org/abs/2406.10999v1 | http://arxiv.org/pdf/2406.10999v1 | Not All Bias is Bad: Balancing Rational Deviations and Cognitive Biases in Large Language Model Reasoning | 2024-06-16T16:25:22 | 2024-06-16T16:25:22 | Liman Wang, Hanyang Zhong | cs.CL | cs.CL, cs.AI | This paper investigates the nuanced role of biases in the decision-making
processes of large language models (LLMs). While conventional research
typically aims to eliminate all biases, our study reveals that not all biases
are detrimental. By examining rational deviations, involving heuristic
shortcuts that enhance dec... | [
-0.5140390396,
-0.7370125651,
0.760427773,
0.1371153444,
-0.1296262443,
-0.0583610125,
-0.5258209705,
-0.3876025379,
-0.0825936422,
0.6565509439,
-0.4380590022,
-0.2248435766,
-0.5842162371,
-0.1183157936,
-0.035389591,
0.8679662943,
-0.0208577644,
0.3191327453,
-0.3463336229,
... |
http://arxiv.org/abs/2406.10996v1 | http://arxiv.org/pdf/2406.10996v1 | THEANINE: Revisiting Memory Management in Long-term Conversations with Timeline-augmented Response Generation | 2024-06-16T16:17:46 | 2024-06-16T16:17:46 | Seo Hyun Kim, Kai Tzu-iunn Ong, Taeyoon Kwon, Namyoung Kim, Keummin Ka, SeongHyeon Bae, Yohan Jo, Seung-won Hwang, Dongha Lee, Jinyoung Yeo | cs.CL | cs.CL | Large language models (LLMs) are capable of processing lengthy dialogue
histories during prolonged interaction with users without additional memory
modules; however, their responses tend to overlook or incorrectly recall
information from the past. In this paper, we revisit memory-augmented response
generation in the er... | [
-0.3484933078,
-1.1728732586,
0.5873211622,
-0.2838815451,
-0.109388873,
0.0834346339,
-0.1195364445,
-0.3436455727,
0.519159615,
0.7483797669,
-0.8087123632,
0.0090108039,
-0.1843334585,
-0.298502475,
-0.3996779025,
0.9334490895,
0.0317929313,
-0.1300593317,
-0.3081113398,
-0.... |
http://arxiv.org/abs/2406.10993v1 | http://arxiv.org/pdf/2406.10993v1 | CoSTA: Code-Switched Speech Translation using Aligned Speech-Text Interleaving | 2024-06-16T16:10:51 | 2024-06-16T16:10:51 | Bhavani Shankar, Preethi Jyothi, Pushpak Bhattacharyya | cs.CL | cs.CL, cs.LG, cs.SD, eess.AS | Code-switching is a widely prevalent linguistic phenomenon in multilingual
societies like India. Building speech-to-text models for code-switched speech
is challenging due to limited availability of datasets. In this work, we focus
on the problem of spoken translation (ST) of code-switched speech in Indian
languages to... | [
-0.0106454557,
-0.5246779919,
0.0177461375,
0.7263356447,
-0.604208529,
0.5082615614,
-0.4544942677,
-0.5145124197,
-0.0300999489,
0.3637061715,
-0.3402118087,
-0.2342381179,
-0.5898008347,
0.1847209483,
-0.6641738415,
1.1021988392,
-0.1799039692,
0.5190832615,
-0.0984183475,
-... |
http://arxiv.org/abs/2406.10991v1 | http://arxiv.org/pdf/2406.10991v1 | Adaptive Query Rewriting: Aligning Rewriters through Marginal Probability of Conversational Answers | 2024-06-16T16:09:05 | 2024-06-16T16:09:05 | Tianhua Zhang, Kun Li, Hongyin Luo, Xixin Wu, James Glass, Helen Meng | cs.CL | cs.CL | Query rewriting is a crucial technique for passage retrieval in open-domain
conversational question answering (CQA). It decontexualizes conversational
queries into self-contained questions suitable for off-the-shelf retrievers.
Existing methods attempt to incorporate retriever's preference during the
training of rewrit... | [
-0.0458992384,
-0.9910895228,
0.4111124873,
-0.1682384908,
-0.2313692868,
-0.0470604077,
-0.0161274169,
0.0012027618,
0.1875000894,
0.8518298268,
-0.6178565025,
-0.4077302217,
-0.0616750978,
0.038064979,
-0.0700391456,
0.9399646521,
-0.0731150508,
-0.0432850569,
-0.5030109286,
... |
http://arxiv.org/abs/2406.15474v1 | http://arxiv.org/pdf/2406.15474v1 | WundtGPT: Shaping Large Language Models To Be An Empathetic, Proactive Psychologist | 2024-06-16T16:06:38 | 2024-06-16T16:06:38 | Chenyu Ren, Yazhou Zhang, Daihai He, Jing Qin | cs.AI | cs.AI, cs.CL, cs.HC | Large language models (LLMs) are raging over the medical domain, and their
momentum has carried over into the mental health domain, leading to the
emergence of few mental health LLMs. Although such mental health LLMs could
provide reasonable suggestions for psychological counseling, how to develop an
authentic and effe... | [
-0.496288389,
-0.7555292249,
1.0282956362,
0.3104133606,
-0.2860238254,
-0.1351680905,
-0.5478987098,
-0.3629230559,
0.1838756502,
0.3229125738,
-0.6377249956,
-0.1386853755,
-0.7978479266,
-0.1546566337,
-0.0385373123,
0.8402399421,
0.0054159891,
0.2790870965,
0.1844886392,
0.... |
http://arxiv.org/abs/2406.10985v1 | http://arxiv.org/pdf/2406.10985v1 | Taking a Deep Breath: Enhancing Language Modeling of Large Language Models with Sentinel Tokens | 2024-06-16T15:50:10 | 2024-06-16T15:50:10 | Weiyao Luo, Suncong Zheng, Heming Xia, Weikang Wang, Yan Lei, Tianyu Liu, Shuang Chen, Zhifang Sui | cs.CL | cs.CL | Large language models (LLMs) have shown promising efficacy across various
tasks, becoming powerful tools in numerous aspects of human life. However,
Transformer-based LLMs suffer a performance degradation when modeling long-term
contexts due to they discard some information to reduce computational overhead.
In this wor... | [
-0.1114068404,
-0.7211717367,
0.7123089433,
0.0590618663,
-0.7570325136,
0.4331852496,
-0.4402517974,
-0.2602036595,
0.3418976665,
0.6327492595,
-0.3013278246,
-0.5079062581,
-0.8046868443,
-0.1759651303,
-0.6531374454,
0.8145049214,
0.1817948818,
-0.3254671395,
-0.0269467384,
... |
http://arxiv.org/abs/2406.10984v1 | http://arxiv.org/pdf/2406.10984v1 | Revisiting Cosine Similarity via Normalized ICA-transformed Embeddings | 2024-06-16T15:44:37 | 2024-06-16T15:44:37 | Hiroaki Yamagiwa, Momose Oyama, Hidetoshi Shimodaira | cs.CL | cs.CL | Cosine similarity is widely used to measure the similarity between two
embeddings, while interpretations based on angle and correlation coefficient
are common. In this study, we focus on the interpretable axes of embeddings
transformed by Independent Component Analysis (ICA), and propose a novel
interpretation of cosin... | [
-0.6836594343,
-0.6076108217,
0.4706057012,
-0.0782828778,
-0.5745902658,
-0.4223728478,
0.0001284976,
-0.4182975292,
0.7759128213,
0.5189374685,
-0.5275756121,
-0.713762939,
-0.5795174241,
0.1203378364,
-0.652769804,
0.7906796336,
0.0668306276,
0.2062381208,
-0.3190580904,
-0.... |
http://arxiv.org/abs/2406.10977v1 | http://arxiv.org/pdf/2406.10977v1 | Toward Optimal LLM Alignments Using Two-Player Games | 2024-06-16T15:24:50 | 2024-06-16T15:24:50 | Rui Zheng, Hongyi Guo, Zhihan Liu, Xiaoying Zhang, Yuanshun Yao, Xiaojun Xu, Zhaoran Wang, Zhiheng Xi, Tao Gui, Qi Zhang, Xuanjing Huang, Hang Li, Yang Liu | cs.CL | cs.CL, cs.AI, 68 | The standard Reinforcement Learning from Human Feedback (RLHF) framework
primarily focuses on optimizing the performance of large language models using
pre-collected prompts. However, collecting prompts that provide comprehensive
coverage is both tedious and challenging, and often fails to include scenarios
that LLMs n... | [
-0.1134684384,
-0.5881931186,
0.6380167007,
-0.0851929635,
0.0106358863,
0.3512018919,
-0.367905736,
-0.1719598025,
-0.1664133668,
0.4454015195,
-0.8587935567,
-0.5022444725,
-0.4554437995,
-0.1442704797,
0.2944786251,
1.237760067,
-0.0944894329,
0.0831610039,
-0.2823777199,
-0... |
http://arxiv.org/abs/2406.10976v1 | http://arxiv.org/pdf/2406.10976v1 | Promoting Data and Model Privacy in Federated Learning through Quantized LoRA | 2024-06-16T15:23:07 | 2024-06-16T15:23:07 | JianHao Zhu, Changze Lv, Xiaohua Wang, Muling Wu, Wenhao Liu, Tianlong Li, Zixuan Ling, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang | cs.LG | cs.LG, cs.CL, cs.CR | Conventional federated learning primarily aims to secure the privacy of data
distributed across multiple edge devices, with the global model dispatched to
edge devices for parameter updates during the learning process. However, the
development of large language models (LLMs) requires substantial data and
computational ... | [
-0.3130512238,
-0.7702240348,
0.1491867006,
0.1636668146,
-0.1423775852,
-0.0475562327,
-0.0969914421,
-0.4574140012,
-0.0319493078,
0.7054268122,
-0.5473936796,
-0.8849936724,
-0.1368733048,
-0.3190152347,
-0.4413782656,
1.1149523258,
0.1851973087,
-0.0208406691,
-0.3666676581,
... |
http://arxiv.org/abs/2406.10974v1 | http://arxiv.org/pdf/2406.10974v1 | Towards Supporting Legal Argumentation with NLP: Is More Data Really All You Need? | 2024-06-16T15:15:44 | 2024-06-16T15:15:44 | T. Y. S. S Santosh, Kevin D. Ashley, Katie Atkinson, Matthias Grabmair | cs.CL | cs.CL, cs.AI | Modeling legal reasoning and argumentation justifying decisions in cases has
always been central to AI & Law, yet contemporary developments in legal NLP
have increasingly focused on statistically classifying legal conclusions from
text. While conceptually simpler, these approaches often fall short in
providing usable j... | [
-0.2031703144,
-0.6967307925,
0.7899125218,
0.125442341,
-0.3067464828,
-0.3232832551,
-0.0434289165,
-0.6915352941,
0.0280604362,
0.4289036989,
-0.2245147973,
-0.1997670233,
-0.6186042428,
-0.0248242989,
-0.3908699751,
1.007121563,
0.1450657248,
0.5101787448,
-0.7568950653,
-0... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.