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/2311.08588v1
CodeScope: An Execution-based Multilingual Multitask Multidimensional Benchmark for Evaluating LLMs on Code Understanding and Generation
Weixiang Yan, Haitian Liu, Yunkun Wang, Yunzhe Li, Qian Chen, Wen Wang, Tingyu Lin, Weishan Zhao, Li Zhu, Shuiguang Deng, Hari Sundaram
cs.CL, cs.AI, cs.SE
Large Language Models (LLMs) have demonstrated remarkable performance on coding related tasks, particularly on assisting humans in programming and facilitating programming automation. However, existing benchmarks for evaluating the code understanding and generation capacities of LLMs suffer from severe limitations. Fir...
http://arxiv.org/pdf/2311.08588v1
2023-11-14T23:18:52
2023-11-14T23:18:52
{ "title": [ -0.066589355, -0.03894043, -0.03253174, -0.032409668, -0.06390381, 0.008178711, -0.075683594, 0.02217102, 0.02571106, 0.024215698, -0.017822266, 0.022766113, -0.002090454, -0.026855469, -0.0062026978, -0.007133484, -0.0041618347, 0.0...
http://arxiv.org/abs/2311.08584v1
Asking More Informative Questions for Grounded Retrieval
Sedrick Keh, Justin T. Chiu, Daniel Fried
cs.CL
When a model is trying to gather information in an interactive setting, it benefits from asking informative questions. However, in the case of a grounded multi-turn image identification task, previous studies have been constrained to polar yes/no questions, limiting how much information the model can gain in a single t...
http://arxiv.org/pdf/2311.08584v1
2023-11-14T23:13:27
2023-11-14T23:13:27
{ "title": [ 0.012451172, 0.020004272, -0.04537964, -0.053619385, 0.020355225, -0.017745972, -0.0074272156, -0.025848389, 0.015945435, 0.01802063, -0.035064697, 0.015266418, -0.016677856, -0.016967773, -0.037200928, 0.007987976, 0.014228821, 0.00...
http://arxiv.org/abs/2311.08579v1
Graph-Induced Syntactic-Semantic Spaces in Transformer-Based Variational AutoEncoders
Yingji Zhang, Marco Valentino, Danilo S. Carvalho, Ian Pratt-Hartmann, André Freitas
cs.CL
The injection of syntactic information in Variational AutoEncoders (VAEs) has been shown to result in an overall improvement of performances and generalisation. An effective strategy to achieve such a goal is to separate the encoding of distributional semantic features and syntactic structures into heterogeneous latent...
http://arxiv.org/pdf/2311.08579v1
2023-11-14T22:47:23
2023-11-14T22:47:23
{ "title": [ -0.005519867, 0.005420685, -0.03729248, -0.04348755, -0.0054969788, -0.00030231476, -0.051849365, -0.0096206665, 0.0005993843, 0.036254883, -0.0054397583, 0.022109985, 0.013244629, -0.021713257, -0.016616821, 0.011177063, 0.002216339, ...
http://arxiv.org/abs/2311.08576v1
Towards Evaluating AI Systems for Moral Status Using Self-Reports
Ethan Perez, Robert Long
cs.LG, cs.AI, cs.CL
As AI systems become more advanced and widely deployed, there will likely be increasing debate over whether AI systems could have conscious experiences, desires, or other states of potential moral significance. It is important to inform these discussions with empirical evidence to the extent possible. We argue that und...
http://arxiv.org/pdf/2311.08576v1
2023-11-14T22:45:44
2023-11-14T22:45:44
{ "title": [ -0.002336502, 0.009216309, -0.030075073, -0.0038986206, 0.020111084, -0.033996582, 0.04763794, -0.024917603, -0.018127441, 0.026184082, -0.03326416, 0.005874634, -0.03250122, -0.0029392242, -0.054107666, -0.018447876, -0.022613525, 0...
http://arxiv.org/abs/2311.08572v1
Parameter-Efficient Multilingual Summarisation: An Empirical Study
Chenxi Whitehouse, Fantine Huot, Jasmijn Bastings, Mostafa Dehghani, Chu-Cheng Lin, Mirella Lapata
cs.CL, cs.AI, cs.LG
With the increasing prevalence of Large Language Models, traditional full fine-tuning approaches face growing challenges, especially in memory-intensive tasks. This paper investigates the potential of Parameter-Efficient Fine-Tuning, focusing on Low-Rank Adaptation (LoRA), for complex and under-explored multilingual su...
http://arxiv.org/pdf/2311.08572v1
2023-11-14T22:32:39
2023-11-14T22:32:39
{ "title": [ -0.011665344, -0.008956909, -0.020736694, -0.09753418, 0.0029449463, 0.021331787, -0.07495117, -0.0015335083, -0.010559082, 0.059265137, 0.020111084, 0.011184692, 0.0085372925, -0.0035972595, -0.055358887, 0.018341064, -0.0068092346, ...
http://arxiv.org/abs/2311.08562v1
MAgIC: Benchmarking Large Language Model Powered Multi-Agent in Cognition, Adaptability, Rationality and Collaboration
Lin Xu, Zhiyuan Hu, Daquan Zhou, Hongyu Ren, Zhen Dong, Kurt Keutzer, See Kiong Ng, Jiashi Feng
cs.CL
Large Language Models (LLMs) have marked a significant advancement in the field of natural language processing, demonstrating exceptional capabilities in reasoning, tool usage, and memory. As their applications extend into multi-agent environments, a need has arisen for a comprehensive evaluation framework that capture...
http://arxiv.org/pdf/2311.08562v1
2023-11-14T21:46:27
2023-11-14T21:46:27
{ "title": [ -0.050476074, 0.0055389404, -0.045135498, -0.01751709, -0.015975952, -0.02897644, -0.04763794, -0.010238647, -0.0036354065, 0.035003662, 0.018798828, 0.003446579, 0.008369446, -0.030593872, -0.024871826, -0.002445221, -0.012550354, 0...
http://arxiv.org/abs/2311.08552v1
UT5: Pretraining Non autoregressive T5 with unrolled denoising
Mahmoud G. Salem, Jiayu Ye, Chu-Cheng Lin, Frederick Liu
cs.CL
Recent advances in Transformer-based Large Language Models have made great strides in natural language generation. However, to decode K tokens, an autoregressive model needs K sequential forward passes, which may be a performance bottleneck for large language models. Many non-autoregressive (NAR) research are aiming to...
http://arxiv.org/pdf/2311.08552v1
2023-11-14T21:28:10
2023-11-14T21:28:10
{ "title": [ -0.008964539, 0.026046753, -0.049713135, -0.06225586, -0.0045204163, -0.0004439354, -0.07183838, -0.027053833, -0.009727478, 0.025436401, -0.0046958923, 0.0015268326, 0.028717041, -0.055603027, 0.006965637, 0.011779785, 0.01701355, 0...
http://arxiv.org/abs/2311.08545v1
Efficient Continual Pre-training for Building Domain Specific Large Language Models
Yong Xie, Karan Aggarwal, Aitzaz Ahmad
cs.CL
Large language models (LLMs) have demonstrated remarkable open-domain capabilities. Traditionally, LLMs tailored for a domain are trained from scratch to excel at handling domain-specific tasks. In this work, we explore an alternative strategy of continual pre-training as a means to develop domain-specific LLMs. We int...
http://arxiv.org/pdf/2311.08545v1
2023-11-14T21:19:14
2023-11-14T21:19:14
{ "title": [ -0.007499695, -0.01625061, -0.042114258, -0.11462402, 0.010520935, 0.0044059753, -0.07684326, 0.02848816, 0.0021133423, 0.029708862, -0.0063934326, 0.015571594, 0.013198853, -0.019744873, -0.026367188, 0.01977539, -0.007358551, 0.042...
http://arxiv.org/abs/2311.08538v1
Extending Multilingual Machine Translation through Imitation Learning
Wen Lai, Viktor Hangya, Alexander Fraser
cs.CL
Despite the growing variety of languages supported by existing multilingual neural machine translation (MNMT) models, most of the world's languages are still being left behind. We aim to extend large-scale MNMT models to a new language, allowing for translation between the newly added and all of the already supported l...
http://arxiv.org/pdf/2311.08538v1
2023-11-14T21:04:03
2023-11-14T21:04:03
{ "title": [ -0.0206604, -0.017074585, 0.010253906, -0.04928589, -0.015792847, 0.021774292, -0.03353882, -0.0050086975, -0.02418518, 0.045532227, -0.010116577, 0.019500732, 0.005722046, -0.015792847, -0.031585693, 0.0023822784, -0.01348114, 0.040...
http://arxiv.org/abs/2311.08533v1
Natural Language Processing for Financial Regulation
Ixandra Achitouv, Dragos Gorduza, Antoine Jacquier
cs.CL, cs.LG, q-fin.CP
This article provides an understanding of Natural Language Processing techniques in the framework of financial regulation, more specifically in order to perform semantic matching search between rules and policy when no dataset is available for supervised learning. We outline how to outperform simple pre-trained sentenc...
http://arxiv.org/pdf/2311.08533v1
2023-11-14T20:58:21
2023-11-14T20:58:21
{ "title": [ -0.010009766, 0.00016343594, -0.036254883, -0.05090332, -0.004802704, -0.004371643, -0.032684326, -0.0026550293, -0.012664795, 0.038757324, -0.047088623, -0.00042653084, 0.016571045, -0.0006055832, -0.041809082, -0.009475708, -0.01679992...
http://arxiv.org/abs/2311.08526v1
GLiNER: Generalist Model for Named Entity Recognition using Bidirectional Transformer
Urchade Zaratiana, Nadi Tomeh, Pierre Holat, Thierry Charnois
cs.CL, cs.AI, cs.LG
Named Entity Recognition (NER) is essential in various Natural Language Processing (NLP) applications. Traditional NER models are effective but limited to a set of predefined entity types. In contrast, Large Language Models (LLMs) can extract arbitrary entities through natural language instructions, offering greater fl...
http://arxiv.org/pdf/2311.08526v1
2023-11-14T20:39:12
2023-11-14T20:39:12
{ "title": [ -0.0135269165, 0.013427734, -0.05279541, -0.08355713, 0.032470703, 0.009162903, -0.045074463, -0.0016565323, 0.005836487, 0.025604248, -0.03265381, 0.01374054, 0.013847351, -0.008010864, -0.028305054, -0.00041270256, -0.01902771, 0.0...
http://arxiv.org/abs/2311.08516v1
LLMs cannot find reasoning errors, but can correct them!
Gladys Tyen, Hassan Mansoor, Peter Chen, Tony Mak, Victor Cărbune
cs.AI, cs.CL, cs.LG
While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023; Madaan et al., 2023), recent attempts to self-correct logical or reasoning errors often cause correct answers to become incorrect, resulting in worse performances overall (Huang et al., 2023). In this...
http://arxiv.org/pdf/2311.08516v1
2023-11-14T20:12:38
2023-11-14T20:12:38
{ "title": [ 0.017822266, -0.011039734, 0.010017395, -0.029815674, -0.019897461, 0.007041931, -0.020339966, -0.03201294, 0.01576233, 0.036376953, -0.047576904, -0.013420105, 0.011543274, -0.019241333, -0.026229858, -0.029525757, -0.041900635, 0.0...
http://arxiv.org/abs/2311.08511v1
CoRE-CoG: Conversational Recommendation of Entities using Constrained Generation
Harshvardhan Srivastava, Kanav Pruthi, Soumen Chakrabarti, Mausam
cs.CL
End-to-end conversational recommendation systems (CRS) generate responses by leveraging both dialog history and a knowledge base (KB). A CRS mainly faces three key challenges: (1) at each turn, it must decide if recommending a KB entity is appropriate; if so, it must identify the most relevant KB entity to recommend; a...
http://arxiv.org/pdf/2311.08511v1
2023-11-14T20:07:34
2023-11-14T20:07:34
{ "title": [ -0.0053520203, -0.009254456, -0.04736328, -0.0513916, 0.014038086, -0.004585266, -0.024856567, 0.00868988, 0.015960693, 0.038330078, -0.03286743, 0.025024414, 0.03152466, -0.008201599, -0.031463623, 0.017501831, -0.023956299, 0.03762...
http://arxiv.org/abs/2311.08505v1
Semi-Structured Chain-of-Thought: Integrating Multiple Sources of Knowledge for Improved Language Model Reasoning
Xin Su, Tiep Le, Steven Bethard, Phillip Howard
cs.CL
An important open question pertaining to the use of large language models for knowledge-intensive tasks is how to effectively integrate knowledge from three sources: the model's parametric memory, external structured knowledge, and external unstructured knowledge. Most existing prompting methods either rely solely on o...
http://arxiv.org/pdf/2311.08505v1
2023-11-14T19:53:53
2023-11-14T19:53:53
{ "title": [ -0.01600647, -0.026519775, -0.025482178, -0.091308594, -0.008346558, 0.013572693, -0.07232666, 0.020355225, 0.01461792, 0.008224487, -0.017929077, 0.006652832, -0.0011005402, -0.024291992, -0.028625488, -0.0022640228, -0.011177063, -...
http://arxiv.org/abs/2311.08487v1
Alignment is not sufficient to prevent large language models from generating harmful information: A psychoanalytic perspective
Zi Yin, Wei Ding, Jia Liu
cs.CL, cs.AI
Large Language Models (LLMs) are central to a multitude of applications but struggle with significant risks, notably in generating harmful content and biases. Drawing an analogy to the human psyche's conflict between evolutionary survival instincts and societal norm adherence elucidated in Freud's psychoanalysis theory...
http://arxiv.org/pdf/2311.08487v1
2023-11-14T19:28:51
2023-11-14T19:28:51
{ "title": [ -0.023986816, 0.002664566, -0.0335083, -0.039611816, -0.008415222, -0.0073165894, -0.031433105, -0.007659912, -0.02810669, 0.041015625, 0.008468628, -0.0024700165, 0.012199402, -0.02923584, -0.02923584, 0.005832672, -0.0010299683, 0....
http://arxiv.org/abs/2311.08481v1
Functionality learning through specification instructions
Pedro Henrique Luz de Araujo, Benjamin Roth
cs.CL
Test suites assess natural language processing models' performance on specific functionalities: cases of interest involving model robustness, fairness, or particular linguistic capabilities. They enable fine-grained evaluations of model aspects that would otherwise go unnoticed in standard evaluation datasets, but they...
http://arxiv.org/pdf/2311.08481v1
2023-11-14T19:15:55
2023-11-14T19:15:55
{ "title": [ -0.012260437, -0.0071411133, -0.017242432, -0.04284668, -0.012008667, -0.008399963, -0.03286743, -0.008468628, -0.021209717, 0.012702942, -0.04562378, 0.0051460266, -0.03756714, -0.011909485, -0.021987915, -0.024215698, 0.012039185, ...
http://arxiv.org/abs/2311.08472v1
Selecting Shots for Demographic Fairness in Few-Shot Learning with Large Language Models
Carlos Aguirre, Kuleen Sasse, Isabel Cachola, Mark Dredze
cs.CL
Recently, work in NLP has shifted to few-shot (in-context) learning, with large language models (LLMs) performing well across a range of tasks. However, while fairness evaluations have become a standard for supervised methods, little is known about the fairness of LLMs as prediction systems. Further, common standard me...
http://arxiv.org/pdf/2311.08472v1
2023-11-14T19:02:03
2023-11-14T19:02:03
{ "title": [ -0.0043754578, -0.022109985, -0.039031982, -0.02305603, -0.008712769, -0.019470215, -0.018722534, 0.012260437, -0.017486572, 0.011405945, -0.04611206, 0.015617371, -0.025344849, -0.020050049, -0.014564514, -0.022445679, -0.008308411, ...
http://arxiv.org/abs/2311.08469v1
UNcommonsense Reasoning: Abductive Reasoning about Uncommon Situations
Wenting Zhao, Justin T Chiu, Jena D. Hwang, Faeze Brahman, Jack Hessel, Sanjiban Choudhury, Yejin Choi, Xiang Lorraine Li, Alane Suhr
cs.CL
Language technologies that accurately model the dynamics of events must perform commonsense reasoning. Existing work evaluating commonsense reasoning focuses on making inferences about common, everyday situations. To instead investigate the ability to model unusual, unexpected, and unlikely situations, we explore the t...
http://arxiv.org/pdf/2311.08469v1
2023-11-14T19:00:55
2023-11-14T19:00:55
{ "title": [ 0.0012168884, -0.012802124, -0.034606934, -0.014656067, 0.0030059814, -0.029693604, 0.020065308, -0.022705078, -0.025527954, -0.0027675629, -0.032226562, 0.0015449524, -0.032989502, -0.04168701, 0.0068740845, -0.039154053, -0.018463135, ...
http://arxiv.org/abs/2311.08402v1
Retrieve and Copy: Scaling ASR Personalization to Large Catalogs
Sai Muralidhar Jayanthi, Devang Kulshreshtha, Saket Dingliwal, Srikanth Ronanki, Sravan Bodapati
cs.CL, cs.IR, cs.SD, eess.AS
Personalization of automatic speech recognition (ASR) models is a widely studied topic because of its many practical applications. Most recently, attention-based contextual biasing techniques are used to improve the recognition of rare words and domain specific entities. However, due to performance constraints, the bia...
http://arxiv.org/pdf/2311.08402v1
2023-11-14T18:59:24
2023-11-14T18:59:24
{ "title": [ -0.0004544258, -0.012214661, -0.05657959, -0.08996582, -0.005367279, -0.009086609, -0.040496826, 0.034240723, -0.044036865, 0.05996704, 0.0110321045, 0.0025367737, 0.026916504, -0.021942139, -0.0070266724, 0.00076293945, 0.023773193, ...
http://arxiv.org/abs/2311.08401v1
Fine-tuning Language Models for Factuality
Katherine Tian, Eric Mitchell, Huaxiu Yao, Christopher D. Manning, Chelsea Finn
cs.CL, cs.AI, cs.LG
The fluency and creativity of large pre-trained language models (LLMs) have led to their widespread use, sometimes even as a replacement for traditional search engines. Yet language models are prone to making convincing but factually inaccurate claims, often referred to as 'hallucinations.' These errors can inadvertent...
http://arxiv.org/pdf/2311.08401v1
2023-11-14T18:59:15
2023-11-14T18:59:15
{ "title": [ -0.01335907, -0.009307861, -0.03164673, -0.070251465, -0.01600647, -0.010475159, -0.053985596, -0.0013866425, -0.0045318604, 0.04046631, -0.006439209, -0.0058555603, 0.0206604, -0.013336182, -0.045043945, 0.026107788, -0.018600464, 0...
http://arxiv.org/abs/2311.08398v1
Are Large Language Models Temporally Grounded?
Yifu Qiu, Zheng Zhao, Yftah Ziser, Anna Korhonen, Edoardo M. Ponti, Shay B. Cohen
cs.CL, cs.AI
Are Large language models (LLMs) temporally grounded? Since LLMs cannot perceive and interact with the environment, it is impossible to answer this question directly. Instead, we provide LLMs with textual narratives and probe them with respect to their common-sense knowledge of the structure and duration of events, the...
http://arxiv.org/pdf/2311.08398v1
2023-11-14T18:57:15
2023-11-14T18:57:15
{ "title": [ -0.002445221, 0.013267517, -0.061035156, -0.09320068, -0.01121521, -0.025650024, -0.069885254, 0.0335083, 0.0027828217, 0.035064697, -0.031677246, -0.018417358, 0.018310547, -0.021652222, -0.03878784, -0.0038051605, 0.00021743774, 0....
http://arxiv.org/abs/2311.08396v1
Zero-shot audio captioning with audio-language model guidance and audio context keywords
Leonard Salewski, Stefan Fauth, A. Sophia Koepke, Zeynep Akata
eess.AS, cs.AI, cs.CL, cs.SD
Zero-shot audio captioning aims at automatically generating descriptive textual captions for audio content without prior training for this task. Different from speech recognition which translates audio content that contains spoken language into text, audio captioning is commonly concerned with ambient sounds, or sounds...
http://arxiv.org/pdf/2311.08396v1
2023-11-14T18:55:48
2023-11-14T18:55:48
{ "title": [ 0.0048980713, -0.0158844, -0.04373169, -0.044769287, -0.025360107, -0.02999878, -0.025115967, 0.013999939, -0.008415222, 0.07623291, -0.0064697266, 0.016342163, 0.045654297, -0.015472412, -0.022903442, 0.030517578, 0.0007958412, 0.01...
http://arxiv.org/abs/2311.08391v1
A Material Lens on Coloniality in NLP
William Held, Camille Harris, Michael Best, Diyi Yang
cs.CL
Coloniality, the continuation of colonial harms beyond "official" colonization, has pervasive effects across society and scientific fields. Natural Language Processing (NLP) is no exception to this broad phenomenon. In this work, we argue that coloniality is implicitly embedded in and amplified by NLP data, algorithms,...
http://arxiv.org/pdf/2311.08391v1
2023-11-14T18:52:09
2023-11-14T18:52:09
{ "title": [ -0.03375244, -0.024749756, -0.004764557, -0.011428833, -0.06530762, -0.0043144226, -0.04168701, -0.0049972534, -0.0079574585, 0.0012426376, -0.049835205, -0.0037517548, -0.024398804, -0.02381897, -0.014419556, -0.026062012, -0.011291504,...
http://arxiv.org/abs/2311.08390v1
On What Basis? Predicting Text Preference Via Structured Comparative Reasoning
Jing Nathan Yan, Tianqi Liu, Justin T Chiu, Jiaming Shen, Zhen Qin, Yue Yu, Yao Zhao, Charu Lakshmanan, Yair Kurzion, Alexander M. Rush, Jialu Liu, Michael Bendersky
cs.CL
Comparative reasoning plays a crucial role in text preference prediction; however, large language models (LLMs) often demonstrate inconsistencies in their reasoning. While approaches like Chain-of-Thought improve accuracy in many other settings, they struggle to consistently distinguish the similarities and differences...
http://arxiv.org/pdf/2311.08390v1
2023-11-14T18:51:38
2023-11-14T18:51:38
{ "title": [ -0.009101868, -0.011940002, -0.014518738, -0.057769775, -0.0077209473, 0.0030021667, -0.00920105, -0.008369446, -0.011123657, 0.011459351, -0.040618896, 0.03164673, -0.027786255, -0.0020999908, -0.03050232, -0.02029419, -0.022918701, ...
http://arxiv.org/abs/2311.08389v1
TSST: A Benchmark and Evaluation Models for Text Speech-Style Transfer
Huashan Sun, Yixiao Wu, Yinghao Li, Jiawei Li, Yizhe Yang, Yang Gao
cs.CL, cs.AI, I.2.7
Text style is highly abstract, as it encompasses various aspects of a speaker's characteristics, habits, logical thinking, and the content they express. However, previous text-style transfer tasks have primarily focused on data-driven approaches, lacking in-depth analysis and research from the perspectives of linguisti...
http://arxiv.org/pdf/2311.08389v1
2023-11-14T18:50:51
2023-11-14T18:50:51
{ "title": [ -0.01411438, -0.010131836, -0.039733887, -0.05734253, 0.012062073, 0.0013504028, -0.049591064, 0.0055389404, 0.012573242, 0.056671143, 0.0015649796, 0.023788452, 0.039489746, -0.011871338, -0.047210693, 0.035003662, -0.0061149597, 0....
http://arxiv.org/abs/2311.08385v2
ChOiRe: Characterizing and Predicting Human Opinions with Chain of Opinion Reasoning
Xuan Long Do, Kenji Kawaguchi, Min-Yen Kan, Nancy F. Chen
cs.CL
Aligning language models (LMs) with human opinion is challenging yet vital to enhance their grasp of human values, preferences, and beliefs. We present ChOiRe, a four-step solution framework to predict human opinion that differentiates between the user explicit personae (i.e. demographic or ideological attributes) that...
http://arxiv.org/pdf/2311.08385v2
2023-11-15T16:40:13
2023-11-14T18:48:27
{ "title": [ -0.0063056946, -0.018692017, -0.039916992, -0.051086426, 0.0029754639, -0.010795593, -0.010307312, -0.0023326874, 0.0107803345, 0.023086548, -0.031799316, -0.0021209717, -0.016052246, -0.0110321045, -0.05078125, -0.009353638, -0.01452636...
http://arxiv.org/abs/2311.08380v1
Direct Preference Optimization for Neural Machine Translation with Minimum Bayes Risk Decoding
Guangyu Yang, Jinghong Chen, Weizhe Lin, Bill Byrne
cs.CL
Minimum Bayes Risk (MBR) decoding can significantly improve translation performance of Multilingual Large Language Models (MLLMs). However, MBR decoding is computationally expensive and in this paper, we show how recently developed Reinforcement Learning (RL) technique, Direct Preference Optimization (DPO) can be used ...
http://arxiv.org/pdf/2311.08380v1
2023-11-14T18:43:51
2023-11-14T18:43:51
{ "title": [ -0.027786255, -0.011543274, -0.0044937134, -0.048187256, -0.013786316, 0.009307861, -0.040802002, 0.0020923615, -0.031799316, 0.049865723, -0.016540527, 0.0093688965, 0.021652222, -0.021514893, -0.03164673, 0.001294136, -0.01600647, ...
http://arxiv.org/abs/2311.08377v1
Learning to Filter Context for Retrieval-Augmented Generation
Zhiruo Wang, Jun Araki, Zhengbao Jiang, Md Rizwan Parvez, Graham Neubig
cs.CL, cs.AI
On-the-fly retrieval of relevant knowledge has proven an essential element of reliable systems for tasks such as open-domain question answering and fact verification. However, because retrieval systems are not perfect, generation models are required to generate outputs given partially or entirely irrelevant passages. T...
http://arxiv.org/pdf/2311.08377v1
2023-11-14T18:41:54
2023-11-14T18:41:54
{ "title": [ 0.008323669, 0.02067566, -0.037384033, -0.06866455, 0.034698486, -0.0065994263, -0.009971619, -0.0085372925, 0.02897644, 0.041809082, -0.018081665, 0.0022296906, 0.0118637085, -0.014015198, -0.038879395, 0.01625061, -0.023590088, 0.0...
http://arxiv.org/abs/2311.08374v1
A Ship of Theseus: Curious Cases of Paraphrasing in LLM-Generated Texts
Nafis Irtiza Tripto, Saranya Venkatraman, Dominik Macko, Robert Moro, Ivan Srba, Adaku Uchendu, Thai Le, Dongwon Lee
cs.CL
In the realm of text manipulation and linguistic transformation, the question of authorship has always been a subject of fascination and philosophical inquiry. Much like the \textbf{Ship of Theseus paradox}, which ponders whether a ship remains the same when each of its original planks is replaced, our research delves ...
http://arxiv.org/pdf/2311.08374v1
2023-11-14T18:40:42
2023-11-14T18:40:42
{ "title": [ -0.0423584, -0.019348145, -0.008674622, -0.0070610046, -0.059631348, 0.011642456, -0.029266357, -0.02368164, -0.043670654, 0.038024902, -0.030593872, 0.04067993, 0.004722595, -0.02748108, -0.0060157776, -0.012336731, -0.034423828, 0....
http://arxiv.org/abs/2311.08370v1
SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models
Bertie Vidgen, Hannah Rose Kirk, Rebecca Qian, Nino Scherrer, Anand Kannappan, Scott A. Hale, Paul Röttger
cs.CL
The past year has seen rapid acceleration in the development of large language models (LLMs). For many tasks, there is now a wide range of open-source and open-access LLMs that are viable alternatives to proprietary models like ChatGPT. Without proper steering and safeguards, however, LLMs will readily follow malicious...
http://arxiv.org/pdf/2311.08370v1
2023-11-14T18:33:43
2023-11-14T18:33:43
{ "title": [ -0.020309448, -0.008201599, -0.05078125, -0.10546875, 0.018554688, 0.000121474266, -0.074279785, 0.008392334, -0.008705139, 0.028945923, -0.0062789917, 0.016052246, 0.00044035912, -0.031158447, -0.022949219, 0.01499176, -0.0052871704, ...
http://arxiv.org/abs/2311.08369v1
How You Prompt Matters! Even Task-Oriented Constraints in Instructions Affect LLM-Generated Text Detection
Ryuto Koike, Masahiro Kaneko, Naoaki Okazaki
cs.CL
Against the misuse (e.g., plagiarism or spreading misinformation) of Large Language Models (LLMs), many recent works have presented LLM-generated-text detectors with promising detection performance. Spotlighting a situation where users instruct LLMs to generate texts (e.g., essay writing), there are various ways to wri...
http://arxiv.org/pdf/2311.08369v1
2023-11-14T18:32:52
2023-11-14T18:32:52
{ "title": [ -0.06518555, -0.00012874603, -0.055541992, 0.024597168, -0.04626465, -0.02331543, -0.057281494, -0.015792847, -0.02482605, 0.013038635, -0.0357666, 0.026824951, -0.0040626526, -0.030227661, 0.02772522, -0.02519226, -0.017028809, 0.02...
http://arxiv.org/abs/2311.08360v2
The Transient Nature of Emergent In-Context Learning in Transformers
Aaditya K. Singh, Stephanie C. Y. Chan, Ted Moskovitz, Erin Grant, Andrew M. Saxe, Felix Hill
cs.LG, cs.AI, cs.CL
Transformer neural networks can exhibit a surprising capacity for in-context learning (ICL) despite not being explicitly trained for it. Prior work has provided a deeper understanding of how ICL emerges in transformers, e.g. through the lens of mechanistic interpretability, Bayesian inference, or by examining the distr...
http://arxiv.org/pdf/2311.08360v2
2023-11-15T04:02:44
2023-11-14T18:03:20
{ "title": [ 0.015541077, 0.009475708, -0.014694214, -0.041412354, 0.02255249, -0.02571106, 0.02696228, -0.00687027, -0.023071289, 0.029190063, -0.048828125, 0.027938843, -0.035705566, -0.009468079, -0.03265381, -0.02053833, 0.0025119781, 0.03640...
http://arxiv.org/abs/2311.08349v1
Artificial Text Boundary Detection with Topological Data Analysis and Sliding Window Techniques
Laida Kushnareva, Tatiana Gaintseva, German Magai, Serguei Barannikov, Dmitry Abulkhanov, Kristian Kuznetsov, Irina Piontkovskaya, Sergey Nikolenko
cs.CL
Due to the rapid development of text generation models, people increasingly often encounter texts that may start out as written by a human but then continue as machine-generated results of large language models. Detecting the boundary between human-written and machine-generated parts of such texts is a very challenging...
http://arxiv.org/pdf/2311.08349v1
2023-11-14T17:48:19
2023-11-14T17:48:19
{ "title": [ -0.023452759, -0.007865906, 0.00037693977, -0.034423828, -0.005748749, 0.01626587, -0.02381897, -0.01890564, -0.01737976, 0.042633057, -0.044647217, 0.009208679, 0.005104065, 0.0029754639, -0.045776367, -0.01448822, 0.0019741058, 0.0...
http://arxiv.org/abs/2311.08348v1
MC^2: A Multilingual Corpus of Minority Languages in China
Chen Zhang, Mingxu Tao, Quzhe Huang, Jiuheng Lin, Zhibin Chen, Yansong Feng
cs.CL
Large-scale corpora play a vital role in the construction of large language models (LLMs). However, existing LLMs exhibit limited abilities in understanding low-resource languages, including the minority languages in China, due to a lack of training data. To improve the accessibility of these languages, we present MC^2...
http://arxiv.org/pdf/2311.08348v1
2023-11-14T17:45:50
2023-11-14T17:45:50
{ "title": [ -0.03262329, 0.0061302185, -0.05078125, -0.021911621, -0.0005683899, -0.020217896, -0.028717041, -0.008613586, -0.018188477, 0.039001465, -0.0063476562, 0.010124207, 0.03756714, -0.048828125, -0.009513855, 0.0076293945, -0.03353882, ...
http://arxiv.org/abs/2311.08329v2
KTRL+F: Knowledge-Augmented In-Document Search
Hanseok Oh, Haebin Shin, Miyoung Ko, Hyunji Lee, Minjoon Seo
cs.CL
We introduce a new problem KTRL+F, a knowledge-augmented in-document search task that necessitates real-time identification of all semantic targets within a document with the awareness of external sources through a single natural query. This task addresses following unique challenges for in-document search: 1) utilizin...
http://arxiv.org/pdf/2311.08329v2
2023-11-15T17:36:55
2023-11-14T17:18:08
{ "title": [ -0.022567749, -0.025314331, -0.036254883, -0.057556152, -0.0057525635, 0.0053634644, -0.031204224, -0.0021858215, 0.024612427, 0.032989502, -0.03366089, -0.0016937256, 0.02015686, -0.0056419373, -0.03945923, 0.0015630722, -0.000685215, ...
http://arxiv.org/abs/2311.08324v1
Anti-LM Decoding for Zero-shot In-context Machine Translation
Suzanna Sia, Alexandra DeLucia, Kevin Duh
cs.CL, cs.AI
Zero-shot In-context learning is the phenomenon where models can perform the task simply given the instructions. However, pre-trained large language models are known to be poorly calibrated for this task. One of the most effective approaches to handling this bias is to adopt a contrastive decoding objective, which acco...
http://arxiv.org/pdf/2311.08324v1
2023-11-14T17:09:43
2023-11-14T17:09:43
{ "title": [ -0.019515991, -0.02053833, -0.010177612, -0.03744507, -0.02468872, 0.004016876, -0.023223877, 0.01574707, 0.004840851, 0.06915283, -0.009468079, 0.012580872, 0.05340576, -0.016677856, -0.03781128, 0.04486084, 0.011886597, 0.02444458,...
http://arxiv.org/abs/2311.08323v1
Open-vocabulary keyword spotting in any language through multilingual contrastive speech-phoneme pretraining
Jian Zhu, Farhan Samir, Changbing Yang, Jahurul Islam
cs.CL, cs.SD, eess.AS
In this paper, we introduce a massively multilingual speech corpora with fine-grained phonemic transcriptions, encompassing more than 115 languages from diverse language families. Based on this multilingual dataset, we propose CLAP-IPA, a multilingual phoneme-speech contrastive embedding model capable of open-vocabular...
http://arxiv.org/pdf/2311.08323v1
2023-11-14T17:09:07
2023-11-14T17:09:07
{ "title": [ 0.01939392, 0.007381439, -0.008583069, -0.030471802, -0.034179688, -0.019836426, -0.03213501, -0.023712158, -0.006023407, 0.037902832, -0.04031372, 0.025909424, 0.011360168, -0.02748108, 0.00687027, -0.026184082, -0.0011806488, 0.032...
http://arxiv.org/abs/2311.08306v1
On-the-Fly Fusion of Large Language Models and Machine Translation
Hieu Hoang, Huda Khayrallah, Marcin Junczys-Dowmunt
cs.CL
We propose the on-the-fly ensembling of a machine translation model with an LLM, prompted on the same task and input. We perform experiments on 4 language pairs (both directions) with varying data amounts. We find that a slightly weaker-at-translation LLM can improve translations of a NMT model, and ensembling with an ...
http://arxiv.org/pdf/2311.08306v1
2023-11-14T16:49:33
2023-11-14T16:49:33
{ "title": [ -0.03149414, -0.01121521, -0.026824951, -0.06744385, -0.00047802925, 0.015670776, -0.0546875, 0.013298035, 0.007183075, 0.04598999, -0.013832092, -0.0031337738, 0.028701782, -0.012275696, -0.037841797, 0.017837524, -0.0024814606, 0.0...
http://arxiv.org/abs/2311.08303v1
Extrinsically-Focused Evaluation of Omissions in Medical Summarization
Elliot Schumacher, Daniel Rosenthal, Varun Nair, Luladay Price, Geoffrey Tso, Anitha Kannan
cs.CL, cs.AI
The goal of automated summarization techniques (Paice, 1990; Kupiec et al, 1995) is to condense text by focusing on the most critical information. Generative large language models (LLMs) have shown to be robust summarizers, yet traditional metrics struggle to capture resulting performance (Goyal et al, 2022) in more po...
http://arxiv.org/pdf/2311.08303v1
2023-11-14T16:46:15
2023-11-14T16:46:15
{ "title": [ -0.0018453598, 0.0104904175, -0.037353516, -0.06329346, 0.0041999817, -0.0014009476, -0.037261963, -0.017059326, -0.0104599, 0.050842285, -0.011489868, 0.027191162, 0.0072898865, 0.00315094, -0.044952393, 0.007434845, -0.009742737, 0...
http://arxiv.org/abs/2311.08300v1
Workflow-Guided Response Generation for Task-Oriented Dialogue
Do June Min, Paloma Sodhi, Ramya Ramakrishnan
cs.CL, cs.AI
Task-oriented dialogue (TOD) systems aim to achieve specific goals through interactive dialogue. Such tasks usually involve following specific workflows, i.e. executing a sequence of actions in a particular order. While prior work has focused on supervised learning methods to condition on past actions, they do not expl...
http://arxiv.org/pdf/2311.08300v1
2023-11-14T16:44:33
2023-11-14T16:44:33
{ "title": [ -0.032196045, -0.022964478, -0.06890869, -0.08605957, -0.0012598038, -0.0019302368, -0.06542969, 0.05505371, -0.00415802, 0.050628662, 0.01864624, 0.004295349, -0.003036499, -0.025527954, -0.061645508, 0.004962921, 0.003293991, 0.014...
http://arxiv.org/abs/2311.08299v1
VERVE: Template-based ReflectiVE Rewriting for MotiVational IntErviewing
Do June Min, Verónica Pérez-Rosas, Kenneth Resnicow, Rada Mihalcea
cs.CL, cs.AI
Reflective listening is a fundamental skill that counselors must acquire to achieve proficiency in motivational interviewing (MI). It involves responding in a manner that acknowledges and explores the meaning of what the client has expressed in the conversation. In this work, we introduce the task of counseling respons...
http://arxiv.org/pdf/2311.08299v1
2023-11-14T16:44:16
2023-11-14T16:44:16
{ "title": [ -0.021316528, -0.0031833649, -0.03302002, -0.072143555, -0.021942139, 0.023269653, -0.086242676, -0.0146484375, -0.012023926, 0.01247406, -0.009651184, 0.008560181, -0.013015747, -0.048919678, -0.017669678, -0.020721436, -0.0039520264, ...
http://arxiv.org/abs/2311.08298v1
A Survey of Language Model Confidence Estimation and Calibration
Jiahui Geng, Fengyu Cai, Yuxia Wang, Heinz Koeppl, Preslav Nakov, Iryna Gurevych
cs.CL, cs.AI
Language models (LMs) have demonstrated remarkable capabilities across a wide range of tasks in various domains. Despite their impressive performance, the reliability of their output is concerning and questionable regarding the demand for AI safety. Assessing the confidence of LM predictions and calibrating them across...
http://arxiv.org/pdf/2311.08298v1
2023-11-14T16:43:29
2023-11-14T16:43:29
{ "title": [ -0.062286377, 0.00006723404, -0.048339844, -0.043792725, -0.01121521, 0.007293701, -0.08282471, -0.020950317, 0.00920105, 0.021942139, -0.02999878, -0.00617218, -0.005264282, -0.029571533, -0.054901123, 0.007987976, -0.019805908, 0.0...
http://arxiv.org/abs/2311.08287v1
How Well Do Large Language Models Understand Syntax? An Evaluation by Asking Natural Language Questions
Houquan Zhou, Yang Hou, Zhenghua Li, Xuebin Wang, Zhefeng Wang, Xinyu Duan, Min Zhang
cs.CL
While recent advancements in large language models (LLMs) bring us closer to achieving artificial general intelligence, the question persists: Do LLMs truly understand language, or do they merely mimic comprehension through pattern recognition? This study seeks to explore this question through the lens of syntax, a cru...
http://arxiv.org/pdf/2311.08287v1
2023-11-14T16:30:36
2023-11-14T16:30:36
{ "title": [ -0.032440186, -0.015731812, -0.022979736, -0.06536865, -0.006038666, 0.011169434, -0.064086914, 0.0032234192, 0.01852417, 0.0020160675, -0.038085938, -0.0068473816, -0.011489868, -0.010894775, -0.04949951, -0.0073242188, -0.015060425, ...
http://arxiv.org/abs/2311.08273v1
Examining Modularity in Multilingual LMs via Language-Specialized Subnetworks
Rochelle Choenni, Ekaterina Shutova, Dan Garrette
cs.CL
Recent work has proposed explicitly inducing language-wise modularity in multilingual LMs via sparse fine-tuning (SFT) on per-language subnetworks as a means of better guiding cross-lingual sharing. In this work, we investigate (1) the degree to which language-wise modularity naturally arises within models with no spec...
http://arxiv.org/pdf/2311.08273v1
2023-11-14T16:11:23
2023-11-14T16:11:23
{ "title": [ -0.057434082, -0.013916016, -0.042877197, -0.043792725, -0.023239136, -0.0057411194, -0.05355835, 0.018585205, -0.060333252, 0.05569458, 0.0042495728, 0.039733887, 0.009979248, -0.021377563, -0.016342163, -0.000333786, -0.014221191, ...
http://arxiv.org/abs/2311.08268v1
A Wolf in Sheep's Clothing: Generalized Nested Jailbreak Prompts can Fool Large Language Models Easily
Peng Ding, Jun Kuang, Dan Ma, Xuezhi Cao, Yunsen Xian, Jiajun Chen, Shujian Huang
cs.CL
Large Language Models (LLMs), such as ChatGPT and GPT-4, are designed to provide useful and safe responses. However, adversarial prompts known as 'jailbreaks' can circumvent safeguards, leading LLMs to generate harmful content. Exploring jailbreak prompts can help to better reveal the weaknesses of LLMs and further ste...
http://arxiv.org/pdf/2311.08268v1
2023-11-14T16:02:16
2023-11-14T16:02:16
{ "title": [ -0.015625, -0.0021705627, -0.030273438, -0.021057129, -0.021942139, -0.021743774, -0.027801514, -0.007835388, -0.04763794, 0.0072021484, -0.016662598, 0.008323669, -0.0048103333, -0.042114258, 0.00983429, -0.024765015, -0.01713562, 0...
http://arxiv.org/abs/2311.08263v1
Fast Chain-of-Thought: A Glance of Future from Parallel Decoding Leads to Answers Faster
Hongxuan Zhang, Zhining Liu, Jiaqi Zheng, Chenyi Zhuang, Jinjie Gu, Guihai Chen
cs.CL
In this work, we propose FastCoT, a model-agnostic framework based on parallel decoding without any further training of an auxiliary model or modification to the LLM itself. FastCoT uses a size-varying context window whose size changes with position to conduct parallel decoding and auto-regressive decoding simultaneous...
http://arxiv.org/pdf/2311.08263v1
2023-11-14T15:56:18
2023-11-14T15:56:18
{ "title": [ 0.017196655, -0.03074646, -0.024475098, -0.055511475, -0.0435791, -0.022354126, -0.014823914, 0.0065574646, 0.010787964, 0.03488159, -0.008720398, 0.008110046, 0.031280518, -0.0435791, -0.018508911, 0.00013291836, 0.028137207, 0.0204...
http://arxiv.org/abs/2311.08252v1
REST: Retrieval-Based Speculative Decoding
Zhenyu He, Zexuan Zhong, Tianle Cai, Jason D Lee, Di He
cs.CL, cs.AI, cs.IR, cs.LG
We introduce Retrieval-Based Speculative Decoding (REST), a novel algorithm designed to speed up language model generation. The key insight driving the development of REST is the observation that the process of text generation often includes certain common phases and patterns. Unlike previous methods that rely on a dra...
http://arxiv.org/pdf/2311.08252v1
2023-11-14T15:43:47
2023-11-14T15:43:47
{ "title": [ -0.045806885, -0.020370483, -0.030136108, -0.06616211, -0.012779236, 0.024261475, -0.044647217, 0.009246826, 0.004096985, 0.06311035, -0.0071983337, -0.027770996, 0.019104004, -0.0062294006, -0.07885742, 0.010429382, 0.004020691, 0.0...
http://arxiv.org/abs/2311.08249v1
On Using Distribution-Based Compositionality Assessment to Evaluate Compositional Generalisation in Machine Translation
Anssi Moisio, Mathias Creutz, Mikko Kurimo
cs.CL
Compositional generalisation (CG), in NLP and in machine learning more generally, has been assessed mostly using artificial datasets. It is important to develop benchmarks to assess CG also in real-world natural language tasks in order to understand the abilities and limitations of systems deployed in the wild. To this...
http://arxiv.org/pdf/2311.08249v1
2023-11-14T15:37:19
2023-11-14T15:37:19
{ "title": [ -0.023025513, -0.011161804, -0.008377075, -0.0637207, -0.013664246, 0.02583313, -0.06036377, 0.0055351257, 0.0007581711, 0.05770874, -0.0011634827, 0.005748749, 0.021652222, -0.017425537, -0.0814209, 0.0062828064, -0.032318115, 0.067...
http://arxiv.org/abs/2311.08240v1
Investigating the Encoding of Words in BERT's Neurons using Feature Textualization
Tanja Baeumel, Soniya Vijayakumar, Josef van Genabith, Guenter Neumann, Simon Ostermann
cs.CL, cs.AI
Pretrained language models (PLMs) form the basis of most state-of-the-art NLP technologies. Nevertheless, they are essentially black boxes: Humans do not have a clear understanding of what knowledge is encoded in different parts of the models, especially in individual neurons. The situation is different in computer vis...
http://arxiv.org/pdf/2311.08240v1
2023-11-14T15:21:49
2023-11-14T15:21:49
{ "title": [ -0.022338867, -0.0074539185, -0.021636963, 0.004508972, -0.037231445, -0.0102005005, -0.017547607, -0.031097412, -0.009628296, 0.06161499, 0.004142761, -0.026947021, 0.034851074, -0.03543091, -0.047668457, 0.015464783, -0.0014705658, ...
http://arxiv.org/abs/2311.08219v1
Eval-GCSC: A New Metric for Evaluating ChatGPT's Performance in Chinese Spelling Correction
Kunting Li, Yong Hu, Shaolei Wang, Hanhan Ma, Liang He, Fandong Meng, Jie Zhou
cs.CL, cs.AI
ChatGPT has demonstrated impressive performance in various downstream tasks. However, in the Chinese Spelling Correction (CSC) task, we observe a discrepancy: while ChatGPT performs well under human evaluation, it scores poorly according to traditional metrics. We believe this inconsistency arises because the tradition...
http://arxiv.org/pdf/2311.08219v1
2023-11-14T14:56:33
2023-11-14T14:56:33
{ "title": [ -0.022766113, -0.008087158, 0.0012617111, 0.017318726, -0.019226074, 0.00061273575, 0.0039634705, -0.030151367, -0.006023407, 0.03756714, -0.024398804, -0.006668091, 0.022659302, -0.020401001, -0.03869629, -0.006362915, -0.04660034, ...
http://arxiv.org/abs/2311.08213v1
Unlock the Power: Competitive Distillation for Multi-Modal Large Language Models
Xinwei Li, Li Lin, Shuai Wang, Chen Qian
cs.CV, cs.CL
Recently, multi-modal content generation has attracted lots of attention from researchers by investigating the utilization of visual instruction tuning based on large language models (LLMs). To enhance the performance and generalization ability of such LLMs, the practice of distilling knowledge from pretrained multi-mo...
http://arxiv.org/pdf/2311.08213v1
2023-11-14T14:49:46
2023-11-14T14:49:46
{ "title": [ -0.014350891, -0.0029449463, -0.04522705, -0.07788086, 0.006061554, 0.0007324219, -0.08618164, 0.009292603, 0.013313293, 0.031173706, -0.015701294, -0.009597778, 0.027114868, -0.0236969, -0.038116455, 0.0029754639, -0.010688782, 0.02...
http://arxiv.org/abs/2311.08206v1
Human-Centric Autonomous Systems With LLMs for User Command Reasoning
Yi Yang, Qingwen Zhang, Ci Li, Daniel Simões Marta, Nazre Batool, John Folkesson
cs.CL, cs.AI, cs.RO
The evolution of autonomous driving has made remarkable advancements in recent years, evolving into a tangible reality. However, a human-centric large-scale adoption hinges on meeting a variety of multifaceted requirements. To ensure that the autonomous system meets the user's intent, it is essential to accurately disc...
http://arxiv.org/pdf/2311.08206v1
2023-11-14T14:42:28
2023-11-14T14:42:28
{ "title": [ -0.030685425, 0.002325058, -0.04660034, -0.051727295, 0.005962372, -0.015899658, -0.037872314, 0.0051002502, -0.02494812, 0.026748657, -0.04159546, 0.014053345, -0.015388489, -0.014709473, -0.022491455, -0.020309448, -0.0051574707, -...
http://arxiv.org/abs/2311.08195v1
Automated Fact-Checking in Dialogue: Are Specialized Models Needed?
Eric Chamoun, Marzieh Saeidi, Andreas Vlachos
cs.CL, cs.AI
Prior research has shown that typical fact-checking models for stand-alone claims struggle with claims made in dialogues. As a solution, fine-tuning these models on labelled dialogue data has been proposed. However, creating separate models for each use case is impractical, and we show that fine-tuning models for dialo...
http://arxiv.org/pdf/2311.08195v1
2023-11-14T14:29:00
2023-11-14T14:29:00
{ "title": [ -0.008094788, -0.009231567, -0.024139404, -0.07348633, -0.0011081696, -0.00017821789, -0.037231445, -0.0024108887, -0.030181885, 0.03466797, -0.0025691986, 0.009841919, 0.003364563, -0.022384644, -0.047943115, -0.0046920776, -0.04095459,...
http://arxiv.org/abs/2311.08191v1
GEC-DePenD: Non-Autoregressive Grammatical Error Correction with Decoupled Permutation and Decoding
Konstantin Yakovlev, Alexander Podolskiy, Andrey Bout, Sergey Nikolenko, Irina Piontkovskaya
cs.CL
Grammatical error correction (GEC) is an important NLP task that is currently usually solved with autoregressive sequence-to-sequence models. However, approaches of this class are inherently slow due to one-by-one token generation, so non-autoregressive alternatives are needed. In this work, we propose a novel non-auto...
http://arxiv.org/pdf/2311.08191v1
2023-11-14T14:24:36
2023-11-14T14:24:36
{ "title": [ -0.020248413, -0.017211914, -0.03744507, -0.058258057, -0.024795532, 0.0005002022, -0.070129395, 0.00598526, -0.0062332153, 0.03488159, -0.033477783, 0.011802673, 0.031829834, -0.01725769, -0.039886475, 0.0050964355, -0.037994385, 0....
http://arxiv.org/abs/2311.08189v2
Unlocking Science: Novel Dataset and Benchmark for Cross-Modality Scientific Information Extraction
Yuhan Li, Jian Wu, Zhiwei Yu, Börje F. Karlsson, Wei Shen, Manabu Okumura, Chin-Yew Lin
cs.CL
Extracting key information from scientific papers has the potential to help researchers work more efficiently and accelerate the pace of scientific progress. Over the last few years, research on Scientific Information Extraction (SciIE) witnessed the release of several new systems and benchmarks. However, existing pape...
http://arxiv.org/pdf/2311.08189v2
2023-11-15T02:57:01
2023-11-14T14:22:47
{ "title": [ -0.048034668, 0.002855301, -0.066467285, -0.036590576, 0.0005235672, -0.012138367, -0.033813477, -0.014839172, 0.021057129, 0.03289795, -0.0024337769, 0.012367249, -0.0051651, -0.012557983, -0.045715332, 0.020935059, -0.012138367, 0....
http://arxiv.org/abs/2311.08182v1
Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning
Shengguang Wu, Keming Lu, Benfeng Xu, Junyang Lin, Qi Su, Chang Zhou
cs.CL, cs.LG
Enhancing the instruction-following ability of Large Language Models (LLMs) primarily demands substantial instruction-tuning datasets. However, the sheer volume of these imposes a considerable computational burden and annotation cost. To investigate a label-efficient instruction tuning method that allows the model itse...
http://arxiv.org/pdf/2311.08182v1
2023-11-14T14:10:40
2023-11-14T14:10:40
{ "title": [ -0.055847168, -0.0032920837, -0.018295288, -0.040283203, -0.0047073364, 0.0062294006, -0.012290955, 0.012191772, 0.0032997131, 0.030380249, -0.03152466, 0.03274536, -0.045837402, -0.0052490234, -0.052520752, 0.00217247, 0.013244629, ...
http://arxiv.org/abs/2311.08166v1
MechAgents: Large language model multi-agent collaborations can solve mechanics problems, generate new data, and integrate knowledge
Bo Ni, Markus J. Buehler
cs.AI, cond-mat.dis-nn, cond-mat.mtrl-sci, cs.CL, cs.LG
Solving mechanics problems using numerical methods requires comprehensive intelligent capability of retrieving relevant knowledge and theory, constructing and executing codes, analyzing the results, a task that has thus far mainly been reserved for humans. While emerging AI methods can provide effective approaches to s...
http://arxiv.org/pdf/2311.08166v1
2023-11-14T13:49:03
2023-11-14T13:49:03
{ "title": [ -0.042510986, -0.027755737, -0.03277588, -0.009857178, -0.042938232, -0.008575439, -0.0390625, 0.012428284, -0.0073890686, 0.015731812, -0.012054443, 0.021514893, -0.02357483, -0.02558899, 0.0035934448, -0.021255493, 0.0070381165, -0...
http://arxiv.org/abs/2311.08154v1
Ask One More Time: Self-Agreement Improves Reasoning of Language Models in (Almost) All Scenarios
Lei Lin, Jiayi Fu, Pengli Liu, Junchen Wan, Fuzheng Zhang, Zhongyuan Wang, Di Zhang, Kun Gai
cs.CL, cs.AI
Although chain-of-thought (CoT) prompting combined with language models has achieved encouraging results on complex reasoning tasks, the naive greedy decoding used in CoT prompting usually causes the repetitiveness and local optimality. To address this shortcoming, ensemble-optimization tries to obtain multiple reasoni...
http://arxiv.org/pdf/2311.08154v1
2023-11-14T13:30:54
2023-11-14T13:30:54
{ "title": [ -0.017959595, -0.019439697, -0.051452637, -0.070007324, -0.016571045, -0.015365601, -0.071899414, 0.016738892, -0.0066108704, 0.0287323, 0.0018854141, 0.0021896362, 0.0039634705, -0.030075073, -0.040985107, 0.015083313, -0.002922058, ...
http://arxiv.org/abs/2311.08152v1
Towards Reasoning in Large Language Models via Multi-Agent Peer Review Collaboration
Zhenran Xu, Senbao Shi, Baotian Hu, Jindi Yu, Dongfang Li, Min Zhang, Yuxiang Wu
cs.CL
Large Language Models (LLMs) have shown remarkable capabilities in general natural language processing tasks but often fall short in complex reasoning tasks. Recent studies have explored human-like problem-solving strategies, such as self-correct, to push further the boundary of single-model reasoning ability. In this ...
http://arxiv.org/pdf/2311.08152v1
2023-11-14T13:27:07
2023-11-14T13:27:07
{ "title": [ -0.058410645, -0.031707764, -0.03555298, -0.062347412, -0.031341553, 0.021362305, -0.0725708, 0.02961731, 0.009750366, 0.031799316, -0.024154663, -0.009819031, 0.017532349, -0.014549255, -0.03781128, -0.0006299019, -0.02923584, 0.050...
http://arxiv.org/abs/2311.08147v1
RECALL: A Benchmark for LLMs Robustness against External Counterfactual Knowledge
Yi Liu, Lianzhe Huang, Shicheng Li, Sishuo Chen, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun
cs.CL, cs.AI
LLMs and AI chatbots have improved people's efficiency in various fields. However, the necessary knowledge for answering the question may be beyond the models' knowledge boundaries. To mitigate this issue, many researchers try to introduce external knowledge, such as knowledge graphs and Internet contents, into LLMs fo...
http://arxiv.org/pdf/2311.08147v1
2023-11-14T13:24:19
2023-11-14T13:24:19
{ "title": [ -0.029525757, 0.016479492, -0.052337646, -0.03555298, -0.0031356812, -0.01096344, -0.061279297, -0.030944824, 0.023788452, 0.01398468, -0.057250977, 0.025680542, -0.016662598, -0.008598328, -0.03982544, -0.012527466, 0.0010614395, 0....
http://arxiv.org/abs/2311.08143v1
Sinkhorn Transformations for Single-Query Postprocessing in Text-Video Retrieval
Konstantin Yakovlev, Gregory Polyakov, Ilseyar Alimova, Alexander Podolskiy, Andrey Bout, Sergey Nikolenko, Irina Piontkovskaya
cs.CL
A recent trend in multimodal retrieval is related to postprocessing test set results via the dual-softmax loss (DSL). While this approach can bring significant improvements, it usually presumes that an entire matrix of test samples is available as DSL input. This work introduces a new postprocessing approach based on S...
http://arxiv.org/pdf/2311.08143v1
2023-11-14T13:20:23
2023-11-14T13:20:23
{ "title": [ 0.02243042, 0.024749756, -0.029525757, -0.036499023, 0.04071045, -0.022888184, 0.021026611, -0.021835327, 0.037719727, 0.07122803, -0.05215454, 0.018737793, 0.03765869, -0.00026893616, -0.0592041, 0.01838684, -0.0054512024, 0.0212554...
http://arxiv.org/abs/2311.08123v1
Memory-efficient Stochastic methods for Memory-based Transformers
Vishwajit Kumar Vishnu, C. Chandra Sekhar
cs.LG, cs.CL
Training Memory-based transformers can require a large amount of memory and can be quite inefficient. We propose a novel two-phase training mechanism and a novel regularization technique to improve the training efficiency of memory-based transformers, which are often used for long-range context problems. For our experi...
http://arxiv.org/pdf/2311.08123v1
2023-11-14T12:37:25
2023-11-14T12:37:25
{ "title": [ -0.012290955, 0.014198303, -0.023544312, -0.086242676, 0.034454346, 0.016921997, -0.02482605, -0.008354187, -0.009429932, 0.043304443, -0.007095337, 0.013023376, 0.0079193115, -0.017364502, -0.028396606, 0.016921997, 0.0048103333, 0....
http://arxiv.org/abs/2311.08117v1
Insights into Classifying and Mitigating LLMs' Hallucinations
Alessandro Bruno, Pier Luigi Mazzeo, Aladine Chetouani, Marouane Tliba, Mohamed Amine Kerkouri
cs.CL
The widespread adoption of large language models (LLMs) across diverse AI applications is proof of the outstanding achievements obtained in several tasks, such as text mining, text generation, and question answering. However, LLMs are not exempt from drawbacks. One of the most concerning aspects regards the emerging pr...
http://arxiv.org/pdf/2311.08117v1
2023-11-14T12:30:28
2023-11-14T12:30:28
{ "title": [ 0.024871826, 0.015052795, -0.019989014, -0.016967773, -0.030807495, -0.037384033, -0.0143585205, -0.03805542, -0.031799316, 0.02267456, 0.00390625, 0.024307251, 0.004337311, -0.05050659, 0.0357666, -0.004875183, 0.0121536255, 0.03170...
http://arxiv.org/abs/2311.08110v1
Improving hateful memes detection via learning hatefulness-aware embedding space through retrieval-guided contrastive learning
Jingbiao Mei, Jinghong Chen, Weizhe Lin, Bill Byrne, Marcus Tomalin
cs.CL, cs.CV
Hateful memes have emerged as a significant concern on the Internet. These memes, which are a combination of image and text, often convey messages vastly different from their individual meanings. Thus, detecting hateful memes requires the system to jointly understand the visual and textual modalities. However, our inve...
http://arxiv.org/pdf/2311.08110v1
2023-11-14T12:14:54
2023-11-14T12:14:54
{ "title": [ 0.07330322, 0.011993408, -0.011062622, -0.00082445145, 0.005252838, -0.050628662, 0.04736328, -0.041229248, 0.03552246, 0.041625977, -0.046844482, 0.02432251, 0.027145386, -0.021224976, -0.04043579, 0.009918213, -0.011581421, 0.04516...
http://arxiv.org/abs/2311.08107v1
SAIE Framework: Support Alone Isn't Enough -- Advancing LLM Training with Adversarial Remarks
Mengsay Loem, Masahiro Kaneko, Naoaki Okazaki
cs.CL
Large Language Models (LLMs) can justify or criticize their predictions through discussion with other models or humans, thereby enhancing their intrinsic understanding of instances. While proactive discussions enhance performance, this approach is currently limited to the inference phase. In this context, we posit a hy...
http://arxiv.org/pdf/2311.08107v1
2023-11-14T12:12:25
2023-11-14T12:12:25
{ "title": [ -0.018173218, -0.010414124, -0.025390625, -0.037109375, -0.03665161, 0.009902954, -0.052459717, -0.007633209, -0.03753662, 0.012748718, -0.012245178, 0.03277588, -0.030136108, -0.04437256, 0.030776978, -0.030670166, -0.0052261353, 0....
http://arxiv.org/abs/2311.08106v1
Carpe Diem: On the Evaluation of World Knowledge in Lifelong Language Models
Yujin Kim, Jaehong Yoon, Seonghyeon Ye, Sung Ju Hwang, Se-young Yun
cs.CL
In an ever-evolving world, the dynamic nature of knowledge presents challenges for language models that are trained on static data, leading to outdated encoded information. However, real-world scenarios require models not only to acquire new knowledge but also to overwrite outdated information into updated ones. To add...
http://arxiv.org/pdf/2311.08106v1
2023-11-14T12:12:02
2023-11-14T12:12:02
{ "title": [ -0.039642334, -0.0018596649, -0.056427002, -0.041809082, -0.009124756, -0.022537231, -0.025604248, 0.0015659332, 0.011756897, 0.03289795, -0.042114258, 0.025268555, -0.0042533875, -0.005607605, -0.053497314, 0.006072998, -0.015548706, ...
http://arxiv.org/abs/2311.08105v1
DiLoCo: Distributed Low-Communication Training of Language Models
Arthur Douillard, Qixuan Feng, Andrei A. Rusu, Rachita Chhaparia, Yani Donchev, Adhiguna Kuncoro, Marc'Aurelio Ranzato, Arthur Szlam, Jiajun Shen
cs.LG, cs.CL
Large language models (LLM) have become a critical component in many applications of machine learning. However, standard approaches to training LLM require a large number of tightly interconnected accelerators, with devices exchanging gradients and other intermediate states at each optimization step. While it is diffic...
http://arxiv.org/pdf/2311.08105v1
2023-11-14T12:05:45
2023-11-14T12:05:45
{ "title": [ -0.04989624, -0.0064697266, -0.05783081, -0.06201172, 0.003250122, 0.00064373016, -0.059448242, 0.011154175, 0.009811401, 0.030853271, -0.019439697, -0.010261536, 0.021621704, -0.02760315, -0.0368042, 0.004558563, -0.018081665, 0.054...
http://arxiv.org/abs/2311.08103v1
Exploring Semi-supervised Hierarchical Stacked Encoder for Legal Judgement Prediction
Nishchal Prasad, Mohand Boughanem, Taoufiq Dkaki
cs.CL, cs.AI, cs.IR
Predicting the judgment of a legal case from its unannotated case facts is a challenging task. The lengthy and non-uniform document structure poses an even greater challenge in extracting information for decision prediction. In this work, we explore and propose a two-level classification mechanism; both supervised and ...
http://arxiv.org/pdf/2311.08103v1
2023-11-14T12:03:26
2023-11-14T12:03:26
{ "title": [ -0.016189575, -0.0019741058, -0.047058105, -0.0657959, -0.002538681, -0.011505127, -0.043060303, 0.023971558, 0.03942871, 0.029678345, -0.036621094, -0.012825012, 0.040039062, -0.012245178, -0.051879883, 0.009864807, -0.015098572, 0....
http://arxiv.org/abs/2311.08097v1
Empowering Multi-step Reasoning across Languages via Tree-of-Thoughts
Leonardo Ranaldi, Fabio Massimo Zanzotto
cs.CL, cs.AI
Chain-of-Thought (CoT) prompting empowers the reasoning abilities of Large Language Models (LLMs), eliciting them to solve complex reasoning tasks step-by-step. However, with the success of CoT methods, the ability to deliver multi-step reasoning remains limited to English due to the imbalance in the distribution of th...
http://arxiv.org/pdf/2311.08097v1
2023-11-14T11:49:43
2023-11-14T11:49:43
{ "title": [ -0.013389587, -0.052001953, -0.025878906, -0.09649658, -0.024429321, 0.0262146, -0.06591797, 0.03878784, 0.014450073, 0.008872986, -0.024627686, -0.0025634766, -0.007675171, -0.02935791, -0.023544312, 0.004020691, 0.0099105835, 0.002...
http://arxiv.org/abs/2311.08093v1
Spot: A Natural Language Interface for Geospatial Searches in OSM
Lynn Khellaf, Ipek Baris Schlicht, Julia Bayer, Ruben Bouwmeester, Tilman Miraß, Tilman Wagner
cs.CL, cs.AI
Investigative journalists and fact-checkers have found OpenStreetMap (OSM) to be an invaluable resource for their work due to its extensive coverage and intricate details of various locations, which play a crucial role in investigating news scenes. Despite its value, OSM's complexity presents considerable accessibility...
http://arxiv.org/pdf/2311.08093v1
2023-11-14T11:35:09
2023-11-14T11:35:09
{ "title": [ -0.0077171326, -0.027679443, 0.002418518, -0.025268555, -0.002483368, 0.0053710938, -0.013549805, -0.039245605, 0.020584106, 0.027801514, -0.021728516, -0.02268982, 0.0028877258, -0.03152466, -0.04244995, -0.00036263466, 0.0018806458, ...
http://arxiv.org/abs/2311.08089v1
Align after Pre-train: Improving Multilingual Generative Models with Cross-lingual Alignment
Chong Li, Shaonan Wang, Jiajun Zhang, Chengqing Zong
cs.CL
Multilingual generative models obtain remarkable cross-lingual capabilities through pre-training on large-scale corpora. However, they still exhibit a performance bias toward high-resource languages, and learn isolated distributions of sentence representations across languages. To bridge this gap, we propose a simple y...
http://arxiv.org/pdf/2311.08089v1
2023-11-14T11:24:08
2023-11-14T11:24:08
{ "title": [ -0.026504517, -0.0050239563, -0.033172607, -0.074279785, 0.00047373772, 0.010353088, -0.085754395, -0.005622864, 0.014450073, 0.037963867, 0.003944397, 0.015083313, 0.008308411, -0.020065308, -0.03543091, 0.020553589, -0.0064697266, ...
http://arxiv.org/abs/2311.08057v1
Data and models for stance and premise detection in COVID-19 tweets: insights from the Social Media Mining for Health (SMM4H) 2022 shared task
Vera Davydova, Huabin Yang, Elena Tutubalina
cs.CL, cs.SI, I.2.7; J.3
The COVID-19 pandemic has sparked numerous discussions on social media platforms, with users sharing their views on topics such as mask-wearing and vaccination. To facilitate the evaluation of neural models for stance detection and premise classification, we organized the Social Media Mining for Health (SMM4H) 2022 Sha...
http://arxiv.org/pdf/2311.08057v1
2023-11-14T10:30:49
2023-11-14T10:30:49
{ "title": [ 0.028762817, 0.006378174, -0.018432617, -0.050689697, -0.011413574, -0.012145996, -0.06365967, -0.01309967, 0.00017464161, 0.029907227, -0.061340332, 0.0340271, -0.010505676, -0.00021314621, -0.0058135986, -0.033477783, 0.0007920265, ...
http://arxiv.org/abs/2311.08045v1
Adversarial Preference Optimization
Pengyu Cheng, Yifan Yang, Jian Li, Yong Dai, Nan Du
cs.CL, cs.AI, cs.LG
Human preference alignment is a crucial training step to improve the interaction quality of large language models (LLMs). Existing aligning methods depend on manually annotated preference data to guide the LLM optimization directions. However, in practice, continuously updating LLMs raises a distribution gap between mo...
http://arxiv.org/pdf/2311.08045v1
2023-11-14T10:10:31
2023-11-14T10:10:31
{ "title": [ -0.012054443, 0.016113281, -0.015945435, -0.03062439, 0.015068054, -0.0007209778, 0.0017347336, -0.04043579, -0.040283203, 0.01776123, 0.00083732605, 0.009887695, -0.047790527, -0.027862549, -0.010177612, -0.008224487, -0.0064086914, ...
http://arxiv.org/abs/2311.08011v1
Forgetting before Learning: Utilizing Parametric Arithmetic for Knowledge Updating in Large Language Models
Shiwen Ni, Dingwei Chen, Chengming Li, Xiping Hu, Ruifeng Xu, Min Yang
cs.CL
Recently Large Language Models (LLMs) have demonstrated their amazing text understanding and generation capabilities. However, even stronger LLMs may still learn incorrect knowledge from the training corpus, as well as some knowledge that is outdated over time. Direct secondary fine-tuning with data containing new know...
http://arxiv.org/pdf/2311.08011v1
2023-11-14T09:12:40
2023-11-14T09:12:40
{ "title": [ -0.025894165, -0.019836426, -0.03164673, -0.07672119, -0.014381409, -0.0026550293, -0.062805176, 0.012611389, 0.013969421, 0.034362793, -0.046051025, -0.015151978, 0.030258179, -0.0016593933, -0.04751587, -0.016113281, 0.011550903, 0...
http://arxiv.org/abs/2311.08010v1
Distantly-Supervised Named Entity Recognition with Uncertainty-aware Teacher Learning and Student-student Collaborative Learning
Helan Hu, Shuzheng Si, Haozhe Zhao, Shuang Zeng, Kaikai An, Zefan Cai, Baobao Chang
cs.CL, cs.AI
Distantly-Supervised Named Entity Recognition (DS-NER) effectively alleviates the burden of annotation, but meanwhile suffers from the label noise. Recent works attempt to adopt the teacher-student framework to gradually refine the training labels and improve the overall robustness. However, we argue that these teacher...
http://arxiv.org/pdf/2311.08010v1
2023-11-14T09:09:58
2023-11-14T09:09:58
{ "title": [ -0.0054779053, 0.017333984, -0.04776001, -0.049713135, 0.022735596, -0.023910522, -0.019302368, -0.00024318695, -0.010871887, -0.005340576, -0.06890869, 0.01889038, -0.018508911, -0.008125305, -0.005168915, -0.050750732, -0.017227173, ...
http://arxiv.org/abs/2311.08002v1
TempTabQA: Temporal Question Answering for Semi-Structured Tables
Vivek Gupta, Pranshu Kandoi, Mahek Bhavesh Vora, Shuo Zhang, Yujie He, Ridho Reinanda, Vivek Srikumar
cs.CL, cs.AI, cs.IR
Semi-structured data, such as Infobox tables, often include temporal information about entities, either implicitly or explicitly. Can current NLP systems reason about such information in semi-structured tables? To tackle this question, we introduce the task of temporal question answering on semi-structured tables. We p...
http://arxiv.org/pdf/2311.08002v1
2023-11-14T08:57:01
2023-11-14T08:57:01
{ "title": [ -0.010047913, -0.014137268, -0.046691895, -0.087402344, 0.014915466, -0.001619339, -0.047973633, 0.030075073, 0.03463745, 0.034729004, -0.027008057, -0.008628845, 0.008354187, 0.0020484924, -0.07745361, -0.0071754456, 0.00039577484, ...
http://arxiv.org/abs/2311.08001v1
A Comparative Analysis of the COVID-19 Infodemic in English and Chinese: Insights from Social Media Textual Data
Jia Luo, Daiyun Peng, Lei Shi, Didier El Baz, Xinran Liu
cs.SI, cs.CL, physics.soc-ph
The COVID-19 infodemic, characterized by the rapid spread of misinformation and unverified claims related to the pandemic, presents a significant challenge. This paper presents a comparative analysis of the COVID-19 infodemic in the English and Chinese languages, utilizing textual data extracted from social media platf...
http://arxiv.org/pdf/2311.08001v1
2023-11-14T08:55:11
2023-11-14T08:55:11
{ "title": [ -0.0018692017, 0.0004274845, -0.04977417, -0.046020508, 0.008491516, -0.025909424, -0.01727295, 0.011436462, -0.054351807, 0.041870117, -0.023712158, 0.03427124, 0.025314331, -0.014823914, 0.0025672913, -0.007888794, 0.00081157684, 0...
http://arxiv.org/abs/2311.07996v1
How Well Do Text Embedding Models Understand Syntax?
Yan Zhang, Zhaopeng Feng, Zhiyang Teng, Zuozhu Liu, Haizhou Li
cs.CL
Text embedding models have significantly contributed to advancements in natural language processing by adeptly capturing semantic properties of textual data. However, the ability of these models to generalize across a wide range of syntactic contexts remains under-explored. In this paper, we first develop an evaluation...
http://arxiv.org/pdf/2311.07996v1
2023-11-14T08:51:00
2023-11-14T08:51:00
{ "title": [ -0.013900757, 0.002922058, -0.0038070679, -0.04269409, -0.036712646, 0.0013380051, -0.06689453, -0.014595032, 0.02810669, 0.0014314651, -0.052215576, -0.016494751, 0.008384705, -0.014076233, -0.023254395, -0.012252808, -0.00919342, 0...
http://arxiv.org/abs/2311.07989v1
A Survey on Language Models for Code
Ziyin Zhang, Chaoyu Chen, Bingchang Liu, Cong Liao, Zi Gong, Hang Yu, Jianguo Li, Rui Wang
cs.CL, cs.AI, cs.SE
In this work we systematically review the recent advancements in code processing with language models, covering 50+ models, 30+ evaluation tasks, and 500 related works. We break down code processing models into general language models represented by the GPT family and specialized models that are specifically pretrained...
http://arxiv.org/pdf/2311.07989v1
2023-11-14T08:34:26
2023-11-14T08:34:26
{ "title": [ -0.050628662, -0.029632568, -0.035614014, -0.0635376, -0.007858276, 0.012077332, -0.07196045, 0.032684326, 0.0079422, 0.035583496, -0.004016876, -0.018356323, 0.00035524368, -0.025421143, -0.057250977, 0.015159607, -0.015777588, 0.02...
http://arxiv.org/abs/2311.07978v1
How good are Large Language Models on African Languages?
Jessica Ojo, Kelechi Ogueji, Pontus Stenetorp, David I. Adelani
cs.CL, cs.AI, cs.LG
Recent advancements in natural language processing have led to the proliferation of large language models (LLMs). These models have been shown to yield good performance, using in-context learning, even on unseen tasks and languages. Additionally, they have been widely adopted as language-model-as-a-service commercial A...
http://arxiv.org/pdf/2311.07978v1
2023-11-14T08:10:14
2023-11-14T08:10:14
{ "title": [ -0.03640747, -0.018966675, 0.006187439, -0.018600464, -0.0769043, 0.027954102, -0.08654785, -0.014373779, -0.01878357, 0.024642944, -0.009925842, -0.009246826, 0.04360962, -0.047454834, 0.019943237, 0.030075073, 0.012435913, 0.022171...
http://arxiv.org/abs/2311.07961v1
The ART of LLM Refinement: Ask, Refine, and Trust
Kumar Shridhar, Koustuv Sinha, Andrew Cohen, Tianlu Wang, Ping Yu, Ram Pasunuru, Mrinmaya Sachan, Jason Weston, Asli Celikyilmaz
cs.CL
In recent years, Large Language Models (LLMs) have demonstrated remarkable generative abilities, but can they judge the quality of their own generations? A popular concept, referred to as self-refinement, postulates that LLMs can detect and correct the errors in their generations when asked to do so. However, recent em...
http://arxiv.org/pdf/2311.07961v1
2023-11-14T07:26:32
2023-11-14T07:26:32
{ "title": [ -0.058410645, -0.012573242, -0.049041748, -0.036132812, -0.026550293, 0.00881958, -0.055725098, -0.004383087, -0.03982544, 0.018844604, -0.03390503, 0.019226074, -0.01158905, -0.0368042, -0.0035953522, -0.020599365, -0.014694214, 0.0...
http://arxiv.org/abs/2311.07954v1
A Closer Look at the Self-Verification Abilities of Large Language Models in Logical Reasoning
Ruixin Hong, Hongming Zhang, Xinyu Pang, Dong Yu, Changshui Zhang
cs.AI, cs.CL
Logical reasoning has been an ongoing pursuit in the field of AI. Despite significant advancements made by large language models (LLMs), they still struggle with complex logical reasoning problems. To enhance reasoning performance, one promising direction is scalable oversight, which requires LLMs to identify their own...
http://arxiv.org/pdf/2311.07954v1
2023-11-14T07:13:10
2023-11-14T07:13:10
{ "title": [ -0.028518677, -0.009002686, -0.05899048, -0.07910156, 0.0056991577, -0.011245728, -0.05606079, 0.020324707, -0.024093628, 0.029632568, -0.014640808, -0.0032081604, 0.0057525635, -0.025634766, -0.02645874, 0.004096985, -0.009468079, 0...
http://arxiv.org/abs/2311.07945v1
First Step Advantage: Importance of Starting Right in Multi-Step Reasoning
Kushal Jain, Kumar Shridhar
cs.CL
Large Language Models (LLMs) can solve complex reasoning tasks by generating rationales for their predictions. Distilling these capabilities into a smaller, compact model can facilitate the creation of specialized, cost-effective models tailored for specific tasks. However, smaller models often face challenges in compl...
http://arxiv.org/pdf/2311.07945v1
2023-11-14T06:45:31
2023-11-14T06:45:31
{ "title": [ -0.0054969788, -0.034942627, -0.06323242, -0.06829834, -0.044036865, -0.02267456, -0.06695557, 0.025772095, -0.025848389, 0.01600647, 0.023071289, -0.0010728836, 0.02758789, -0.061645508, 0.0385437, -0.0024814606, 0.009666443, 0.0200...
http://arxiv.org/abs/2311.07941v1
Non-autoregressive Machine Translation with Probabilistic Context-free Grammar
Shangtong Gui, Chenze Shao, Zhengrui Ma, Xishan Zhang, Yunji Chen, Yang Feng
cs.CL, cs.AI
Non-autoregressive Transformer(NAT) significantly accelerates the inference of neural machine translation. However, conventional NAT models suffer from limited expression power and performance degradation compared to autoregressive (AT) models due to the assumption of conditional independence among target tokens. To ad...
http://arxiv.org/pdf/2311.07941v1
2023-11-14T06:39:04
2023-11-14T06:39:04
{ "title": [ -0.032196045, 0.000049233437, -0.036010742, -0.07727051, 0.00894928, 0.018585205, -0.057678223, 0.013908386, -0.0012693405, 0.04522705, 0.0055236816, 0.0069351196, 0.016159058, -0.0151901245, -0.050323486, 0.016220093, -0.0074920654, ...
http://arxiv.org/abs/2311.07930v1
It's All Relative! -- A Synthetic Query Generation Approach for Improving Zero-Shot Relevance Prediction
Aditi Chaudhary, Karthik Raman, Michael Bendersky
cs.CL
Recent developments in large language models (LLMs) have shown promise in their ability to generate synthetic query-document pairs by prompting with as few as 8 demonstrations. This has enabled building better IR models, especially for tasks with no training data readily available. Typically, such synthetic query gener...
http://arxiv.org/pdf/2311.07930v1
2023-11-14T06:16:49
2023-11-14T06:16:49
{ "title": [ 0.024520874, -0.014755249, -0.028533936, -0.06323242, -0.00050497055, -0.021652222, 0.00069379807, 0.0119018555, 0.011779785, 0.058441162, -0.0074920654, 0.011505127, -0.013389587, -0.008728027, -0.07507324, -0.0037517548, 0.004436493, ...
http://arxiv.org/abs/2311.07925v1
Brain-Driven Representation Learning Based on Diffusion Model
Soowon Kim, Seo-Hyun Lee, Young-Eun Lee, Ji-Won Lee, Ji-Ha Park, Seong-Whan Lee
cs.CL, cs.AI
Interpreting EEG signals linked to spoken language presents a complex challenge, given the data's intricate temporal and spatial attributes, as well as the various noise factors. Denoising diffusion probabilistic models (DDPMs), which have recently gained prominence in diverse areas for their capabilities in representa...
http://arxiv.org/pdf/2311.07925v1
2023-11-14T05:59:58
2023-11-14T05:59:58
{ "title": [ 0.009315491, 0.014465332, -0.035125732, -0.024215698, 0.011421204, -0.022750854, 0.014564514, -0.01335907, 0.012870789, 0.022903442, -0.010047913, -0.010520935, -0.01701355, -0.033050537, -0.04031372, 0.0013074875, 0.0059394836, 0.04...
http://arxiv.org/abs/2311.07919v1
Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models
Yunfei Chu, Jin Xu, Xiaohuan Zhou, Qian Yang, Shiliang Zhang, Zhijie Yan, Chang Zhou, Jingren Zhou
eess.AS, cs.CL, cs.LG
Recently, instruction-following audio-language models have received broad attention for audio interaction with humans. However, the absence of pre-trained audio models capable of handling diverse audio types and tasks has hindered progress in this field. Consequently, most existing works have only been able to support ...
http://arxiv.org/pdf/2311.07919v1
2023-11-14T05:34:50
2023-11-14T05:34:50
{ "title": [ -0.007789612, -0.017333984, -0.029785156, -0.07147217, -0.00077676773, 0.00415802, -0.057769775, 0.007423401, 0.007663727, 0.042144775, -0.013877869, -0.0052261353, 0.040802002, -0.024932861, -0.032409668, 0.018218994, -0.012420654, ...
http://arxiv.org/abs/2311.07918v1
Automated title and abstract screening for scoping reviews using the GPT-4 Large Language Model
David Wilkins
cs.CL
Scoping reviews, a type of literature review, require intensive human effort to screen large numbers of scholarly sources for their relevance to the review objectives. This manuscript introduces GPTscreenR, a package for the R statistical programming language that uses the GPT-4 Large Language Model (LLM) to automatica...
http://arxiv.org/pdf/2311.07918v1
2023-11-14T05:30:43
2023-11-14T05:30:43
{ "title": [ -0.04043579, -0.0069236755, -0.011528015, -0.0357666, -0.024887085, 0.008117676, -0.044769287, -0.03363037, -0.017166138, 0.009941101, -0.008468628, 0.02659607, -0.030471802, -0.026809692, -0.015594482, -0.008102417, -0.01864624, 0.0...
http://arxiv.org/abs/2311.07914v1
Can Knowledge Graphs Reduce Hallucinations in LLMs? : A Survey
Garima Agrawal, Tharindu Kumarage, Zeyad Alghami, Huan Liu
cs.CL, cs.LG
The contemporary LLMs are prone to producing hallucinations, stemming mainly from the knowledge gaps within the models. To address this critical limitation, researchers employ diverse strategies to augment the LLMs by incorporating external knowledge, aiming to reduce hallucinations and enhance reasoning accuracy. Amon...
http://arxiv.org/pdf/2311.07914v1
2023-11-14T05:21:57
2023-11-14T05:21:57
{ "title": [ -0.047607422, -0.010223389, -0.064331055, -0.040130615, -0.02998352, -0.011428833, -0.06335449, 0.0090789795, -0.0071754456, 0.029342651, -0.029266357, 0.030883789, -0.016723633, -0.023666382, -0.033477783, -0.012512207, -0.0027999878, ...
http://arxiv.org/abs/2311.07911v1
Instruction-Following Evaluation for Large Language Models
Jeffrey Zhou, Tianjian Lu, Swaroop Mishra, Siddhartha Brahma, Sujoy Basu, Yi Luan, Denny Zhou, Le Hou
cs.CL, cs.AI, cs.LG, 68T50 (Primary) 68T99 (Secondary), I.2.7
One core capability of Large Language Models (LLMs) is to follow natural language instructions. However, the evaluation of such abilities is not standardized: Human evaluations are expensive, slow, and not objectively reproducible, while LLM-based auto-evaluation is potentially biased or limited by the ability of the e...
http://arxiv.org/pdf/2311.07911v1
2023-11-14T05:13:55
2023-11-14T05:13:55
{ "title": [ -0.03768921, -0.0033626556, -0.034088135, -0.06149292, 0.0020656586, -0.0014810562, -0.06060791, 0.0013504028, 0.013023376, 0.021881104, -0.02897644, 0.006706238, -0.01020813, -0.006729126, -0.047729492, 0.0023002625, -0.023483276, 0...
http://arxiv.org/abs/2311.07897v1
CPopQA: Ranking Cultural Concept Popularity by LLMs
Ming Jiang, Mansi Joshi
cs.CL
Prior work has demonstrated large language models' (LLMs) potential to discern statistical tendencies within their pre-training corpora. Despite that, many examinations of LLMs' knowledge capacity focus on knowledge explicitly appearing in the training data or implicitly inferable from similar contexts. How well an LLM...
http://arxiv.org/pdf/2311.07897v1
2023-11-14T04:10:40
2023-11-14T04:10:40
{ "title": [ -0.0023040771, -0.010688782, -0.011894226, 0.026687622, -0.01638794, -0.04055786, 0.018157959, -0.041168213, 0.008880615, 0.015571594, -0.043121338, 0.016174316, 0.016036987, -0.044921875, -0.0017900467, -0.009361267, -0.03112793, 0....
http://arxiv.org/abs/2311.07884v1
Fair Abstractive Summarization of Diverse Perspectives
Yusen Zhang, Nan Zhang, Yixin Liu, Alexander Fabbri, Junru Liu, Ryo Kamoi, Xiaoxin Lu, Caiming Xiong, Jieyu Zhao, Dragomir Radev, Kathleen McKeown, Rui Zhang
cs.CL
People from different social and demographic groups express diverse perspectives and conflicting opinions on a broad set of topics such as product reviews, healthcare, law, and politics. A fair summary should provide a comprehensive coverage of diverse perspectives without underrepresenting certain groups. However, cur...
http://arxiv.org/pdf/2311.07884v1
2023-11-14T03:38:55
2023-11-14T03:38:55
{ "title": [ 0.030471802, 0.009536743, -0.03881836, -0.10076904, -0.035705566, -0.019256592, -0.0748291, 0.00015294552, -0.0034103394, 0.056243896, -0.03805542, 0.008033752, 0.0016908646, -0.025466919, -0.038208008, 0.0010929108, 0.012420654, 0.0...
http://arxiv.org/abs/2311.07879v1
Toxicity Detection is NOT all you Need: Measuring the Gaps to Supporting Volunteer Content Moderators
Yang Trista Cao, Lovely-Frances Domingo, Sarah Ann Gilbert, Michelle Mazurek, Katie Shilton, Hal Daumé III
cs.CL, cs.AI
Extensive efforts in automated approaches for content moderation have been focused on developing models to identify toxic, offensive, and hateful content -- with the aim of lightening the load for moderators. Yet, it remains uncertain whether improvements on those tasks truly address the needs that moderators have in a...
http://arxiv.org/pdf/2311.07879v1
2023-11-14T03:18:28
2023-11-14T03:18:28
{ "title": [ 0.032928467, 0.019699097, -0.02218628, -0.07165527, 0.041107178, -0.024398804, -0.011070251, -0.026138306, -0.0028324127, 0.027816772, -0.030670166, 0.025039673, 0.012077332, -0.012413025, -0.027297974, -0.0033397675, -0.0072517395, ...
http://arxiv.org/abs/2311.07853v1
Learning Mutually Informed Representations for Characters and Subwords
Yilin Wang, Xinyi Hu, Matthew R. Gormley
cs.CL, cs.LG
Most pretrained language models rely on subword tokenization, which processes text as a sequence of subword tokens. However, different granularities of text, such as characters, subwords, and words, can contain different kinds of information. Previous studies have shown that incorporating multiple input granularities i...
http://arxiv.org/pdf/2311.07853v1
2023-11-14T02:09:10
2023-11-14T02:09:10
{ "title": [ -0.013191223, 0.0013742447, -0.026245117, -0.058258057, -0.010131836, -0.006084442, -0.038269043, 0.0011930466, 0.023208618, 0.039215088, -0.029693604, 0.006378174, 0.038757324, -0.0102005005, -0.029541016, 0.019012451, -0.0011997223, ...
http://arxiv.org/abs/2311.07850v1
Bring Your Own KG: Self-Supervised Program Synthesis for Zero-Shot KGQA
Dhruv Agarwal, Rajarshi Das, Sopan Khosla, Rashmi Gangadharaiah
cs.CL, cs.AI, cs.DB, cs.LG
We present BYOKG, a universal question-answering (QA) system that can operate on any knowledge graph (KG), requires no human-annotated training data, and can be ready to use within a day -- attributes that are out-of-scope for current KGQA systems. BYOKG draws inspiration from the remarkable ability of humans to compre...
http://arxiv.org/pdf/2311.07850v1
2023-11-14T02:05:29
2023-11-14T02:05:29
{ "title": [ -0.04888916, -0.021865845, -0.024398804, -0.04260254, -0.05557251, 0.007446289, -0.07910156, -0.000030636787, -0.0071907043, 0.0234375, -0.0014400482, 0.001455307, -0.038330078, -0.033416748, -0.032165527, -0.007331848, 0.0025177002, ...
http://arxiv.org/abs/2311.07838v1
LLatrieval: LLM-Verified Retrieval for Verifiable Generation
Xiaonan Li, Changtai Zhu, Linyang Li, Zhangyue Yin, Tianxiang Sun, Xipeng Qiu
cs.CL, cs.AI, cs.IR
Verifiable generation aims to let the large language model (LLM) generate text with corresponding supporting documents, which enables the user to flexibly verify the answer and makes it more trustworthy. Its evaluation not only measures the correctness of the answer, but also the answer's verifiability, i.e., how well ...
http://arxiv.org/pdf/2311.07838v1
2023-11-14T01:38:02
2023-11-14T01:38:02
{ "title": [ -0.019485474, -0.008354187, -0.06463623, -0.103515625, 0.023239136, 0.013389587, -0.07507324, 0.016281128, -0.007881165, 0.03579712, -0.0045661926, 0.02432251, 0.015808105, -0.025405884, -0.013908386, 0.025665283, -0.024932861, 0.018...
http://arxiv.org/abs/2311.07820v1
On the Analysis of Cross-Lingual Prompt Tuning for Decoder-based Multilingual Model
Nohil Park, Joonsuk Park, Kang Min Yoo, Sungroh Yoon
cs.CL
An exciting advancement in the field of multilingual models is the emergence of autoregressive models with zero- and few-shot capabilities, a phenomenon widely reported in large-scale language models. To further improve model adaptation to cross-lingual tasks, another trend is to further fine-tune the language models w...
http://arxiv.org/pdf/2311.07820v1
2023-11-14T00:43:33
2023-11-14T00:43:33
{ "title": [ -0.04071045, -0.013031006, -0.023742676, -0.040527344, -0.024230957, 0.0020866394, -0.090270996, -0.021942139, -0.003944397, 0.054840088, 0.010398865, -0.007160187, 0.045959473, -0.031463623, -0.03265381, 0.030807495, -0.010856628, 0...
http://arxiv.org/abs/2311.07811v1
In-context Learning Generalizes, But Not Always Robustly: The Case of Syntax
Aaron Mueller, Albert Webson, Jackson Petty, Tal Linzen
cs.CL
In-context learning (ICL) is now a common method for supervising large language models (LLMs): given labeled examples in the input context, the LLM learns to perform the task without weight updates. Despite ICL's prevalence and utility, we understand little about whether models supervised in this manner represent the u...
http://arxiv.org/pdf/2311.07811v1
2023-11-13T23:52:43
2023-11-13T23:52:43
{ "title": [ 0.010192871, -0.0005078316, -0.025863647, -0.06008911, 0.0069770813, -0.017318726, -0.012641907, 0.0029239655, -0.015045166, 0.025039673, -0.054473877, 0.029449463, -0.027923584, -0.0003888607, -0.038085938, -0.028625488, -0.0038375854, ...
http://arxiv.org/abs/2311.07804v1
IruMozhi: Automatically classifying diglossia in Tamil
Kabilan Prasanna, Aryaman Arora
cs.CL, I.2.7
Tamil, a Dravidian language of South Asia, is a highly diglossic language with two very different registers in everyday use: Literary Tamil (preferred in writing and formal communication) and Spoken Tamil (confined to speech and informal media). Spoken Tamil is under-supported in modern NLP systems. In this paper, we r...
http://arxiv.org/pdf/2311.07804v1
2023-11-13T23:36:35
2023-11-13T23:36:35
{ "title": [ -0.01235199, -0.008911133, -0.038330078, -0.029876709, -0.029006958, -0.0030155182, -0.034606934, -0.01448822, 0.0077819824, 0.044677734, -0.05718994, -0.011642456, 0.023361206, -0.029388428, -0.042907715, -0.021011353, -0.022323608, ...
http://arxiv.org/abs/2311.07772v2
In-context Learning and Gradient Descent Revisited
Tomer Bar Natan, Gilad Deutch, Nadav Magar, Guy Dar
cs.CL, cs.LG
In-context learning (ICL) has shown impressive results in few-shot learning tasks, yet its underlying mechanism is still not fully understood. Recent works suggest that ICL can be thought of as a gradient descent (GD) based optimization process. While promising, these results mainly focus on simplified settings of ICL ...
http://arxiv.org/pdf/2311.07772v2
2023-11-15T15:29:05
2023-11-13T21:42:38
{ "title": [ 0.0075035095, -0.0005259514, -0.03527832, -0.04901123, 0.032684326, -0.026062012, 0.012458801, -0.016983032, -0.0030994415, 0.022583008, -0.030212402, 0.011672974, -0.027664185, -0.011772156, -0.030090332, -0.0042266846, 0.0035324097, ...