id
stringlengths
33
34
title
stringlengths
6
218
authors
stringlengths
5
6.98k
categories
stringlengths
5
188
summary
stringlengths
34
3.26k
link_pdf
stringlengths
33
34
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/2310.03211v1
On the Performance of Multimodal Language Models
Utsav Garg, Erhan Bas
cs.CL, cs.AI, cs.CV
Instruction-tuned large language models (LLMs) have demonstrated promising zero-shot generalization capabilities across various downstream tasks. Recent research has introduced multimodal capabilities to LLMs by integrating independently pretrained vision encoders through model grafting. These multimodal variants under...
http://arxiv.org/pdf/2310.03211v1
2023-10-04T23:33:36
2023-10-04T23:33:36
{ "title": [ -0.038757324, 0.0051231384, -0.039978027, -0.05770874, 0.0010213852, -0.0010080338, -0.070617676, -0.0027885437, 0.0069770813, 0.04888916, -0.008880615, -0.011428833, 0.021957397, -0.013244629, -0.05859375, 0.014953613, -0.014915466, ...
http://arxiv.org/abs/2310.03210v1
Can Language Models Employ the Socratic Method? Experiments with Code Debugging
Erfan Al-Hossami, Razvan Bunescu, Justin Smith, Ryan Teehan
cs.CL, cs.CY
When employing the Socratic method of teaching, instructors guide students toward solving a problem on their own rather than providing the solution directly. While this strategy can substantially improve learning outcomes, it is usually time-consuming and cognitively demanding. Automated Socratic conversational agents ...
http://arxiv.org/pdf/2310.03210v1
2023-10-04T23:32:33
2023-10-04T23:32:33
{ "title": [ -0.03390503, -0.010345459, -0.025390625, -0.010955811, -0.02067566, -0.017181396, -0.035247803, -0.016525269, -0.003944397, 0.010620117, 0.00081539154, -0.024871826, -0.014083862, -0.03564453, -0.041381836, -0.009742737, -0.025115967, ...
http://arxiv.org/abs/2310.03193v1
The Rise of Open Science: Tracking the Evolution and Perceived Value of Data and Methods Link-Sharing Practices
Hancheng Cao, Jesse Dodge, Kyle Lo, Daniel A. McFarland, Lucy Lu Wang
cs.DL, cs.CL, cs.CY, physics.hist-ph, physics.soc-ph
In recent years, funding agencies and journals increasingly advocate for open science practices (e.g. data and method sharing) to improve the transparency, access, and reproducibility of science. However, quantifying these practices at scale has proven difficult. In this work, we leverage a large-scale dataset of 1.1M ...
http://arxiv.org/pdf/2310.03193v1
2023-10-04T22:34:56
2023-10-04T22:34:56
{ "title": [ -0.0035934448, 0.017547607, -0.06298828, -0.010116577, 0.024337769, -0.046142578, 0.005619049, 0.0025730133, -0.03564453, 0.016983032, -0.0043907166, 0.011810303, -0.024856567, -0.05130005, 0.0018510818, -0.028076172, 0.012214661, 0....
http://arxiv.org/abs/2310.03184v2
Retrieval-augmented Generation to Improve Math Question-Answering: Trade-offs Between Groundedness and Human Preference
Zachary Levonian, Chenglu Li, Wangda Zhu, Anoushka Gade, Owen Henkel, Millie-Ellen Postle, Wanli Xing
cs.CL, cs.HC
For middle-school math students, interactive question-answering (QA) with tutors is an effective way to learn. The flexibility and emergent capabilities of generative large language models (LLMs) has led to a surge of interest in automating portions of the tutoring process - including interactive QA to support conceptu...
http://arxiv.org/pdf/2310.03184v2
2023-11-11T01:26:22
2023-10-04T22:09:28
{ "title": [ 0.024673462, 0.0069847107, -0.02230835, -0.026443481, -0.0014486313, -0.03466797, 0.032287598, -0.0073776245, 0.012626648, 0.049743652, -0.021026611, 0.0045547485, 0.009185791, -0.0016231537, -0.06542969, -0.0013713837, -0.01108551, ...
http://arxiv.org/abs/2310.03182v1
Robust and Interpretable Medical Image Classifiers via Concept Bottleneck Models
An Yan, Yu Wang, Yiwu Zhong, Zexue He, Petros Karypis, Zihan Wang, Chengyu Dong, Amilcare Gentili, Chun-Nan Hsu, Jingbo Shang, Julian McAuley
cs.CV, cs.CL, cs.LG
Medical image classification is a critical problem for healthcare, with the potential to alleviate the workload of doctors and facilitate diagnoses of patients. However, two challenges arise when deploying deep learning models to real-world healthcare applications. First, neural models tend to learn spurious correlatio...
http://arxiv.org/pdf/2310.03182v1
2023-10-04T21:57:09
2023-10-04T21:57:09
{ "title": [ 0.008071899, 0.028152466, -0.03552246, -0.019683838, 0.03543091, -0.017471313, -0.008132935, -0.046691895, 0.022369385, 0.020996094, -0.026550293, 0.010253906, -0.036315918, -0.002696991, -0.06341553, -0.017028809, 0.0121536255, 0.03...
http://arxiv.org/abs/2310.03173v1
$\mathcal{B}$-Coder: Value-Based Deep Reinforcement Learning for Program Synthesis
Zishun Yu, Yunzhe Tao, Liyu Chen, Tao Sun, Hongxia Yang
cs.CL
Program synthesis aims to create accurate, executable code from natural language descriptions. This field has leveraged the power of reinforcement learning (RL) in conjunction with large language models (LLMs), significantly enhancing code generation capabilities. This integration focuses on directly optimizing functio...
http://arxiv.org/pdf/2310.03173v1
2023-10-04T21:40:36
2023-10-04T21:40:36
{ "title": [ -0.040039062, -0.016448975, -0.007171631, -0.027175903, -0.02015686, 0.0049819946, -0.04232788, -0.010719299, -0.012390137, 0.0031032562, -0.040405273, 0.036254883, -0.038360596, -0.019332886, 0.014152527, -0.022613525, -0.023666382, ...
http://arxiv.org/abs/2310.03128v3
MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use
Yue Huang, Jiawen Shi, Yuan Li, Chenrui Fan, Siyuan Wu, Qihui Zhang, Yixin Liu, Pan Zhou, Yao Wan, Neil Zhenqiang Gong, Lichao Sun
cs.SE, cs.CL
Large language models (LLMs) have garnered significant attention due to their impressive natural language processing (NLP) capabilities. Recently, many studies have focused on the tool utilization ability of LLMs. They primarily investigated how LLMs effectively collaborate with given specific tools. However, in scenar...
http://arxiv.org/pdf/2310.03128v3
2023-10-23T18:28:56
2023-10-04T19:39:26
{ "title": [ -0.054504395, -0.00920105, -0.035614014, -0.055236816, -0.019577026, 0.0061798096, -0.087402344, -0.011985779, -0.0056610107, 0.040374756, -0.007194519, 0.0059432983, 0.0082092285, -0.03274536, -0.03479004, 0.025619507, -0.0057754517, ...
http://arxiv.org/abs/2310.03094v2
Large Language Model Cascades with Mixture of Thoughts Representations for Cost-efficient Reasoning
Murong Yue, Jie Zhao, Min Zhang, Liang Du, Ziyu Yao
cs.CL, cs.AI, cs.LG
Large language models (LLMs) such as GPT-4 have exhibited remarkable performance in a variety of tasks, but this strong performance often comes with the high expense of using paid API services. In this paper, we are motivated to study building an LLM cascade to save the cost of using LLMs, particularly for performing r...
http://arxiv.org/pdf/2310.03094v2
2023-10-07T01:16:45
2023-10-04T18:21:17
{ "title": [ -0.019958496, -0.030548096, -0.032226562, -0.07867432, -0.029159546, 0.011749268, -0.064575195, 0.025512695, 0.007865906, 0.017929077, -0.031921387, 0.0076293945, 0.024383545, -0.01763916, -0.016921997, 0.0062294006, -0.004840851, 0....
http://arxiv.org/abs/2310.03084v1
Discovering Knowledge-Critical Subnetworks in Pretrained Language Models
Deniz Bayazit, Negar Foroutan, Zeming Chen, Gail Weiss, Antoine Bosselut
cs.CL, cs.AI, cs.LG
Pretrained language models (LMs) encode implicit representations of knowledge in their parameters. However, localizing these representations and disentangling them from each other remains an open problem. In this work, we investigate whether pretrained language models contain various knowledge-critical subnetworks: par...
http://arxiv.org/pdf/2310.03084v1
2023-10-04T18:02:01
2023-10-04T18:02:01
{ "title": [ -0.039123535, -0.017913818, -0.055877686, -0.038085938, -0.025970459, -0.0149002075, -0.0579834, 0.011375427, -0.016311646, 0.027694702, -0.010650635, 0.00023710728, 0.009536743, -0.024383545, -0.03543091, -0.010620117, -0.011756897, ...
http://arxiv.org/abs/2310.03026v2
LanguageMPC: Large Language Models as Decision Makers for Autonomous Driving
Hao Sha, Yao Mu, Yuxuan Jiang, Li Chen, Chenfeng Xu, Ping Luo, Shengbo Eben Li, Masayoshi Tomizuka, Wei Zhan, Mingyu Ding
cs.RO, cs.AI, cs.CL, cs.CV, cs.LG
Existing learning-based autonomous driving (AD) systems face challenges in comprehending high-level information, generalizing to rare events, and providing interpretability. To address these problems, this work employs Large Language Models (LLMs) as a decision-making component for complex AD scenarios that require hum...
http://arxiv.org/pdf/2310.03026v2
2023-10-13T16:13:43
2023-10-04T17:59:49
{ "title": [ -0.038238525, -0.01663208, -0.04159546, -0.04260254, 0.00060606, -0.013793945, -0.02029419, 0.012435913, -0.02192688, 0.026382446, -0.03704834, 0.017791748, -0.015792847, -0.0039043427, -0.017028809, -0.011398315, -0.016647339, 0.031...
http://arxiv.org/abs/2310.03025v1
Retrieval meets Long Context Large Language Models
Peng Xu, Wei Ping, Xianchao Wu, Lawrence McAfee, Chen Zhu, Zihan Liu, Sandeep Subramanian, Evelina Bakhturina, Mohammad Shoeybi, Bryan Catanzaro
cs.CL, cs.AI, cs.IR, cs.LG
Extending the context window of large language models (LLMs) is getting popular recently, while the solution of augmenting LLMs with retrieval has existed for years. The natural questions are: i) Retrieval-augmentation versus long context window, which one is better for downstream tasks? ii) Can both methods be combine...
http://arxiv.org/pdf/2310.03025v1
2023-10-04T17:59:41
2023-10-04T17:59:41
{ "title": [ -0.004814148, -0.0007328987, -0.03527832, -0.0680542, 0.009246826, -0.0034122467, -0.03338623, 0.005466461, 0.023223877, 0.04647827, -0.031204224, -0.011871338, 0.026504517, -0.010314941, -0.041503906, 0.019897461, -0.0014696121, 0.0...
http://arxiv.org/abs/2310.03018v1
Zero Resource Code-switched Speech Benchmark Using Speech Utterance Pairs For Multiple Spoken Languages
Kuan-Po Huang, Chih-Kai Yang, Yu-Kuan Fu, Ewan Dunbar, Hung-yi Lee
eess.AS, cs.CL, cs.SD
We introduce a new zero resource code-switched speech benchmark designed to directly assess the code-switching capabilities of self-supervised speech encoders. We showcase a baseline system of language modeling on discrete units to demonstrate how the code-switching abilities of speech encoders can be assessed in a zer...
http://arxiv.org/pdf/2310.03018v1
2023-10-04T17:58:11
2023-10-04T17:58:11
{ "title": [ -0.02268982, -0.011199951, -0.04095459, -0.04159546, -0.02281189, -0.013618469, -0.05871582, 0.01374054, -0.005466461, 0.045684814, 0.010871887, -0.009857178, 0.024291992, -0.043395996, -0.031234741, 0.016418457, -0.014884949, 0.0372...
http://arxiv.org/abs/2310.03017v1
Multimodal Question Answering for Unified Information Extraction
Yuxuan Sun, Kai Zhang, Yu Su
cs.CL
Multimodal information extraction (MIE) aims to extract structured information from unstructured multimedia content. Due to the diversity of tasks and settings, most current MIE models are task-specific and data-intensive, which limits their generalization to real-world scenarios with diverse task requirements and limi...
http://arxiv.org/pdf/2310.03017v1
2023-10-04T17:58:05
2023-10-04T17:58:05
{ "title": [ -0.032562256, 0.007598877, -0.040496826, -0.06439209, 0.0000022053719, 0.0041542053, -0.050628662, 0.0022201538, 0.014060974, 0.02861023, -0.017044067, 0.010093689, -0.003030777, -0.0062713623, -0.04849243, 0.006755829, -0.016967773, ...
http://arxiv.org/abs/2310.03016v1
Understanding In-Context Learning in Transformers and LLMs by Learning to Learn Discrete Functions
Satwik Bhattamishra, Arkil Patel, Phil Blunsom, Varun Kanade
cs.LG, cs.CL
In order to understand the in-context learning phenomenon, recent works have adopted a stylized experimental framework and demonstrated that Transformers can learn gradient-based learning algorithms for various classes of real-valued functions. However, the limitations of Transformers in implementing learning algorithm...
http://arxiv.org/pdf/2310.03016v1
2023-10-04T17:57:33
2023-10-04T17:57:33
{ "title": [ 0.023361206, 0.019134521, -0.04458618, -0.046295166, 0.009468079, -0.040374756, -0.0121154785, -0.0077934265, -0.028121948, 0.014099121, -0.07977295, 0.028686523, -0.016189575, -0.013061523, 0.005596161, -0.042388916, -0.0003118515, ...
http://arxiv.org/abs/2310.03003v1
From Words to Watts: Benchmarking the Energy Costs of Large Language Model Inference
Siddharth Samsi, Dan Zhao, Joseph McDonald, Baolin Li, Adam Michaleas, Michael Jones, William Bergeron, Jeremy Kepner, Devesh Tiwari, Vijay Gadepally
cs.CL, cs.DC
Large language models (LLMs) have exploded in popularity due to their new generative capabilities that go far beyond prior state-of-the-art. These technologies are increasingly being leveraged in various domains such as law, finance, and medicine. However, these models carry significant computational challenges, especi...
http://arxiv.org/pdf/2310.03003v1
2023-10-04T17:41:59
2023-10-04T17:41:59
{ "title": [ -0.039733887, -0.030334473, -0.028213501, -0.06756592, -0.018005371, 0.0049705505, -0.07208252, 0.009880066, -0.0074768066, 0.025512695, -0.021362305, 0.028259277, 0.0095825195, -0.021148682, -0.019485474, 0.015945435, 0.0005970001, ...
http://arxiv.org/abs/2310.02998v1
ECoFLaP: Efficient Coarse-to-Fine Layer-Wise Pruning for Vision-Language Models
Yi-Lin Sung, Jaehong Yoon, Mohit Bansal
cs.CV, cs.AI, cs.CL, cs.LG
Large Vision-Language Models (LVLMs) can understand the world comprehensively by integrating rich information from different modalities, achieving remarkable performance improvements on various multimodal downstream tasks. However, deploying LVLMs is often problematic due to their massive computational/energy costs and...
http://arxiv.org/pdf/2310.02998v1
2023-10-04T17:34:00
2023-10-04T17:34:00
{ "title": [ -0.02558899, -0.00969696, -0.031921387, -0.058013916, -0.0011339188, 0.009384155, -0.05508423, 0.013000488, 0.037841797, 0.016464233, -0.022949219, -0.0016460419, 0.020065308, -0.01914978, -0.040985107, 0.0019035339, 0.010032654, 0.0...
http://arxiv.org/abs/2310.02992v1
Kosmos-G: Generating Images in Context with Multimodal Large Language Models
Xichen Pan, Li Dong, Shaohan Huang, Zhiliang Peng, Wenhu Chen, Furu Wei
cs.CV, cs.CL
Recent advancements in text-to-image (T2I) and vision-language-to-image (VL2I) generation have made significant strides. However, the generation from generalized vision-language inputs, especially involving multiple images, remains under-explored. This paper presents Kosmos-G, a model that leverages the advanced percep...
http://arxiv.org/pdf/2310.02992v1
2023-10-04T17:28:44
2023-10-04T17:28:44
{ "title": [ -0.0126571655, -0.004386902, -0.03277588, -0.054504395, -0.0016860962, 0.0029468536, -0.049560547, -0.015197754, 0.030593872, 0.03515625, -0.013160706, 0.006958008, 0.012496948, -0.020233154, -0.028289795, 0.017364502, -0.0026931763, ...
http://arxiv.org/abs/2310.02989v1
xVal: A Continuous Number Encoding for Large Language Models
Siavash Golkar, Mariel Pettee, Michael Eickenberg, Alberto Bietti, Miles Cranmer, Geraud Krawezik, Francois Lanusse, Michael McCabe, Ruben Ohana, Liam Parker, Bruno Régaldo-Saint Blancard, Tiberiu Tesileanu, Kyunghyun Cho, Shirley Ho
stat.ML, cs.AI, cs.CL, cs.LG
Large Language Models have not yet been broadly adapted for the analysis of scientific datasets due in part to the unique difficulties of tokenizing numbers. We propose xVal, a numerical encoding scheme that represents any real number using just a single token. xVal represents a given real number by scaling a dedicated...
http://arxiv.org/pdf/2310.02989v1
2023-10-04T17:26:16
2023-10-04T17:26:16
{ "title": [ -0.04208374, -0.018356323, -0.054534912, -0.06750488, 0.0074157715, 0.0023059845, -0.05368042, 0.026275635, 0.0052719116, 0.02746582, -0.0099105835, -0.00026893616, 0.013160706, -0.018554688, -0.03427124, 0.015464783, -0.019561768, 0...
http://arxiv.org/abs/2310.02984v1
Scaling Laws for Associative Memories
Vivien Cabannes, Elvis Dohmatob, Alberto Bietti
stat.ML, cs.AI, cs.CL, cs.LG, cs.NE, I.2.6; G.1.6
Learning arguably involves the discovery and memorization of abstract rules. The aim of this paper is to study associative memory mechanisms. Our model is based on high-dimensional matrices consisting of outer products of embeddings, which relates to the inner layers of transformer language models. We derive precise sc...
http://arxiv.org/pdf/2310.02984v1
2023-10-04T17:20:34
2023-10-04T17:20:34
{ "title": [ -0.01890564, -0.004611969, -0.020523071, -0.030593872, 0.00046157837, -0.0015039444, 0.008605957, -0.0052871704, -0.010185242, 0.05935669, -0.024810791, -0.0036678314, 0.005622864, -0.0007724762, -0.060058594, 0.0040626526, 0.0011463165,...
http://arxiv.org/abs/2310.02980v1
Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors
Ido Amos, Jonathan Berant, Ankit Gupta
cs.LG, cs.CL
Modeling long-range dependencies across sequences is a longstanding goal in machine learning and has led to architectures, such as state space models, that dramatically outperform Transformers on long sequences. However, these impressive empirical gains have been by and large demonstrated on benchmarks (e.g. Long Range...
http://arxiv.org/pdf/2310.02980v1
2023-10-04T17:17:06
2023-10-04T17:17:06
{ "title": [ 0.013092041, -0.003742218, -0.031188965, -0.07171631, -0.014801025, -0.040008545, -0.03161621, -0.011138916, -0.03543091, -0.0026283264, -0.033477783, 0.012252808, -0.053894043, -0.039642334, -0.00059604645, -0.045776367, 0.0079956055, ...
http://arxiv.org/abs/2310.02977v1
T$^3$Bench: Benchmarking Current Progress in Text-to-3D Generation
Yuze He, Yushi Bai, Matthieu Lin, Wang Zhao, Yubin Hu, Jenny Sheng, Ran Yi, Juanzi Li, Yong-Jin Liu
cs.CV, cs.CL, cs.LG
Recent methods in text-to-3D leverage powerful pretrained diffusion models to optimize NeRF. Notably, these methods are able to produce high-quality 3D scenes without training on 3D data. Due to the open-ended nature of the task, most studies evaluate their results with subjective case studies and user experiments, the...
http://arxiv.org/pdf/2310.02977v1
2023-10-04T17:12:18
2023-10-04T17:12:18
{ "title": [ -0.043548584, 0.00034976006, -0.033721924, -0.029937744, -0.019638062, 0.00007325411, -0.050201416, -0.022872925, -0.0104522705, 0.055267334, -0.026672363, 0.0345459, 0.022399902, -0.017944336, -0.021942139, 0.0035133362, 0.013801575, ...
http://arxiv.org/abs/2310.02973v1
UniverSLU: Universal Spoken Language Understanding for Diverse Classification and Sequence Generation Tasks with a Single Network
Siddhant Arora, Hayato Futami, Jee-weon Jung, Yifan Peng, Roshan Sharma, Yosuke Kashiwagi, Emiru Tsunoo, Shinji Watanabe
cs.CL, cs.SD, eess.AS
Recent studies have demonstrated promising outcomes by employing large language models with multi-tasking capabilities. They utilize prompts to guide the model's behavior and surpass performance of task-specific models. Motivated by this, we ask: can we build a single model that jointly perform various spoken language ...
http://arxiv.org/pdf/2310.02973v1
2023-10-04T17:10:23
2023-10-04T17:10:23
{ "title": [ -0.030914307, -0.0096588135, -0.047851562, -0.073913574, -0.012207031, 0.002029419, -0.070495605, 0.026000977, -0.004760742, 0.020324707, -0.009902954, -0.004135132, 0.004436493, -0.024047852, -0.030471802, 0.00365448, -0.014175415, ...
http://arxiv.org/abs/2310.02971v3
Prompting and Adapter Tuning for Self-supervised Encoder-Decoder Speech Model
Kai-Wei Chang, Ming-Hsin Chen, Yun-Ping Lin, Jing Neng Hsu, Paul Kuo-Ming Huang, Chien-yu Huang, Shang-Wen Li, Hung-yi Lee
eess.AS, cs.CL, eess.SP
Prompting and adapter tuning have emerged as efficient alternatives to fine-tuning (FT) methods. However, existing studies on speech prompting focused on classification tasks and failed on more complex sequence generation tasks. Besides, adapter tuning is primarily applied with a focus on encoder-only self-supervised m...
http://arxiv.org/pdf/2310.02971v3
2023-11-14T21:15:01
2023-10-04T17:07:32
{ "title": [ -0.02104187, 0.012817383, -0.049316406, -0.049438477, -0.020706177, -0.012451172, -0.06463623, -0.00472641, -0.03729248, 0.041931152, -0.0040740967, -0.0035381317, 0.021896362, -0.033599854, -0.030349731, 0.0027980804, 0.011253357, 0...
http://arxiv.org/abs/2310.02954v4
DQ-LoRe: Dual Queries with Low Rank Approximation Re-ranking for In-Context Learning
Jing Xiong, Zixuan Li, Chuanyang Zheng, Zhijiang Guo, Yichun Yin, Enze Xie, Zhicheng Yang, Qingxing Cao, Haiming Wang, Xiongwei Han, Jing Tang, Chengming Li, Xiaodan Liang
cs.CL
Recent advances in natural language processing, primarily propelled by Large Language Models (LLMs), have showcased their remarkable capabilities grounded in in-context learning. A promising avenue for guiding LLMs in intricate reasoning tasks involves the utilization of intermediate reasoning steps within the Chain-of...
http://arxiv.org/pdf/2310.02954v4
2023-10-20T07:50:50
2023-10-04T16:44:37
{ "title": [ 0.010398865, 0.0074691772, -0.02519226, -0.04196167, 0.02355957, -0.011131287, 0.012107849, -0.035339355, -0.0002155304, 0.03778076, -0.052947998, 0.022613525, -0.002696991, 0.0011072159, -0.020431519, -0.004776001, -0.003446579, 0.0...
http://arxiv.org/abs/2310.02953v1
JsonTuning: Towards Generalizable, Robust, and Controllable Instruction Tuning
Chang Gao, Wenxuan Zhang, Guizhen Chen, Wai Lam
cs.CL
Instruction tuning has emerged as a crucial process for harnessing the capabilities of large language models (LLMs) by providing explicit task instructions, leading to improved performance in various tasks. However, prevalent text-to-text instruction tuning (TextTuning) methods suffer from limitations in generalization...
http://arxiv.org/pdf/2310.02953v1
2023-10-04T16:44:23
2023-10-04T16:44:23
{ "title": [ -0.045074463, -0.007789612, -0.04574585, -0.08996582, 0.016998291, 0.01184082, -0.051086426, 0.02520752, -0.01852417, 0.040924072, -0.015914917, 0.017990112, -0.018310547, 0.00031733513, -0.05050659, -0.0115737915, 0.00010603666, 0.0...
http://arxiv.org/abs/2310.02949v1
Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models
Xianjun Yang, Xiao Wang, Qi Zhang, Linda Petzold, William Yang Wang, Xun Zhao, Dahua Lin
cs.CL, cs.AI, cs.CR, cs.LG
Warning: This paper contains examples of harmful language, and reader discretion is recommended. The increasing open release of powerful large language models (LLMs) has facilitated the development of downstream applications by reducing the essential cost of data annotation and computation. To ensure AI safety, extensi...
http://arxiv.org/pdf/2310.02949v1
2023-10-04T16:39:31
2023-10-04T16:39:31
{ "title": [ -0.027282715, -0.022079468, -0.05026245, -0.0635376, -0.010910034, 0.003036499, -0.06451416, -0.0043029785, -0.0018453598, 0.03488159, 0.002960205, -0.009239197, 0.02444458, -0.032806396, -0.03060913, 0.0138168335, -0.006729126, 0.01...
http://arxiv.org/abs/2310.12321v1
A Survey of GPT-3 Family Large Language Models Including ChatGPT and GPT-4
Katikapalli Subramanyam Kalyan
cs.CL
Large language models (LLMs) are a special class of pretrained language models obtained by scaling model size, pretraining corpus and computation. LLMs, because of their large size and pretraining on large volumes of text data, exhibit special abilities which allow them to achieve remarkable performances without any ta...
http://arxiv.org/pdf/2310.12321v1
2023-10-04T16:37:05
2023-10-04T16:37:05
{ "title": [ -0.033569336, -0.0040130615, -0.04437256, -0.035705566, -0.0018148422, -0.009338379, -0.0569458, -0.0059928894, -0.006828308, 0.022369385, -0.00440979, -0.0061302185, 0.013504028, -0.04083252, -0.012931824, 0.020492554, -0.020812988, ...
http://arxiv.org/abs/2310.02943v1
LibriSpeech-PC: Benchmark for Evaluation of Punctuation and Capitalization Capabilities of end-to-end ASR Models
Aleksandr Meister, Matvei Novikov, Nikolay Karpov, Evelina Bakhturina, Vitaly Lavrukhin, Boris Ginsburg
cs.CL
Traditional automatic speech recognition (ASR) models output lower-cased words without punctuation marks, which reduces readability and necessitates a subsequent text processing model to convert ASR transcripts into a proper format. Simultaneously, the development of end-to-end ASR models capable of predicting punctuat...
http://arxiv.org/pdf/2310.02943v1
2023-10-04T16:23:37
2023-10-04T16:23:37
{ "title": [ -0.041992188, 0.005367279, -0.035858154, -0.03451538, -0.033081055, 0.0010461807, -0.08325195, -0.0002799034, -0.04119873, 0.038482666, 0.009056091, -0.025924683, 0.034423828, -0.05493164, -0.014640808, -0.013832092, -0.0043296814, 0...
http://arxiv.org/abs/2310.02932v1
Assessing Large Language Models on Climate Information
Jannis Bulian, Mike S. Schäfer, Afra Amini, Heidi Lam, Massimiliano Ciaramita, Ben Gaiarin, Michelle Chen Huebscher, Christian Buck, Niels Mede, Markus Leippold, Nadine Strauss
cs.CL, cs.AI, cs.CY, cs.LG
Understanding how climate change affects us and learning about available solutions are key steps toward empowering individuals and communities to mitigate and adapt to it. As Large Language Models (LLMs) rise in popularity, it is necessary to assess their capability in this domain. In this study, we present a comprehen...
http://arxiv.org/pdf/2310.02932v1
2023-10-04T16:09:48
2023-10-04T16:09:48
{ "title": [ -0.04421997, -0.014503479, -0.050628662, -0.05078125, -0.015609741, -0.0054626465, -0.034423828, 0.011230469, -0.009643555, 0.03930664, -0.014198303, -0.00944519, -0.02279663, -0.020324707, -0.04345703, 0.0107040405, 0.008178711, 0.0...
http://arxiv.org/abs/2310.02876v1
Hate Speech Detection in Limited Data Contexts using Synthetic Data Generation
Aman Khullar, Daniel Nkemelu, Cuong V. Nguyen, Michael L. Best
cs.CL, cs.CY
A growing body of work has focused on text classification methods for detecting the increasing amount of hate speech posted online. This progress has been limited to only a select number of highly-resourced languages causing detection systems to either under-perform or not exist in limited data contexts. This is majorl...
http://arxiv.org/pdf/2310.02876v1
2023-10-04T15:10:06
2023-10-04T15:10:06
{ "title": [ 0.0062942505, 0.0066184998, -0.049987793, -0.041259766, 0.014831543, -0.029693604, -0.0041046143, -0.0021686554, -0.011009216, 0.026123047, -0.03866577, 0.00018715858, -0.008178711, -0.015563965, -0.035064697, -0.025726318, -0.022094727,...
http://arxiv.org/abs/2310.02842v2
Sweeping Heterogeneity with Smart MoPs: Mixture of Prompts for LLM Task Adaptation
Chen Dun, Mirian Hipolito Garcia, Guoqing Zheng, Ahmed Hassan Awadallah, Anastasios Kyrillidis, Robert Sim
cs.CL, cs.AI
Large Language Models (LLMs) have the ability to solve a variety of tasks, such as text summarization and mathematical questions, just out of the box, but they are often trained with a single task in mind. Due to high computational costs, the current trend is to use prompt instruction tuning to better adjust monolithic...
http://arxiv.org/pdf/2310.02842v2
2023-10-05T21:13:11
2023-10-04T14:11:12
{ "title": [ -0.059020996, -0.020202637, -0.08190918, -0.04321289, -0.022369385, -0.021087646, -0.044555664, 0.040405273, 0.018295288, 0.031082153, -0.00844574, 0.0075569153, -0.00080633163, -0.03213501, -0.015914917, -0.0016679764, -0.0013799667, ...
http://arxiv.org/abs/2310.02832v1
Out-of-Distribution Detection by Leveraging Between-Layer Transformation Smoothness
Fran Jelenić, Josip Jukić, Martin Tutek, Mate Puljiz, Jan Šnajder
cs.LG, cs.CL
Effective OOD detection is crucial for reliable machine learning models, yet most current methods are limited in practical use due to requirements like access to training data or intervention in training. We present a novel method for detecting OOD data in deep neural networks based on transformation smoothness between...
http://arxiv.org/pdf/2310.02832v1
2023-10-04T13:59:45
2023-10-04T13:59:45
{ "title": [ 0.021255493, 0.03564453, -0.02532959, -0.011787415, 0.036071777, -0.012664795, 0.031982422, -0.03326416, 0.0037231445, 0.02671814, -0.0028495789, -0.008041382, 0.008224487, -0.037628174, -0.027679443, -0.022598267, -0.008888245, 0.02...
http://arxiv.org/abs/2310.04456v1
Multimodal Prompt Transformer with Hybrid Contrastive Learning for Emotion Recognition in Conversation
Shihao Zou, Xianying Huang, Xudong Shen
cs.CL, cs.SD, eess.AS
Emotion Recognition in Conversation (ERC) plays an important role in driving the development of human-machine interaction. Emotions can exist in multiple modalities, and multimodal ERC mainly faces two problems: (1) the noise problem in the cross-modal information fusion process, and (2) the prediction problem of less ...
http://arxiv.org/pdf/2310.04456v1
2023-10-04T13:54:46
2023-10-04T13:54:46
{ "title": [ 0.007537842, 0.0078086853, -0.03262329, -0.04647827, 0.0030326843, -0.017669678, -0.034332275, -0.0110321045, 0.025268555, 0.029907227, -0.02708435, -0.00472641, 0.018432617, -0.029205322, -0.044158936, -0.0077209473, -0.012283325, 0...
http://arxiv.org/abs/2310.02804v1
DOMINO: A Dual-System for Multi-step Visual Language Reasoning
Peifang Wang, Olga Golovneva, Armen Aghajanyan, Xiang Ren, Muhao Chen, Asli Celikyilmaz, Maryam Fazel-Zarandi
cs.CL, cs.CV, cs.LG
Visual language reasoning requires a system to extract text or numbers from information-dense images like charts or plots and perform logical or arithmetic reasoning to arrive at an answer. To tackle this task, existing work relies on either (1) an end-to-end vision-language model trained on a large amount of data, or ...
http://arxiv.org/pdf/2310.02804v1
2023-10-04T13:29:47
2023-10-04T13:29:47
{ "title": [ -0.022399902, -0.03326416, -0.027038574, -0.08312988, -0.02078247, 0.022628784, -0.06359863, 0.018081665, 0.028747559, 0.010025024, -0.037841797, 0.004383087, -0.012290955, -0.016860962, -0.03564453, -0.010650635, 0.011161804, 0.0085...
http://arxiv.org/abs/2310.02790v1
Low Resource Summarization using Pre-trained Language Models
Mubashir Munaf, Hammad Afzal, Naima Iltaf, Khawir Mahmood
cs.CL
With the advent of Deep Learning based Artificial Neural Networks models, Natural Language Processing (NLP) has witnessed significant improvements in textual data processing in terms of its efficiency and accuracy. However, the research is mostly restricted to high-resource languages such as English and low-resource la...
http://arxiv.org/pdf/2310.02790v1
2023-10-04T13:09:39
2023-10-04T13:09:39
{ "title": [ -0.004447937, -0.009994507, -0.030227661, -0.098083496, -0.011154175, 0.010284424, -0.086242676, 0.003326416, -0.003250122, 0.058624268, 0.0054397583, 0.004425049, 0.006385803, -0.0072517395, -0.058685303, 0.018615723, 0.004207611, 0...
http://arxiv.org/abs/2310.02778v2
Integrating UMLS Knowledge into Large Language Models for Medical Question Answering
Rui Yang, Edison Marrese-Taylor, Yuhe Ke, Lechao Cheng, Qingyu Chen, Irene Li
cs.CL, cs.AI
Large language models (LLMs) have demonstrated powerful text generation capabilities, bringing unprecedented innovation to the healthcare field. While LLMs hold immense promise for applications in healthcare, applying them to real clinical scenarios presents significant challenges, as these models may generate content ...
http://arxiv.org/pdf/2310.02778v2
2023-10-13T12:10:01
2023-10-04T12:50:26
{ "title": [ -0.0034809113, -0.020996094, -0.023834229, -0.089782715, 0.003786087, 0.020233154, -0.050811768, 0.010124207, -0.004535675, 0.04260254, 0.0022468567, -0.02456665, 0.025100708, -0.016784668, -0.050872803, 0.0031719208, 0.010154724, 0....
http://arxiv.org/abs/2310.02777v1
The Role of Linguistic Priors in Measuring Compositional Generalization of Vision-Language Models
Chenwei Wu, Li Erran Li, Stefano Ermon, Patrick Haffner, Rong Ge, Zaiwei Zhang
cs.CL
Compositionality is a common property in many modalities including natural languages and images, but the compositional generalization of multi-modal models is not well-understood. In this paper, we identify two sources of visual-linguistic compositionality: linguistic priors and the interplay between images and texts. ...
http://arxiv.org/pdf/2310.02777v1
2023-10-04T12:48:33
2023-10-04T12:48:33
{ "title": [ -0.030548096, -0.012992859, -0.01576233, -0.060302734, -0.035064697, 0.013893127, -0.07598877, -0.007320404, 0.013458252, 0.030090332, -0.018966675, -0.0013008118, -0.018997192, -0.017593384, -0.06085205, -0.014091492, 0.007637024, 0...
http://arxiv.org/abs/2310.02759v1
Comparative Study and Framework for Automated Summariser Evaluation: LangChain and Hybrid Algorithms
Bagiya Lakshmi S, Sanjjushri Varshini R, Rohith Mahadevan, Raja CSP Raman
cs.LG, cs.CL
Automated Essay Score (AES) is proven to be one of the cutting-edge technologies. Scoring techniques are used for various purposes. Reliable scores are calculated based on influential variables. Such variables can be computed by different methods based on the domain. The research is concentrated on the user's understan...
http://arxiv.org/pdf/2310.02759v1
2023-10-04T12:14:43
2023-10-04T12:14:43
{ "title": [ -0.026412964, 0.0051841736, -0.041107178, -0.06750488, -0.005329132, 0.0017185211, -0.056671143, -0.008598328, -0.025314331, 0.06347656, -0.01889038, 0.0385437, 0.005340576, 0.004058838, -0.057678223, 0.005241394, -0.019119263, 0.028...
http://arxiv.org/abs/2310.02754v2
LC-Score: Reference-less estimation of Text Comprehension Difficulty
Paul Tardy, Charlotte Roze, Paul Poupet
cs.CL
Being able to read and understand written text is critical in a digital era. However, studies shows that a large fraction of the population experiences comprehension issues. In this context, further initiatives in accessibility are required to improve the audience text comprehension. However, writers are hardly assiste...
http://arxiv.org/pdf/2310.02754v2
2023-10-05T14:28:20
2023-10-04T11:49:37
{ "title": [ -0.044830322, -0.022064209, -0.029525757, -0.010299683, -0.039215088, -0.005088806, -0.01499176, -0.021194458, 0.022750854, 0.017364502, -0.044799805, -0.0030269623, -0.012817383, -0.016113281, -0.05899048, -0.013961792, -0.02279663, ...
http://arxiv.org/abs/2310.04455v1
Inclusive Data Representation in Federated Learning: A Novel Approach Integrating Textual and Visual Prompt
Zihao Zhao, Zhenpeng Shi, Yang Liu, Wenbo Ding
cs.LG, cs.AI, cs.CL
Federated Learning (FL) is often impeded by communication overhead issues. Prompt tuning, as a potential solution, has been introduced to only adjust a few trainable parameters rather than the whole model. However, current single-modality prompt tuning approaches fail to comprehensively portray local clients' data. To ...
http://arxiv.org/pdf/2310.04455v1
2023-10-04T11:20:28
2023-10-04T11:20:28
{ "title": [ 0.001663208, 0.018844604, -0.07208252, -0.08123779, 0.046051025, -0.007320404, -0.021774292, 0.010429382, -0.007858276, 0.040100098, -0.06542969, 0.022018433, -0.017623901, 0.015960693, -0.057037354, -0.034057617, -0.013191223, 0.018...
http://arxiv.org/abs/2310.04454v1
Spherical Position Encoding for Transformers
Eren Unlu
cs.CL, cs.AI
Position encoding is the primary mechanism which induces notion of sequential order for input tokens in transformer architectures. Even though this formulation in the original transformer paper has yielded plausible performance for general purpose language understanding and generation, several new frameworks such as Ro...
http://arxiv.org/pdf/2310.04454v1
2023-10-04T09:28:59
2023-10-04T09:28:59
{ "title": [ 0.005268097, 0.007419586, -0.0030517578, -0.037750244, -0.017501831, 0.01625061, -0.03375244, -0.038360596, 0.0018348694, 0.072265625, -0.027832031, -0.0058021545, 0.064941406, -0.019866943, -0.023910522, 0.028640747, 0.03756714, -0....
http://arxiv.org/abs/2310.04453v1
COVID-19 South African Vaccine Hesitancy Models Show Boost in Performance Upon Fine-Tuning on M-pox Tweets
Nicholas Perikli, Srimoy Bhattacharya, Blessing Ogbuokiri, Zahra Movahedi Nia, Benjamin Lieberman, Nidhi Tripathi, Salah-Eddine Dahbi, Finn Stevenson, Nicola Bragazzi, Jude Kong, Bruce Mellado
cs.CL, cs.LG, cs.SI
Very large numbers of M-pox cases have, since the start of May 2022, been reported in non-endemic countries leading many to fear that the M-pox Outbreak would rapidly transition into another pandemic, while the COVID-19 pandemic ravages on. Given the similarities of M-pox with COVID-19, we chose to test the performance...
http://arxiv.org/pdf/2310.04453v1
2023-10-04T08:30:22
2023-10-04T08:30:22
{ "title": [ 0.0063591003, 0.0026626587, -0.008155823, -0.042663574, -0.0006327629, 0.0015659332, -0.0014820099, 0.017822266, -0.03778076, 0.05319214, -0.023635864, 0.047729492, 0.014282227, -0.015640259, -0.00036597252, -0.029174805, 0.0024471283, ...
http://arxiv.org/abs/2310.02655v1
AGIR: Automating Cyber Threat Intelligence Reporting with Natural Language Generation
Filippo Perrina, Francesco Marchiori, Mauro Conti, Nino Vincenzo Verde
cs.CR, cs.CL
Cyber Threat Intelligence (CTI) reporting is pivotal in contemporary risk management strategies. As the volume of CTI reports continues to surge, the demand for automated tools to streamline report generation becomes increasingly apparent. While Natural Language Processing techniques have shown potential in handling te...
http://arxiv.org/pdf/2310.02655v1
2023-10-04T08:25:37
2023-10-04T08:25:37
{ "title": [ -0.03955078, -0.017318726, -0.055664062, -0.040161133, -0.0030727386, 0.0048103333, -0.031707764, 0.0072135925, -0.029327393, 0.02758789, -0.033050537, 0.0124053955, -0.008888245, -0.011398315, -0.04360962, -0.020690918, -0.020019531, ...
http://arxiv.org/abs/2310.03051v1
How FaR Are Large Language Models From Agents with Theory-of-Mind?
Pei Zhou, Aman Madaan, Srividya Pranavi Potharaju, Aditya Gupta, Kevin R. McKee, Ari Holtzman, Jay Pujara, Xiang Ren, Swaroop Mishra, Aida Nematzadeh, Shyam Upadhyay, Manaal Faruqui
cs.CL, cs.AI
"Thinking is for Doing." Humans can infer other people's mental states from observations--an ability called Theory-of-Mind (ToM)--and subsequently act pragmatically on those inferences. Existing question answering benchmarks such as ToMi ask models questions to make inferences about beliefs of characters in a story, bu...
http://arxiv.org/pdf/2310.03051v1
2023-10-04T06:47:58
2023-10-04T06:47:58
{ "title": [ -0.0030593872, -0.02267456, 0.008323669, -0.0141067505, -0.051513672, -0.016693115, -0.008590698, -0.00856781, -0.00907135, 0.007904053, -0.028671265, -0.029067993, 0.01739502, -0.034332275, -0.010696411, -0.020004272, -0.004802704, ...
http://arxiv.org/abs/2310.02594v1
I$^2$KD-SLU: An Intra-Inter Knowledge Distillation Framework for Zero-Shot Cross-Lingual Spoken Language Understanding
Tianjun Mao, Chenghong Zhang
cs.CL
Spoken language understanding (SLU) typically includes two subtasks: intent detection and slot filling. Currently, it has achieved great success in high-resource languages, but it still remains challenging in low-resource languages due to the scarcity of labeled training data. Hence, there is a growing interest in zero...
http://arxiv.org/pdf/2310.02594v1
2023-10-04T05:45:23
2023-10-04T05:45:23
{ "title": [ -0.00031518936, -0.02558899, -0.04650879, -0.093322754, -0.021270752, 0.001042366, -0.088378906, 0.03253174, 0.030593872, 0.023880005, -0.051727295, -0.01725769, 0.013595581, -0.015808105, -0.038970947, -0.028884888, 0.0010061264, 0....
http://arxiv.org/abs/2310.02567v1
Improving Automatic VQA Evaluation Using Large Language Models
Oscar Mañas, Benno Krojer, Aishwarya Agrawal
cs.CV, cs.AI, cs.CL, cs.LG
8 years after the visual question answering (VQA) task was proposed, accuracy remains the primary metric for automatic evaluation. VQA Accuracy has been effective so far in the IID evaluation setting. However, our community is undergoing a shift towards open-ended generative models and OOD evaluation. In this new parad...
http://arxiv.org/pdf/2310.02567v1
2023-10-04T03:59:57
2023-10-04T03:59:57
{ "title": [ -0.025238037, 0.005191803, -0.048034668, -0.045013428, 0.00995636, -0.010192871, -0.039215088, -0.01838684, 0.00818634, 0.02418518, 0.0019388199, -0.036193848, -0.018814087, -0.036499023, -0.066223145, -0.010520935, -0.0309906, 0.049...
http://arxiv.org/abs/2310.02556v1
NOLA: Networks as Linear Combination of Low Rank Random Basis
Soroush Abbasi Koohpayegani, KL Navaneet, Parsa Nooralinejad, Soheil Kolouri, Hamed Pirsiavash
cs.CL, cs.CV
Large Language Models (LLMs) have recently gained popularity due to their impressive few-shot performance across various downstream tasks. However, fine-tuning all parameters and storing a unique model for each downstream task or domain becomes impractical because of the massive size of checkpoints (e.g., 350GB in GPT-...
http://arxiv.org/pdf/2310.02556v1
2023-10-04T03:30:24
2023-10-04T03:30:24
{ "title": [ -0.045318604, 0.0011453629, -0.050750732, -0.020751953, -0.00623703, -0.008255005, -0.015281677, -0.020553589, -0.053863525, 0.014816284, -0.020828247, 0.042022705, -0.040496826, -0.035247803, 0.020431519, -0.01600647, 0.008804321, 0...
http://arxiv.org/abs/2310.02527v1
CITING: Large Language Models Create Curriculum for Instruction Tuning
Tao Feng, Zifeng Wang, Jimeng Sun
cs.CL, cs.AI
The recent advancement of large language models (LLMs) has been achieved through a combo of instruction tuning and human alignment. However, building manually crafted instruction datasets and performing human alignment become the bottleneck for scaling the development of LLMs. In this paper, we exploit the idea of leve...
http://arxiv.org/pdf/2310.02527v1
2023-10-04T01:58:34
2023-10-04T01:58:34
{ "title": [ -0.01777649, -0.02407837, -0.024993896, -0.04058838, -0.053955078, -0.014808655, -0.06512451, 0.0006136894, -0.07287598, 0.010917664, -0.011260986, 0.0017728806, -0.019317627, -0.04611206, 0.056854248, -0.049621582, -0.0061683655, 0....
http://arxiv.org/abs/2310.02489v1
ResidualTransformer: Residual Low-rank Learning with Weight-sharing for Transformer Layers
Yiming Wang, Jinyu Li
cs.CL, cs.LG, cs.SD, eess.AS
Memory constraint of always-on devices is one of the major concerns when deploying speech processing models on these devices. While larger models trained with sufficiently large amount of data generally perform better, making them fit in the device memory is a demanding challenge. In this paper, we aim to reduce model ...
http://arxiv.org/pdf/2310.02489v1
2023-10-03T23:31:48
2023-10-03T23:31:48
{ "title": [ -0.010810852, -0.0006580353, -0.0435791, -0.04626465, 0.014671326, -0.0047302246, -0.019424438, -0.016433716, -0.008621216, 0.021957397, -0.007751465, 0.015182495, 0.014587402, -0.038146973, -0.004047394, 0.02015686, 0.016921997, 0.0...
http://arxiv.org/abs/2310.02469v1
Large Language Models Can Be Good Privacy Protection Learners
Yijia Xiao, Yiqiao Jin, Yushi Bai, Yue Wu, Xianjun Yang, Xiao Luo, Wenchao Yu, Xujiang Zhao, Yanchi Liu, Haifeng Chen, Wei Wang, Wei Cheng
cs.CL, cs.AI, cs.LG
The proliferation of Large Language Models (LLMs) has driven considerable interest in fine-tuning them with domain-specific data to create specialized language models. Nevertheless, such domain-specific fine-tuning data often contains sensitive personally identifiable information (PII). Direct fine-tuning LLMs on this ...
http://arxiv.org/pdf/2310.02469v1
2023-10-03T22:37:01
2023-10-03T22:37:01
{ "title": [ 0.010688782, -0.007965088, -0.04019165, -0.07220459, -0.03048706, -0.03164673, -0.037902832, 0.031921387, -0.045410156, 0.02545166, -0.022583008, -0.022979736, 0.008354187, -0.050964355, 0.026870728, -0.025665283, -0.0042419434, 0.04...
http://arxiv.org/abs/2310.04452v1
Short text classification with machine learning in the social sciences: The case of climate change on Twitter
Karina Shyrokykh, Maksym Girnyk, Lisa Dellmuth
cs.CL, cs.LG, stat.ME
To analyse large numbers of texts, social science researchers are increasingly confronting the challenge of text classification. When manual labeling is not possible and researchers have to find automatized ways to classify texts, computer science provides a useful toolbox of machine-learning methods whose performance ...
http://arxiv.org/pdf/2310.04452v1
2023-10-03T22:09:43
2023-10-03T22:09:43
{ "title": [ -0.009780884, 0.013595581, -0.025772095, -0.024658203, 0.009208679, -0.011688232, 0.004322052, -0.022491455, -0.03427124, 0.026184082, -0.036071777, 0.036102295, -0.033325195, -0.004459381, -0.04272461, -0.02960205, -0.019805908, 0.0...
http://arxiv.org/abs/2310.02457v2
The Empty Signifier Problem: Towards Clearer Paradigms for Operationalising "Alignment" in Large Language Models
Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, Scott A. Hale
cs.CL, cs.CY
In this paper, we address the concept of "alignment" in large language models (LLMs) through the lens of post-structuralist socio-political theory, specifically examining its parallels to empty signifiers. To establish a shared vocabulary around how abstract concepts of alignment are operationalised in empirical datase...
http://arxiv.org/pdf/2310.02457v2
2023-11-15T18:02:03
2023-10-03T22:02:17
{ "title": [ -0.04248047, 0.0014333725, -0.041625977, -0.05783081, -0.0031795502, 0.022857666, -0.0692749, -0.010910034, 0.003917694, 0.04269409, -0.008163452, -0.005458832, 0.011299133, -0.01828003, -0.061157227, 0.0006785393, -0.02166748, 0.027...
http://arxiv.org/abs/2310.02451v1
Backdoor Adjustment of Confounding by Provenance for Robust Text Classification of Multi-institutional Clinical Notes
Xiruo Ding, Zhecheng Sheng, Meliha Yetişgen, Serguei Pakhomov, Trevor Cohen
cs.CL
Natural Language Processing (NLP) methods have been broadly applied to clinical tasks. Machine learning and deep learning approaches have been used to improve the performance of clinical NLP. However, these approaches require sufficiently large datasets for training, and trained models have been shown to transfer poorl...
http://arxiv.org/pdf/2310.02451v1
2023-10-03T21:40:44
2023-10-03T21:40:44
{ "title": [ -0.014328003, -0.015052795, -0.015808105, -0.04284668, -0.02444458, 0.0068359375, -0.052459717, -0.012962341, -0.0009417534, 0.028808594, -0.03250122, 0.053527832, -0.021575928, -0.007659912, -0.038848877, -0.019378662, -0.013916016, ...
http://arxiv.org/abs/2310.02446v1
Low-Resource Languages Jailbreak GPT-4
Zheng-Xin Yong, Cristina Menghini, Stephen H. Bach
cs.CL, cs.AI, cs.CR, cs.LG
AI safety training and red-teaming of large language models (LLMs) are measures to mitigate the generation of unsafe content. Our work exposes the inherent cross-lingual vulnerability of these safety mechanisms, resulting from the linguistic inequality of safety training data, by successfully circumventing GPT-4's safe...
http://arxiv.org/pdf/2310.02446v1
2023-10-03T21:30:56
2023-10-03T21:30:56
{ "title": [ -0.015129089, -0.012687683, -0.049438477, -0.02003479, -0.03100586, -0.02017212, -0.018920898, 0.031341553, -0.036499023, 0.04019165, 0.017684937, -0.0072746277, 0.011245728, -0.05682373, 0.012832642, -0.013946533, 0.005455017, 0.005...
http://arxiv.org/abs/2310.02439v1
Novice Learner and Expert Tutor: Evaluating Math Reasoning Abilities of Large Language Models with Misconceptions
Naiming Liu, Shashank Sonkar, Zichao Wang, Simon Woodhead, Richard G. Baraniuk
cs.CL
We propose novel evaluations for mathematical reasoning capabilities of Large Language Models (LLMs) based on mathematical misconceptions. Our primary approach is to simulate LLMs as a novice learner and an expert tutor, aiming to identify the incorrect answer to math question resulted from a specific misconception and...
http://arxiv.org/pdf/2310.02439v1
2023-10-03T21:19:50
2023-10-03T21:19:50
{ "title": [ -0.0069007874, -0.008232117, -0.016723633, -0.02279663, -0.019897461, -0.017318726, -0.015625, -0.012260437, -0.0005354881, 0.009384155, -0.0317688, -0.030303955, 0.0037174225, -0.033111572, -0.018997192, -0.043182373, -0.019943237, ...
http://arxiv.org/abs/2310.02431v1
Can Large Language Models Provide Security & Privacy Advice? Measuring the Ability of LLMs to Refute Misconceptions
Yufan Chen, Arjun Arunasalam, Z. Berkay Celik
cs.HC, cs.CL
Users seek security & privacy (S&P) advice from online resources, including trusted websites and content-sharing platforms. These resources help users understand S&P technologies and tools and suggest actionable strategies. Large Language Models (LLMs) have recently emerged as trusted information sources. However, thei...
http://arxiv.org/pdf/2310.02431v1
2023-10-03T20:54:29
2023-10-03T20:54:29
{ "title": [ -0.037231445, -0.01461792, -0.052612305, -0.007381439, -0.041015625, -0.025848389, -0.03555298, 0.001376152, -0.031402588, 0.00046038628, -0.04421997, -0.0013914108, 0.0048713684, -0.045410156, 0.02482605, -0.048034668, -0.0030784607, ...
http://arxiv.org/abs/2310.02421v1
Can a student Large Language Model perform as well as it's teacher?
Sia Gholami, Marwan Omar
cs.LG, cs.AI, cs.CL
The burgeoning complexity of contemporary deep learning models, while achieving unparalleled accuracy, has inadvertently introduced deployment challenges in resource-constrained environments. Knowledge distillation, a technique aiming to transfer knowledge from a high-capacity "teacher" model to a streamlined "student"...
http://arxiv.org/pdf/2310.02421v1
2023-10-03T20:34:59
2023-10-03T20:34:59
{ "title": [ -0.000259161, 0.007820129, -0.02116394, -0.018432617, -0.02645874, -0.009941101, -0.014793396, -0.009506226, -0.020996094, 0.027069092, -0.030441284, -0.028793335, -0.019058228, -0.01876831, -0.017120361, -0.04711914, 0.0029640198, 0...
http://arxiv.org/abs/2310.02410v1
Mixture of Quantized Experts (MoQE): Complementary Effect of Low-bit Quantization and Robustness
Young Jin Kim, Raffy Fahim, Hany Hassan Awadalla
cs.LG, cs.CL
Large Mixture of Experts (MoE) models could achieve state-of-the-art quality on various language tasks, including machine translation task, thanks to the efficient model scaling capability with expert parallelism. However, it has brought a fundamental issue of larger memory consumption and increased memory bandwidth bo...
http://arxiv.org/pdf/2310.02410v1
2023-10-03T20:11:23
2023-10-03T20:11:23
{ "title": [ -0.021087646, 0.017578125, -0.06323242, -0.020141602, 0.006175995, -0.048675537, 0.00881958, -0.019210815, -0.006668091, 0.091918945, 0.02557373, 0.014732361, 0.05984497, -0.028625488, -0.037750244, 0.044830322, 0.012168884, 0.016250...
http://arxiv.org/abs/2310.02409v1
Nugget 2D: Dynamic Contextual Compression for Scaling Decoder-only Language Models
Guanghui Qin, Corby Rosset, Ethan C. Chau, Nikhil Rao, Benjamin Van Durme
cs.CL, cs.AI, cs.LG, I.2.7; I.2.6
Standard Transformer-based language models (LMs) scale poorly to long contexts. We propose a solution based on dynamic contextual compression, which extends the Nugget approach of Qin & Van Durme (2023) from BERT-like frameworks to decoder-only LMs. Our method models history as compressed "nuggets" which are trained to...
http://arxiv.org/pdf/2310.02409v1
2023-10-03T20:07:06
2023-10-03T20:07:06
{ "title": [ -0.046020508, -0.014411926, -0.048431396, -0.07281494, -0.011009216, 0.014961243, -0.07824707, 0.021072388, -0.0020256042, 0.055114746, 0.003900528, -0.0121154785, 0.035705566, -0.027297974, -0.044799805, 0.02784729, 0.005039215, 0.0...
http://arxiv.org/abs/2310.02408v1
MindTheDApp: A Toolchain for Complex Network-Driven Structural Analysis of Ethereum-based Decentralised Applications
Giacomo Ibba, Sabrina Aufiero, Silvia Bartolucci, Rumyana Neykova, Marco Ortu, Roberto Tonelli, Giuseppe Destefanis
cs.IT, cs.CL, math.IT
This paper presents MindTheDApp, a toolchain designed specifically for the structural analysis of Ethereum-based Decentralized Applications (DApps), with a distinct focus on a complex network-driven approach. Unlike existing tools, our toolchain combines the power of ANTLR4 and Abstract Syntax Tree (AST) traversal tech...
http://arxiv.org/pdf/2310.02408v1
2023-10-03T20:03:08
2023-10-03T20:03:08
{ "title": [ -0.03540039, -0.0068244934, -0.043670654, -0.021240234, -0.0073623657, 0.00042796135, -0.038208008, -0.008621216, -0.0038547516, 0.04937744, -0.012290955, 0.010856628, 0.031921387, -0.014350891, -0.033447266, 0.012680054, -0.0034713745, ...
http://arxiv.org/abs/2310.04451v1
AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models
Xiaogeng Liu, Nan Xu, Muhao Chen, Chaowei Xiao
cs.CL, cs.AI
The aligned Large Language Models (LLMs) are powerful language understanding and decision-making tools that are created through extensive alignment with human feedback. However, these large models remain susceptible to jailbreak attacks, where adversaries manipulate prompts to elicit malicious outputs that should not b...
http://arxiv.org/pdf/2310.04451v1
2023-10-03T19:44:37
2023-10-03T19:44:37
{ "title": [ -0.03677368, -0.011787415, -0.04949951, -0.040222168, -0.015060425, -0.0054092407, -0.05770874, -0.0031394958, -0.03137207, 0.0184021, -0.004547119, 0.015426636, 0.0039138794, -0.036834717, -0.0021820068, -0.00097608566, -0.014022827, ...
http://arxiv.org/abs/2310.02382v1
Unsupervised Speech Recognition with N-Skipgram and Positional Unigram Matching
Liming Wang, Mark Hasegawa-Johnson, Chang D. Yoo
cs.CL, cs.SD, eess.AS
Training unsupervised speech recognition systems presents challenges due to GAN-associated instability, misalignment between speech and text, and significant memory demands. To tackle these challenges, we introduce a novel ASR system, ESPUM. This system harnesses the power of lower-order N-skipgrams (up to N=3) combine...
http://arxiv.org/pdf/2310.02382v1
2023-10-03T19:05:32
2023-10-03T19:05:32
{ "title": [ -0.009414673, 0.027862549, -0.046051025, -0.08428955, 0.037902832, 0.01751709, -0.06085205, 0.00031995773, -0.011100769, 0.04321289, -0.041778564, -0.013008118, -0.0005874634, -0.0031719208, -0.055725098, -0.015022278, -0.011459351, ...
http://arxiv.org/abs/2310.02374v2
Conversational Health Agents: A Personalized LLM-Powered Agent Framework
Mahyar Abbasian, Iman Azimi, Amir M. Rahmani, Ramesh Jain
cs.CL
Conversational Health Agents (CHAs) are interactive systems designed to enhance personal healthcare services by engaging in empathetic conversations and processing multimodal data. While current CHAs, especially those utilizing Large Language Models (LLMs), primarily focus on conversation, they often need more comprehe...
http://arxiv.org/pdf/2310.02374v2
2023-10-21T01:23:26
2023-10-03T18:54:10
{ "title": [ -0.023925781, -0.007232666, -0.04824829, -0.07763672, -0.0011291504, 0.01133728, -0.07086182, 0.014289856, -0.013870239, 0.026565552, -0.021133423, 0.019012451, -0.013343811, -0.026763916, -0.015609741, -0.0045433044, -0.015151978, 0...
http://arxiv.org/abs/2310.02372v1
ProtoNER: Few shot Incremental Learning for Named Entity Recognition using Prototypical Networks
Ritesh Kumar, Saurabh Goyal, Ashish Verma, Vatche Isahagian
cs.CL, cs.AI
Key value pair (KVP) extraction or Named Entity Recognition(NER) from visually rich documents has been an active area of research in document understanding and data extraction domain. Several transformer based models such as LayoutLMv2, LayoutLMv3, and LiLT have emerged achieving state of the art results. However, addi...
http://arxiv.org/pdf/2310.02372v1
2023-10-03T18:52:19
2023-10-03T18:52:19
{ "title": [ -0.004611969, -0.010070801, -0.04714966, -0.06750488, 0.006164551, -0.010032654, -0.036712646, 0.010292053, 0.028198242, 0.006855011, -0.04827881, 0.023742676, -0.010032654, -0.011184692, -0.02407837, 0.0035743713, 0.0029945374, 0.03...
http://arxiv.org/abs/2310.02357v3
On the definition of toxicity in NLP
Sergey Berezin, Reza Farahbakhsh, Noel Crespi
cs.CL, cs.AI, cs.LG
The fundamental problem in toxicity detection task lies in the fact that the toxicity is ill-defined. This causes us to rely on subjective and vague data in models' training, which results in non-robust and non-accurate results: garbage in - garbage out. This work suggests a new, stress-level-based definition of toxi...
http://arxiv.org/pdf/2310.02357v3
2023-10-19T18:38:09
2023-10-03T18:32:34
{ "title": [ 0.029815674, -0.006801605, 0.010002136, -0.058898926, -0.023498535, -0.0054779053, -0.032440186, -0.019897461, -0.03857422, 0.026885986, -0.04876709, 0.0069503784, 0.009994507, -0.015731812, 0.003753662, -0.030044556, -0.013549805, 0...
http://arxiv.org/abs/2310.02263v1
Contrastive Post-training Large Language Models on Data Curriculum
Canwen Xu, Corby Rosset, Luciano Del Corro, Shweti Mahajan, Julian McAuley, Jennifer Neville, Ahmed Hassan Awadallah, Nikhil Rao
cs.CL, cs.AI, cs.LG
Alignment serves as an important step to steer large language models (LLMs) towards human preferences. In this paper, we explore contrastive post-training techniques for alignment by automatically constructing preference pairs from multiple models of varying strengths (e.g., InstructGPT, ChatGPT and GPT-4). We carefull...
http://arxiv.org/pdf/2310.02263v1
2023-10-03T17:59:46
2023-10-03T17:59:46
{ "title": [ -0.006793976, -0.009185791, -0.02659607, -0.07147217, -0.013023376, -0.008979797, -0.037231445, 0.0044174194, 0.016998291, 0.016571045, -0.022537231, 0.004146576, -0.015701294, -0.018447876, -0.009597778, -0.0061073303, -0.019577026, ...
http://arxiv.org/abs/2310.02264v1
Generalizable Long-Horizon Manipulations with Large Language Models
Haoyu Zhou, Mingyu Ding, Weikun Peng, Masayoshi Tomizuka, Lin Shao, Chuang Gan
cs.RO, cs.CL, cs.CV, cs.LG
This work introduces a framework harnessing the capabilities of Large Language Models (LLMs) to generate primitive task conditions for generalizable long-horizon manipulations with novel objects and unseen tasks. These task conditions serve as guides for the generation and adjustment of Dynamic Movement Primitives (DMP...
http://arxiv.org/pdf/2310.02264v1
2023-10-03T17:59:46
2023-10-03T17:59:46
{ "title": [ -0.008422852, 0.00019466877, -0.038848877, -0.08862305, 0.005168915, 0.0075569153, -0.0569458, 0.01158905, 0.0050811768, 0.03491211, -0.019729614, -0.00014853477, 0.0007529259, -0.0026855469, -0.037506104, 0.001578331, -0.016784668, ...
http://arxiv.org/abs/2310.02304v1
Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation
Eric Zelikman, Eliana Lorch, Lester Mackey, Adam Tauman Kalai
cs.CL, cs.AI, cs.LG, stat.ML
Several recent advances in AI systems (e.g., Tree-of-Thoughts and Program-Aided Language Models) solve problems by providing a "scaffolding" program that structures multiple calls to language models to generate better outputs. A scaffolding program is written in a programming language such as Python. In this work, we u...
http://arxiv.org/pdf/2310.02304v1
2023-10-03T17:59:32
2023-10-03T17:59:32
{ "title": [ -0.023254395, -0.0061454773, -0.010978699, -0.0496521, -0.028259277, 0.0046463013, -0.048339844, -0.0009303093, -0.02078247, 0.010147095, 0.0063438416, -0.005455017, -0.0071983337, -0.049438477, -0.0009160042, 0.008110046, 0.025650024, ...
http://arxiv.org/abs/2310.02255v2
MathVista: Evaluating Math Reasoning in Visual Contexts with GPT-4V, Bard, and Other Large Multimodal Models
Pan Lu, Hritik Bansal, Tony Xia, Jiacheng Liu, Chunyuan Li, Hannaneh Hajishirzi, Hao Cheng, Kai-Wei Chang, Michel Galley, Jianfeng Gao
cs.CV, cs.AI, cs.CL, cs.LG
Large Language Models (LLMs) and Large Multimodal Models (LMMs) exhibit impressive problem-solving skills in many tasks and domains, but their ability in mathematical reasoning in visual contexts has not been systematically studied. To bridge this gap, we present MathVista, a benchmark designed to combine challenges fr...
http://arxiv.org/pdf/2310.02255v2
2023-10-25T20:22:24
2023-10-03T17:57:24
{ "title": [ 0.01423645, 0.0069122314, -0.01701355, -0.018447876, 0.002746582, -0.025283813, -0.0010461807, -0.028137207, 0.023757935, 0.0028381348, -0.027496338, -0.0060043335, -0.0005578995, -0.03744507, -0.024353027, -0.007457733, -0.01612854, ...
http://arxiv.org/abs/2310.02249v1
Harnessing Pre-Trained Sentence Transformers for Offensive Language Detection in Indian Languages
Ananya Joshi, Raviraj Joshi
cs.CL, cs.LG
In our increasingly interconnected digital world, social media platforms have emerged as powerful channels for the dissemination of hate speech and offensive content. This work delves into the domain of hate speech detection, placing specific emphasis on three low-resource Indian languages: Bengali, Assamese, and Gujar...
http://arxiv.org/pdf/2310.02249v1
2023-10-03T17:53:09
2023-10-03T17:53:09
{ "title": [ 0.012039185, 0.006996155, -0.02720642, -0.05630493, 0.012504578, -0.013191223, -0.029342651, -0.0042381287, -0.0023670197, 0.018173218, -0.03555298, 0.0019569397, 0.015655518, -0.010810852, -0.04107666, -0.015037537, -0.016311646, 0....
http://arxiv.org/abs/2310.02238v2
Who's Harry Potter? Approximate Unlearning in LLMs
Ronen Eldan, Mark Russinovich
cs.CL, cs.AI
Large language models (LLMs) are trained on massive internet corpora that often contain copyrighted content. This poses legal and ethical challenges for the developers and users of these models, as well as the original authors and publishers. In this paper, we propose a novel technique for unlearning a subset of the tr...
http://arxiv.org/pdf/2310.02238v2
2023-10-04T05:20:19
2023-10-03T17:48:14
{ "title": [ -0.017166138, -0.020370483, -0.066345215, 0.017333984, -0.048431396, -0.03024292, -0.024276733, -0.002313614, -0.048309326, 0.042999268, -0.0017461777, 0.047943115, 0.0034065247, -0.036071777, 0.03366089, -0.02142334, 0.0093307495, 0...
http://arxiv.org/abs/2310.02235v1
Automatic Quality Assessment of Wikipedia Articles -- A Systematic Literature Review
Pedro Miguel Moás, Carla Teixeira Lopes
cs.CL, cs.AI, cs.CY, cs.LG, cs.SI
Wikipedia is the world's largest online encyclopedia, but maintaining article quality through collaboration is challenging. Wikipedia designed a quality scale, but with such a manual assessment process, many articles remain unassessed. We review existing methods for automatically measuring the quality of Wikipedia arti...
http://arxiv.org/pdf/2310.02235v1
2023-10-03T17:45:39
2023-10-03T17:45:39
{ "title": [ -0.011245728, 0.020370483, -0.03012085, -0.031799316, 0.015716553, -0.00073575974, -0.019927979, -0.048095703, -0.020004272, 0.043823242, -0.018966675, 0.027236938, -0.008270264, -0.0015649796, -0.048736572, 0.0012359619, -0.018173218, ...
http://arxiv.org/abs/2310.02229v2
Extraction of Medication and Temporal Relation from Clinical Text using Neural Language Models
Hangyu Tu, Lifeng Han, Goran Nenadic
cs.CL, cs.AI
Clinical texts, represented in electronic medical records (EMRs), contain rich medical information and are essential for disease prediction, personalised information recommendation, clinical decision support, and medication pattern mining and measurement. Relation extractions between medication mentions and temporal in...
http://arxiv.org/pdf/2310.02229v2
2023-10-08T21:17:54
2023-10-03T17:37:22
{ "title": [ -0.012680054, -0.0025539398, -0.0513916, -0.08526611, 0.0007362366, 0.0031166077, -0.070617676, 0.009147644, 0.0028476715, 0.034210205, -0.001739502, 0.00022745132, -0.010131836, -0.01789856, -0.05596924, 0.0037631989, -0.01210022, 0...
http://arxiv.org/abs/2310.02226v1
Think before you speak: Training Language Models With Pause Tokens
Sachin Goyal, Ziwei Ji, Ankit Singh Rawat, Aditya Krishna Menon, Sanjiv Kumar, Vaishnavh Nagarajan
cs.CL, cs.AI, cs.LG
Language models generate responses by producing a series of tokens in immediate succession: the $(K+1)^{th}$ token is an outcome of manipulating $K$ hidden vectors per layer, one vector per preceding token. What if instead we were to let the model manipulate say, $K+10$ hidden vectors, before it outputs the $(K+1)^{th}...
http://arxiv.org/pdf/2310.02226v1
2023-10-03T17:32:41
2023-10-03T17:32:41
{ "title": [ -0.029083252, -0.013999939, -0.032073975, -0.045562744, -0.015792847, -0.0016202927, -0.051696777, 0.006034851, 0.0015277863, 0.0003156662, -0.042877197, 0.01309967, -0.011497498, -0.01007843, -0.03390503, -0.0044670105, -0.014022827, ...
http://arxiv.org/abs/2310.02224v1
Can Language Models be Instructed to Protect Personal Information?
Yang Chen, Ethan Mendes, Sauvik Das, Wei Xu, Alan Ritter
cs.CL
Large multimodal language models have proven transformative in numerous applications. However, these models have been shown to memorize and leak pre-training data, raising serious user privacy and information security concerns. While data leaks should be prevented, it is also crucial to examine the trade-off between th...
http://arxiv.org/pdf/2310.02224v1
2023-10-03T17:30:33
2023-10-03T17:30:33
{ "title": [ -0.00409317, -0.01436615, -0.0009226799, -0.047332764, -0.03100586, 0.02281189, -0.05230713, -0.0009393692, -0.04788208, 0.02961731, 0.006072998, 0.00157547, 0.0038814545, -0.046783447, 0.025741577, -0.017578125, -0.0072669983, 0.054...
http://arxiv.org/abs/2310.02207v1
Language Models Represent Space and Time
Wes Gurnee, Max Tegmark
cs.LG, cs.AI, cs.CL
The capabilities of large language models (LLMs) have sparked debate over whether such systems just learn an enormous collection of superficial statistics or a coherent model of the data generating process -- a world model. We find evidence for the latter by analyzing the learned representations of three spatial datase...
http://arxiv.org/pdf/2310.02207v1
2023-10-03T17:06:52
2023-10-03T17:06:52
{ "title": [ 0.022460938, -0.022842407, -0.018005371, -0.058563232, -0.051971436, -0.024734497, -0.052612305, 0.018341064, 0.007850647, 0.06536865, -0.02279663, -0.052947998, 0.05505371, -0.011161804, -0.024856567, -0.008026123, 0.035339355, -0.0...
http://arxiv.org/abs/2310.18322v1
What's Next in Affective Modeling? Large Language Models
Nutchanon Yongsatianchot, Tobias Thejll-Madsen, Stacy Marsella
cs.CL, cs.AI
Large Language Models (LLM) have recently been shown to perform well at various tasks from language understanding, reasoning, storytelling, and information search to theory of mind. In an extension of this work, we explore the ability of GPT-4 to solve tasks related to emotion prediction. GPT-4 performs well across mul...
http://arxiv.org/pdf/2310.18322v1
2023-10-03T16:39:20
2023-10-03T16:39:20
{ "title": [ -0.018249512, -0.0071792603, -0.031799316, -0.052124023, -0.0066070557, -0.0076904297, -0.033966064, -0.009170532, 0.02796936, 0.024215698, -0.034576416, -0.01473999, 0.021347046, -0.0289917, -0.040527344, 0.01109314, -0.0048713684, ...
http://arxiv.org/abs/2310.04450v1
Investigating Large Language Models' Perception of Emotion Using Appraisal Theory
Nutchanon Yongsatianchot, Parisa Ghanad Torshizi, Stacy Marsella
cs.CL, cs.AI
Large Language Models (LLM) like ChatGPT have significantly advanced in recent years and are now being used by the general public. As more people interact with these systems, improving our understanding of these black box models is crucial, especially regarding their understanding of human psychological aspects. In thi...
http://arxiv.org/pdf/2310.04450v1
2023-10-03T16:34:47
2023-10-03T16:34:47
{ "title": [ -0.021514893, -0.0034637451, -0.031341553, -0.037384033, -0.014503479, -0.009284973, -0.029281616, -0.0017051697, 0.034210205, 0.029937744, -0.025436401, -0.027893066, 0.01763916, -0.025009155, -0.06726074, 0.0024871826, -0.018447876, ...
http://arxiv.org/abs/2310.02174v1
Ask Again, Then Fail: Large Language Models' Vacillations in Judgement
Qiming Xie, Zengzhi Wang, Yi Feng, Rui Xia
cs.CL, cs.AI, cs.LG
With the emergence of generative conversational large language models (LLMs) like ChatGPT, serving as virtual assistants in various fields, the stability and reliability of their responses have become crucial. However, during usage, it has been observed that these models tend to waver in their judgements when confronte...
http://arxiv.org/pdf/2310.02174v1
2023-10-03T16:08:41
2023-10-03T16:08:41
{ "title": [ -0.023468018, -0.007411957, -0.027282715, -0.034820557, -0.003774643, -0.013641357, -0.019851685, -0.008003235, 0.021194458, 0.022445679, -0.06341553, -0.010139465, 0.004508972, 0.0014324188, -0.068603516, -0.02279663, -0.020050049, ...
http://arxiv.org/abs/2310.02170v1
Dynamic LLM-Agent Network: An LLM-agent Collaboration Framework with Agent Team Optimization
Zijun Liu, Yanzhe Zhang, Peng Li, Yang Liu, Diyi Yang
cs.CL, cs.AI, cs.MA
Large language model (LLM) agents have been shown effective on a wide range of tasks, and by ensembling multiple LLM agents, their performances could be further improved. Existing approaches employ a fixed set of agents to interact with each other in a static architecture, which limits their generalizability to various...
http://arxiv.org/pdf/2310.02170v1
2023-10-03T16:05:48
2023-10-03T16:05:48
{ "title": [ -0.08312988, -0.014503479, -0.07849121, -0.043884277, -0.020629883, 0.0055389404, -0.059143066, 0.029312134, -0.0236969, 0.020370483, -0.020095825, 0.02444458, -0.033569336, -0.020950317, -0.015419006, -0.010353088, -0.022994995, 0.0...
http://arxiv.org/abs/2310.02168v1
Editing Personality for LLMs
Shengyu Mao, Ningyu Zhang, Xiaohan Wang, Mengru Wang, Yunzhi Yao, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen
cs.CL, cs.AI, cs.CY, cs.LG, cs.MA
This paper introduces an innovative task focused on editing the personality traits of Large Language Models (LLMs). This task seeks to adjust the models' responses to opinion-related questions on specified topics since an individual's personality often manifests in the form of their expressed opinions, thereby showcasi...
http://arxiv.org/pdf/2310.02168v1
2023-10-03T16:02:36
2023-10-03T16:02:36
{ "title": [ -0.03955078, -0.019836426, -0.027114868, 0.023010254, -0.0713501, -0.0038642883, -0.033843994, -0.0056495667, -0.045440674, 0.06341553, -0.017868042, 0.0004439354, 0.040496826, -0.025161743, 0.022125244, -0.001458168, -0.0036201477, ...
http://arxiv.org/abs/2310.02166v1
Large Language Models Meet Knowledge Graphs to Answer Factoid Questions
Mikhail Salnikov, Hai Le, Prateek Rajput, Irina Nikishina, Pavel Braslavski, Valentin Malykh, Alexander Panchenko
cs.CL
Recently, it has been shown that the incorporation of structured knowledge into Large Language Models significantly improves the results for a variety of NLP tasks. In this paper, we propose a method for exploring pre-trained Text-to-Text Language Models enriched with additional information from Knowledge Graphs for an...
http://arxiv.org/pdf/2310.02166v1
2023-10-03T15:57:00
2023-10-03T15:57:00
{ "title": [ -0.0051193237, -0.01928711, -0.0552063, -0.0892334, -0.042510986, -0.009773254, -0.07324219, 0.040130615, -0.007850647, 0.011230469, -0.044189453, -0.008239746, 0.027954102, -0.034698486, 0.025177002, -0.018569946, 0.013679504, 0.039...
http://arxiv.org/abs/2310.02129v1
Unveiling the Pitfalls of Knowledge Editing for Large Language Models
Zhoubo Li, Ningyu Zhang, Yunzhi Yao, Mengru Wang, Xi Chen, Huajun Chen
cs.CL, cs.AI, cs.CV, cs.DB, cs.LG
As the cost associated with fine-tuning Large Language Models (LLMs) continues to rise, recent research efforts have pivoted towards developing methodologies to edit implicit knowledge embedded within LLMs. Yet, there's still a dark cloud lingering overhead -- will knowledge editing trigger butterfly effect? since it i...
http://arxiv.org/pdf/2310.02129v1
2023-10-03T15:10:46
2023-10-03T15:10:46
{ "title": [ -0.047790527, -0.035705566, -0.04940796, -0.06298828, -0.016403198, 0.0017366409, -0.05215454, 0.020385742, 0.0033359528, 0.027114868, -0.023635864, -0.017150879, 0.014205933, -0.00982666, -0.048095703, -0.0005245209, -0.0154953, 0.0...
http://arxiv.org/abs/2310.02124v1
Exploring Collaboration Mechanisms for LLM Agents: A Social Psychology View
Jintian Zhang, Xin Xu, Shumin Deng
cs.CL, cs.AI, cs.CY, cs.LG, cs.MA
As Natural Language Processing (NLP) systems are increasingly employed in intricate social environments, a pressing query emerges: Can these NLP systems mirror human-esque collaborative intelligence, in a multi-agent society consisting of multiple large language models (LLMs)? This paper probes the collaboration mechan...
http://arxiv.org/pdf/2310.02124v1
2023-10-03T15:05:52
2023-10-03T15:05:52
{ "title": [ -0.052825928, -0.013572693, -0.051513672, -0.01914978, -0.029312134, -0.011909485, -0.04949951, 0.010131836, -0.026306152, 0.015731812, -0.014709473, 0.027450562, -0.035491943, -0.024902344, -0.014465332, -0.007411957, -0.033233643, ...
http://arxiv.org/abs/2310.02118v1
TWIZ: The Wizard of Multimodal Conversational-Stimulus
Rafael Ferreira, Diogo Tavares, Diogo Silva, Rodrigo Valério, João Bordalo, Inês Simões, Vasco Ramos, David Semedo, João Magalhães
cs.CL, cs.AI, I.2.7
In this report, we describe the vision, challenges, and scientific contributions of the Task Wizard team, TWIZ, in the Alexa Prize TaskBot Challenge 2022. Our vision, is to build TWIZ bot as an helpful, multimodal, knowledgeable, and engaging assistant that can guide users towards the successful completion of complex m...
http://arxiv.org/pdf/2310.02118v1
2023-10-03T14:59:35
2023-10-03T14:59:35
{ "title": [ 0.016983032, -0.03314209, -0.027954102, -0.01436615, -0.06329346, -0.028289795, -0.030838013, 0.0033397675, -0.03010559, 0.068847656, 0.0042381287, 0.0046653748, 0.03262329, -0.031311035, -0.013130188, 0.010223389, -0.006702423, 0.02...
http://arxiv.org/abs/2310.02107v2
Instance Needs More Care: Rewriting Prompts for Instances Yields Better Zero-Shot Performance
Saurabh Srivastava, Chengyue Huang, Weiguo Fan, Ziyu Yao
cs.CL
Enabling large language models (LLMs) to perform tasks in zero-shot has been an appealing goal owing to its labor-saving (i.e., requiring no task-specific annotations); as such, zero-shot prompting approaches also enjoy better task generalizability. To improve LLMs' zero-shot performance, prior work has focused on devi...
http://arxiv.org/pdf/2310.02107v2
2023-10-05T22:46:24
2023-10-03T14:51:34
{ "title": [ 0.034942627, -0.035888672, -0.0047187805, -0.058044434, -0.049621582, -0.00198555, -0.027328491, -0.0040626526, -0.017349243, 0.020690918, -0.032928467, 0.03326416, 0.010383606, -0.03717041, 0.030548096, -0.022033691, 0.023925781, -0...
http://arxiv.org/abs/2310.05976v1
An evolutionary model of personality traits related to cooperative behavior using a large language model
Reiji Suzuki, Takaya Arita
physics.soc-ph, cs.AI, cs.CL, cs.GT, cs.MA, cs.NE, q-bio.PE, 68W50, 91A22
This paper aims to shed light on the evolutionary dynamics of diverse and social populations by introducing the rich expressiveness of generative models into the trait expression of social agent-based evolutionary models. Specifically, we focus on the evolution of personality traits in the context of a game-theoretic r...
http://arxiv.org/pdf/2310.05976v1
2023-10-03T14:35:27
2023-10-03T14:35:27
{ "title": [ -0.045928955, -0.0069007874, -0.031188965, -0.012367249, -0.01008606, -0.011772156, -0.017730713, -0.0034599304, -0.01763916, 0.04458618, 0.015586853, -0.008544922, 0.0067443848, -0.026763916, -0.043792725, 0.021499634, -0.024353027, ...
http://arxiv.org/abs/2310.02071v3
Towards End-to-End Embodied Decision Making via Multi-modal Large Language Model: Explorations with GPT4-Vision and Beyond
Liang Chen, Yichi Zhang, Shuhuai Ren, Haozhe Zhao, Zefan Cai, Yuchi Wang, Peiyi Wang, Tianyu Liu, Baobao Chang
cs.AI, cs.CL, cs.CV, cs.RO
In this study, we explore the potential of Multimodal Large Language Models (MLLMs) in improving embodied decision-making processes for agents. While Large Language Models (LLMs) have been widely used due to their advanced reasoning skills and vast world knowledge, MLLMs like GPT4-Vision offer enhanced visual understan...
http://arxiv.org/pdf/2310.02071v3
2023-10-16T12:08:00
2023-10-03T14:13:36
{ "title": [ -0.0059318542, 0.0029029846, 0.000008165836, -0.019714355, -0.02230835, -0.011444092, -0.02178955, -0.034454346, 0.0127334595, 0.011161804, -0.02722168, 0.0002644062, 0.0062446594, -0.027053833, -0.012001038, -0.0015487671, -0.008613586,...
http://arxiv.org/abs/2310.02053v1
Controlling Topic-Focus Articulation in Meaning-to-Text Generation using Graph Neural Networks
Chunliu Wang, Rik van Noord, Johan Bos
cs.CL
A bare meaning representation can be expressed in various ways using natural language, depending on how the information is structured on the surface level. We are interested in finding ways to control topic-focus articulation when generating text from meaning. We focus on distinguishing active and passive voice for sen...
http://arxiv.org/pdf/2310.02053v1
2023-10-03T13:51:01
2023-10-03T13:51:01
{ "title": [ -0.05130005, -0.010528564, -0.020462036, -0.008361816, -0.049835205, 0.016815186, -0.05886841, -0.0071754456, -0.028701782, 0.043914795, 0.017990112, 0.036193848, 0.0038833618, -0.027572632, -0.02545166, 0.009292603, -0.023986816, 0....
http://arxiv.org/abs/2310.02050v1
Tuning Large language model for End-to-end Speech Translation
Hao Zhang, Nianwen Si, Yaqi Chen, Wenlin Zhang, Xukui Yang, Dan Qu, Xiaolin Jiao
cs.CL, cs.CV
With the emergence of large language models (LLMs), multimodal models based on LLMs have demonstrated significant potential. Models such as LLaSM, X-LLM, and SpeechGPT exhibit an impressive ability to comprehend and generate human instructions. However, their performance often falters when faced with complex tasks like...
http://arxiv.org/pdf/2310.02050v1
2023-10-03T13:43:50
2023-10-03T13:43:50
{ "title": [ -0.040893555, -0.00762558, -0.027023315, -0.06237793, -0.0070266724, 0.008628845, -0.058898926, 0.018203735, -0.010948181, 0.048675537, -0.0011167526, 0.0010128021, 0.022750854, -0.019378662, -0.039489746, 0.01751709, -0.007209778, 0...
http://arxiv.org/abs/2310.02040v1
Jury: A Comprehensive Evaluation Toolkit
Devrim Cavusoglu, Ulas Sert, Secil Sen, Sinan Altinuc
cs.CL, cs.AI, I.2.7; D.1.3
Evaluation plays a critical role in deep learning as a fundamental block of any prediction-based system. However, the vast number of Natural Language Processing (NLP) tasks and the development of various metrics have led to challenges in evaluating different systems with different metrics. To address these challenges, ...
http://arxiv.org/pdf/2310.02040v1
2023-10-03T13:31:28
2023-10-03T13:31:28
{ "title": [ -0.01651001, -0.014793396, -0.035125732, -0.046783447, 0.020874023, 0.00006920099, -0.026138306, -0.032592773, 0.00484848, 0.0071258545, -0.0052833557, -0.013626099, -0.031341553, -0.047576904, -0.040924072, -0.0055732727, -0.019866943, ...
http://arxiv.org/abs/2310.02031v4
OceanGPT: A Large Language Model for Ocean Science Tasks
Zhen Bi, Ningyu Zhang, Yida Xue, Yixin Ou, Daxiong Ji, Guozhou Zheng, Huajun Chen
cs.CL, cs.AI, cs.CE, cs.LG, cs.RO
Ocean science, which delves into the oceans that are reservoirs of life and biodiversity, is of great significance given that oceans cover over 70% of our planet's surface. Recently, advances in Large Language Models (LLMs) have transformed the paradigm in science. Despite the success in other domains, current LLMs oft...
http://arxiv.org/pdf/2310.02031v4
2023-10-25T10:55:38
2023-10-03T13:17:35
{ "title": [ -0.044281006, -0.03186035, -0.050994873, -0.034729004, -0.0012655258, -0.0051002502, -0.036346436, 0.0009870529, 0.027709961, 0.026489258, -0.020019531, -0.021743774, -0.0004930496, -0.022872925, -0.06008911, 0.010574341, -0.012519836, ...
http://arxiv.org/abs/2310.01991v1
Fill in the Blank: Exploring and Enhancing LLM Capabilities for Backward Reasoning in Math Word Problems
Aniruddha Deb, Neeva Oza, Sarthak Singla, Dinesh Khandelwal, Dinesh Garg, Parag Singla
cs.CL, cs.AI, cs.LG, I.2.3
While forward reasoning (i.e. find the answer given the question) has been explored extensively in the recent literature, backward reasoning is relatively unexplored. We examine the backward reasoning capabilities of LLMs on Math Word Problems (MWPs): given a mathematical question and its answer, with some details omit...
http://arxiv.org/pdf/2310.01991v1
2023-10-03T12:03:06
2023-10-03T12:03:06
{ "title": [ -0.01777649, -0.023834229, -0.031097412, -0.020523071, -0.04510498, -0.008651733, -0.040130615, 0.00021326542, 0.018234253, 0.0060577393, -0.026000977, 0.008003235, -0.020446777, -0.030380249, -0.015335083, -0.022888184, -0.0121154785, ...
http://arxiv.org/abs/2310.01960v1
Language Models as Knowledge Bases for Visual Word Sense Disambiguation
Anastasia Kritharoula, Maria Lymperaiou, Giorgos Stamou
cs.CL, cs.AI
Visual Word Sense Disambiguation (VWSD) is a novel challenging task that lies between linguistic sense disambiguation and fine-grained multimodal retrieval. The recent advancements in the development of visiolinguistic (VL) transformers suggest some off-the-self implementations with encouraging results, which however w...
http://arxiv.org/pdf/2310.01960v1
2023-10-03T11:11:55
2023-10-03T11:11:55
{ "title": [ -0.020309448, -0.0012588501, -0.012069702, -0.041503906, -0.021713257, 0.016967773, -0.05517578, -0.011131287, 0.05709839, 0.044281006, -0.053527832, -0.003200531, 0.01663208, 0.0074806213, -0.0836792, 0.0021648407, 0.010444641, 0.01...
http://arxiv.org/abs/2310.01957v2
Driving with LLMs: Fusing Object-Level Vector Modality for Explainable Autonomous Driving
Long Chen, Oleg Sinavski, Jan Hünermann, Alice Karnsund, Andrew James Willmott, Danny Birch, Daniel Maund, Jamie Shotton
cs.RO, cs.AI, cs.CL, cs.CV
Large Language Models (LLMs) have shown promise in the autonomous driving sector, particularly in generalization and interpretability. We introduce a unique object-level multimodal LLM architecture that merges vectorized numeric modalities with a pre-trained LLM to improve context understanding in driving situations. W...
http://arxiv.org/pdf/2310.01957v2
2023-10-13T21:59:27
2023-10-03T11:05:14
{ "title": [ -0.008552551, -0.0071258545, -0.021911621, -0.043121338, -0.0058631897, -0.018539429, -0.0056533813, -0.017211914, -0.034576416, 0.014343262, -0.04019165, 0.045196533, -0.026626587, -0.013679504, 0.011657715, -0.022766113, 0.00038027763,...
http://arxiv.org/abs/2310.01929v1
Navigating Cultural Chasms: Exploring and Unlocking the Cultural POV of Text-To-Image Models
Mor Ventura, Eyal Ben-David, Anna Korhonen, Roi Reichart
cs.CL, cs.AI, cs.LG
Text-To-Image (TTI) models, exemplified by DALL-E and StableDiffusion, have recently gained prominence for their remarkable zero-shot capabilities in generating images guided by textual prompts. Language, as a conduit of culture, plays a pivotal role in these models' multilingual capabilities, which in turn shape their...
http://arxiv.org/pdf/2310.01929v1
2023-10-03T10:13:36
2023-10-03T10:13:36
{ "title": [ -0.0041122437, -0.013145447, 0.028503418, -0.01928711, -0.013427734, -0.0015907288, -0.015449524, -0.04434204, 0.025115967, 0.03152466, -0.03274536, 0.019729614, 0.01222229, -0.01638794, -0.034454346, 0.023025513, 0.0057678223, 0.003...
http://arxiv.org/abs/2310.01917v2
Hierarchical Evaluation Framework: Best Practices for Human Evaluation
Iva Bojic, Jessica Chen, Si Yuan Chang, Qi Chwen Ong, Shafiq Joty, Josip Car
cs.CL, cs.HC
Human evaluation plays a crucial role in Natural Language Processing (NLP) as it assesses the quality and relevance of developed systems, thereby facilitating their enhancement. However, the absence of widely accepted human evaluation metrics in NLP hampers fair comparisons among different systems and the establishment...
http://arxiv.org/pdf/2310.01917v2
2023-10-12T07:59:56
2023-10-03T09:46:02
{ "title": [ -0.0016479492, 0.004638672, -0.0413208, -0.06561279, 0.028213501, -0.0154800415, -0.016418457, -0.026000977, -0.019439697, 0.03189087, 0.00065660477, 0.009552002, -0.010879517, -0.0335083, -0.035003662, 0.008514404, -0.018585205, 0.0...
http://arxiv.org/abs/2310.01889v3
Ring Attention with Blockwise Transformers for Near-Infinite Context
Hao Liu, Matei Zaharia, Pieter Abbeel
cs.CL
Transformers have emerged as the architecture of choice for many state-of-the-art AI models, showcasing exceptional performance across a wide range of AI applications. However, the memory demands imposed by Transformers limit their ability to handle long sequences, thereby creating challenges for tasks involving extend...
http://arxiv.org/pdf/2310.01889v3
2023-10-12T01:00:09
2023-10-03T08:44:50
{ "title": [ 0.021850586, 0.023483276, -0.04660034, -0.10443115, 0.037963867, -0.0019369125, -0.039093018, 0.00011622906, 0.004180908, 0.053527832, -0.047790527, -0.0038013458, 0.024353027, 0.007408142, -0.052612305, -0.010574341, -0.0038871765, ...
http://arxiv.org/abs/2310.01886v2
Effective and Parameter-Efficient Reusing Fine-Tuned Models
Weisen Jiang, Baijiong Lin, Han Shi, Yu Zhang, Zhenguo Li, James T. Kwok
cs.LG, cs.CL, cs.CV
Many pre-trained large-scale models provided online have become highly effective in transferring to downstream tasks. At the same time, various task-specific models fine-tuned on these pre-trained models are available online for public use. In practice, as collecting task-specific data is labor-intensive and fine-tunin...
http://arxiv.org/pdf/2310.01886v2
2023-10-04T02:30:27
2023-10-03T08:39:33
{ "title": [ -0.00074481964, -0.0074539185, -0.015525818, -0.10534668, 0.005065918, 0.02130127, -0.06567383, -0.008735657, -0.0034809113, 0.046966553, -0.009918213, -0.008728027, 0.010856628, -0.008270264, -0.04425049, 0.023101807, 0.0037517548, ...
http://arxiv.org/abs/2310.01854v1
Fine-tuned vs. Prompt-tuned Supervised Representations: Which Better Account for Brain Language Representations?
Jingyuan Sun, Marie-Francine Moens
cs.AI, cs.CL
To decipher the algorithm underlying the human brain's language representation, previous work probed brain responses to language input with pre-trained artificial neural network (ANN) models fine-tuned on NLU tasks. However, full fine-tuning generally updates the entire parametric space and distorts pre-trained feature...
http://arxiv.org/pdf/2310.01854v1
2023-10-03T07:34:30
2023-10-03T07:34:30
{ "title": [ 0.0019683838, -0.0107040405, -0.017074585, -0.0496521, -0.009918213, -0.01285553, -0.036621094, -0.00009250641, -0.028244019, 0.019012451, 0.0019035339, -0.0073165894, -0.0077171326, -0.03894043, -0.028671265, -0.004886627, 0.002155304, ...