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/2310.13996v1
Emulating the Human Mind: A Neural-symbolic Link Prediction Model with Fast and Slow Reasoning and Filtered Rules
Mohammad Hossein Khojasteh, Najmeh Torabian, Ali Farjami, Saeid Hosseini, Behrouz Minaei-Bidgoli
cs.CL
Link prediction is an important task in addressing the incompleteness problem of knowledge graphs (KG). Previous link prediction models suffer from issues related to either performance or explanatory capability. Furthermore, models that are capable of generating explanations, often struggle with erroneous paths or reas...
http://arxiv.org/pdf/2310.13996v1
2023-10-21T12:45:11
2023-10-21T12:45:11
{ "title": [ 0.0033073425, -0.010375977, -0.013412476, -0.035614014, -0.01852417, -0.0155181885, -0.0059051514, -0.024124146, -0.0138168335, 0.033050537, 0.017410278, -0.021347046, 0.005039215, -0.047454834, -0.031585693, -0.004058838, -0.0011787415,...
http://arxiv.org/abs/2310.13995v1
On Bilingual Lexicon Induction with Large Language Models
Yaoyiran Li, Anna Korhonen, Ivan Vulić
cs.CL, cs.AI, cs.IR, cs.LG
Bilingual Lexicon Induction (BLI) is a core task in multilingual NLP that still, to a large extent, relies on calculating cross-lingual word representations. Inspired by the global paradigm shift in NLP towards Large Language Models (LLMs), we examine the potential of the latest generation of LLMs for the development o...
http://arxiv.org/pdf/2310.13995v1
2023-10-21T12:43:27
2023-10-21T12:43:27
{ "title": [ -0.045043945, -0.016296387, -0.02923584, -0.060516357, -0.011222839, 0.028411865, -0.06896973, 0.013420105, 0.0056495667, 0.02319336, -0.039916992, 0.0026855469, -0.01612854, -0.0050201416, -0.036987305, -0.011001587, -0.028671265, 0...
http://arxiv.org/abs/2310.13990v1
A Novel Information-Theoretic Objective to Disentangle Representations for Fair Classification
Pierre Colombo, Nathan Noiry, Guillaume Staerman, Pablo Piantanida
cs.LG, cs.CL
One of the pursued objectives of deep learning is to provide tools that learn abstract representations of reality from the observation of multiple contextual situations. More precisely, one wishes to extract disentangled representations which are (i) low dimensional and (ii) whose components are independent and corresp...
http://arxiv.org/pdf/2310.13990v1
2023-10-21T12:35:48
2023-10-21T12:35:48
{ "title": [ -0.0018014908, 0.013679504, -0.053710938, -0.048797607, 0.026412964, -0.01826477, -0.0004339218, -0.009101868, 0.012123108, 0.029373169, -0.043273926, 0.020050049, -0.0066604614, -0.009742737, -0.026550293, -0.0003437996, -0.006549835, ...
http://arxiv.org/abs/2310.13988v1
GEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4
Tom Kocmi, Christian Federmann
cs.CL
This paper introduces GEMBA-MQM, a GPT-based evaluation metric designed to detect translation quality errors, specifically for the quality estimation setting without the need for human reference translations. Based on the power of large language models (LLM), GEMBA-MQM employs a fixed three-shot prompting technique, qu...
http://arxiv.org/pdf/2310.13988v1
2023-10-21T12:30:33
2023-10-21T12:30:33
{ "title": [ -0.025726318, -0.008529663, 0.0024528503, -0.024734497, -0.01423645, 0.013908386, -0.040222168, -0.03793335, 0.0071907043, 0.03112793, 0.0004415512, 0.015541077, 0.017456055, -0.03237915, -0.037506104, 0.031188965, -0.025894165, 0.03...
http://arxiv.org/abs/2310.13985v1
HateRephrase: Zero- and Few-Shot Reduction of Hate Intensity in Online Posts using Large Language Models
Vibhor Agarwal, Yu Chen, Nishanth Sastry
cs.CL
Hate speech has become pervasive in today's digital age. Although there has been considerable research to detect hate speech or generate counter speech to combat hateful views, these approaches still cannot completely eliminate the potential harmful societal consequences of hate speech -- hate speech, even when detecte...
http://arxiv.org/pdf/2310.13985v1
2023-10-21T12:18:29
2023-10-21T12:18:29
{ "title": [ 0.022216797, -0.018814087, -0.029678345, -0.03378296, 0.0028839111, -0.03161621, -0.00009948015, -0.0023403168, 0.007286072, 0.04425049, -0.017303467, 0.0035095215, 0.045562744, -0.011878967, -0.04815674, 0.013404846, -0.0033473969, ...
http://arxiv.org/abs/2310.13974v1
Automatic Pronunciation Assessment -- A Review
Yassine El Kheir, Ahmed Ali, Shammur Absar Chowdhury
cs.CL, cs.SD, eess.AS
Pronunciation assessment and its application in computer-aided pronunciation training (CAPT) have seen impressive progress in recent years. With the rapid growth in language processing and deep learning over the past few years, there is a need for an updated review. In this paper, we review methods employed in pronunci...
http://arxiv.org/pdf/2310.13974v1
2023-10-21T11:26:24
2023-10-21T11:26:24
{ "title": [ -0.021957397, 0.008850098, -0.03048706, -0.0546875, 0.002784729, 0.009117126, -0.024017334, 0.017196655, 0.0026664734, 0.0519104, -0.0181427, -0.0029830933, 0.003353119, -0.016723633, -0.06591797, 0.010299683, -0.018249512, 0.0422363...
http://arxiv.org/abs/2310.18336v1
AITA Generating Moral Judgements of the Crowd with Reasoning
Osama Bsher, Ameer Sabri
cs.CL, cs.LG
Morality is a fundamental aspect of human behavior and ethics, influencing how we interact with each other and the world around us. When faced with a moral dilemma, a person's ability to make clear moral judgments can be clouded. Due to many factors such as personal biases, emotions and situational factors people can f...
http://arxiv.org/pdf/2310.18336v1
2023-10-21T10:27:22
2023-10-21T10:27:22
{ "title": [ 0.016693115, 0.012268066, -0.027923584, -0.035217285, -0.02734375, -0.000033915043, 0.0034599304, 0.008415222, -0.044799805, 0.039276123, -0.026489258, -0.00055885315, 0.0060424805, -0.01878357, 0.0055618286, -0.028259277, -0.0033512115,...
http://arxiv.org/abs/2310.13961v1
Ensemble-Instruct: Generating Instruction-Tuning Data with a Heterogeneous Mixture of LMs
Young-Suk Lee, Md Arafat Sultan, Yousef El-Kurdi, Tahira Naseem Asim Munawar, Radu Florian, Salim Roukos, Ramón Fernandez Astudillo
cs.CL, cs.AI
Using in-context learning (ICL) for data generation, techniques such as Self-Instruct (Wang et al., 2023) or the follow-up Alpaca (Taori et al., 2023) can train strong conversational agents with only a small amount of human supervision. One limitation of these approaches is that they resort to very large language model...
http://arxiv.org/pdf/2310.13961v1
2023-10-21T10:21:17
2023-10-21T10:21:17
{ "title": [ -0.06817627, -0.01979065, -0.047698975, -0.024795532, -0.029418945, -0.0016508102, -0.03869629, 0.010673523, -0.0054969788, 0.018035889, -0.06488037, 0.036956787, -0.037841797, 0.0010890961, -0.024795532, -0.0140686035, -0.00092697144, ...
http://arxiv.org/abs/2310.13960v2
Linguistically Motivated Sign Language Segmentation
Amit Moryossef, Zifan Jiang, Mathias Müller, Sarah Ebling, Yoav Goldberg
cs.CL, cs.CV
Sign language segmentation is a crucial task in sign language processing systems. It enables downstream tasks such as sign recognition, transcription, and machine translation. In this work, we consider two kinds of segmentation: segmentation into individual signs and segmentation into phrases, larger units comprising s...
http://arxiv.org/pdf/2310.13960v2
2023-10-30T13:51:39
2023-10-21T10:09:34
{ "title": [ -0.034088135, 0.0009908676, -0.033996582, -0.035308838, -0.022628784, 0.0046958923, -0.050598145, -0.0070648193, -0.01146698, 0.019424438, -0.036956787, 0.007850647, -0.02758789, -0.025146484, -0.021331787, -0.022705078, -0.0028324127, ...
http://arxiv.org/abs/2310.13915v1
Values, Ethics, Morals? On the Use of Moral Concepts in NLP Research
Karina Vida, Judith Simon, Anne Lauscher
cs.CL, cs.CY
With language technology increasingly affecting individuals' lives, many recent works have investigated the ethical aspects of NLP. Among other topics, researchers focused on the notion of morality, investigating, for example, which moral judgements language models make. However, there has been little to no discussion ...
http://arxiv.org/pdf/2310.13915v1
2023-10-21T06:04:10
2023-10-21T06:04:10
{ "title": [ -0.014633179, 0.0025806427, -0.033447266, -0.012992859, 0.0057525635, -0.028167725, 0.0030822754, -0.023284912, -0.010299683, 0.032104492, -0.039154053, 0.020126343, -0.0033054352, 0.008743286, -0.0496521, 0.0008587837, -0.038513184, ...
http://arxiv.org/abs/2310.13895v1
RTSUM: Relation Triple-based Interpretable Summarization with Multi-level Salience Visualization
Seonglae Cho, Yonggi Cho, HoonJae Lee, Myungha Jang, Jinyoung Yeo, Dongha Lee
cs.CL, cs.LG
In this paper, we present RTSUM, an unsupervised summarization framework that utilizes relation triples as the basic unit for summarization. Given an input document, RTSUM first selects salient relation triples via multi-level salience scoring and then generates a concise summary from the selected relation triples by u...
http://arxiv.org/pdf/2310.13895v1
2023-10-21T02:46:03
2023-10-21T02:46:03
{ "title": [ 0.015525818, -0.012168884, -0.030380249, -0.099243164, -0.014427185, 0.013702393, -0.0703125, -0.007003784, 0.041046143, 0.04168701, -0.018478394, 0.0017728806, -0.0005431175, 0.011978149, -0.08703613, -0.0027637482, 0.0071487427, 0....
http://arxiv.org/abs/2310.13864v1
RECAP: Towards Precise Radiology Report Generation via Dynamic Disease Progression Reasoning
Wenjun Hou, Yi Cheng, Kaishuai Xu, Wenjie Li, Jiang Liu
cs.CL
Automating radiology report generation can significantly alleviate radiologists' workloads. Previous research has primarily focused on realizing highly concise observations while neglecting the precise attributes that determine the severity of diseases (e.g., small pleural effusion). Since incorrect attributes will lea...
http://arxiv.org/pdf/2310.13864v1
2023-10-21T00:05:32
2023-10-21T00:05:32
{ "title": [ 0.0076179504, -0.00642395, -0.043518066, -0.10064697, 0.012557983, 0.022033691, -0.056488037, 0.02279663, 0.00022482872, 0.009170532, -0.0063972473, 0.006450653, -0.030761719, -0.032440186, -0.029159546, 0.0033874512, 0.014923096, 0....
http://arxiv.org/abs/2310.13859v1
Not all Fake News is Written: A Dataset and Analysis of Misleading Video Headlines
Yoo Yeon Sung, Jordan Boyd-Graber, Naeemul Hassan
cs.CL, cs.CV
Polarization and the marketplace for impressions have conspired to make navigating information online difficult for users, and while there has been a significant effort to detect false or misleading text, multimodal datasets have received considerably less attention. To complement existing resources, we present multimo...
http://arxiv.org/pdf/2310.13859v1
2023-10-20T23:47:01
2023-10-20T23:47:01
{ "title": [ -0.00687027, 0.013343811, -0.04171753, -0.037322998, 0.026062012, -0.017120361, -0.00018775463, -0.028137207, -0.050720215, 0.02243042, -0.025634766, 0.031707764, -0.02468872, -0.0067214966, -0.026351929, -0.027862549, -0.024505615, ...
http://arxiv.org/abs/2310.13856v1
Implications of Annotation Artifacts in Edge Probing Test Datasets
Sagnik Ray Choudhury, Jushaan Kalra
cs.CL, I.2.7
Edge probing tests are classification tasks that test for grammatical knowledge encoded in token representations coming from contextual encoders such as large language models (LLMs). Many LLM encoders have shown high performance in EP tests, leading to conjectures about their ability to encode linguistic knowledge. How...
http://arxiv.org/pdf/2310.13856v1
2023-10-20T23:19:35
2023-10-20T23:19:35
{ "title": [ -0.023956299, -0.0027217865, -0.020629883, -0.041168213, -0.0051879883, -0.0072669983, -0.014953613, -0.0006928444, -0.004310608, 0.04006958, -0.00945282, 0.0035991669, -0.02532959, -0.023880005, -0.061950684, -0.0012559891, 0.002670288,...
http://arxiv.org/abs/2310.13855v1
Evoke: Evoking Critical Thinking Abilities in LLMs via Reviewer-Author Prompt Editing
Xinyu Hu, Pengfei Tang, Simiao Zuo, Zihan Wang, Bowen Song, Qiang Lou, Jian Jiao, Denis Charles
cs.CL, cs.AI
Large language models (LLMs) have made impressive progress in natural language processing. These models rely on proper human instructions (or prompts) to generate suitable responses. However, the potential of LLMs are not fully harnessed by commonly-used prompting methods: many human-in-the-loop algorithms employ ad-ho...
http://arxiv.org/pdf/2310.13855v1
2023-10-20T23:15:59
2023-10-20T23:15:59
{ "title": [ -0.043304443, -0.03086853, -0.026397705, 0.036895752, -0.06774902, -0.017150879, -0.015029907, -0.03366089, -0.020828247, -0.0021076202, -0.047088623, 0.020751953, -0.014419556, -0.036315918, -0.010650635, -0.048614502, -0.039855957, ...
http://arxiv.org/abs/2310.13850v1
Ecologically Valid Explanations for Label Variation in NLI
Nan-Jiang Jiang, Chenhao Tan, Marie-Catherine de Marneffe
cs.CL
Human label variation, or annotation disagreement, exists in many natural language processing (NLP) tasks, including natural language inference (NLI). To gain direct evidence of how NLI label variation arises, we build LiveNLI, an English dataset of 1,415 ecologically valid explanations (annotators explain the NLI labe...
http://arxiv.org/pdf/2310.13850v1
2023-10-20T22:52:19
2023-10-20T22:52:19
{ "title": [ -0.038085938, -0.0131073, -0.027786255, -0.006603241, -0.030151367, -0.011489868, -0.019073486, -0.0076560974, -0.018981934, 0.009788513, -0.02558899, 0.0041389465, -0.030258179, -0.03201294, -0.024642944, -0.016479492, -0.027236938, ...
http://arxiv.org/abs/2310.13836v1
Foundation Model's Embedded Representations May Detect Distribution Shift
Adam Tsou, Max Vargas, Andrew Engel, Tony Chiang
cs.LG, cs.CL
Distribution shifts between train and test datasets obscure our ability to understand the generalization capacity of neural network models. This topic is especially relevant given the success of pre-trained foundation models as starting points for transfer learning (TL) models across tasks and contexts. We present a ca...
http://arxiv.org/pdf/2310.13836v1
2023-10-20T22:20:50
2023-10-20T22:20:50
{ "title": [ 0.035186768, 0.029312134, -0.01890564, 0.012748718, -0.0038986206, -0.027328491, -0.0008921623, -0.060546875, -0.004371643, 0.02217102, -0.03338623, -0.018127441, 0.008605957, -0.040405273, 0.017684937, -0.016143799, 0.011383057, 0.0...
http://arxiv.org/abs/2310.13824v1
Plausibility Processing in Transformer Language Models: Focusing on the Role of Attention Heads in GPT
Soo Hyun Ryu
cs.CL
The goal of this paper is to explore how Transformer language models process semantic knowledge, especially regarding the plausibility of noun-verb relations. First, I demonstrate GPT2 exhibits a higher degree of similarity with humans in plausibility processing compared to other Transformer language models. Next, I de...
http://arxiv.org/pdf/2310.13824v1
2023-10-20T21:31:19
2023-10-20T21:31:19
{ "title": [ 0.010375977, 0.022201538, -0.007549286, -0.027053833, -0.0006837845, -0.018630981, -0.031799316, -0.04071045, -0.008712769, 0.0051002502, -0.014808655, -0.013031006, 0.039978027, -0.037963867, 0.0028839111, -0.0013256073, -0.035980225, ...
http://arxiv.org/abs/2310.13812v1
Yet Another Model for Arabic Dialect Identification
Ajinkya Kulkarni, Hanan Aldarmaki
cs.CL, cs.SD, eess.AS
In this paper, we describe a spoken Arabic dialect identification (ADI) model for Arabic that consistently outperforms previously published results on two benchmark datasets: ADI-5 and ADI-17. We explore two architectural variations: ResNet and ECAPA-TDNN, coupled with two types of acoustic features: MFCCs and features...
http://arxiv.org/pdf/2310.13812v1
2023-10-20T20:58:45
2023-10-20T20:58:45
{ "title": [ 0.011199951, 0.0038852692, -0.01348114, -0.014175415, -0.037750244, -0.002275467, -0.012138367, -0.008460999, -0.047943115, 0.043273926, 0.014259338, -0.022628784, 0.0029277802, -0.07672119, 0.014472961, 0.00014138222, 0.028839111, 0...
http://arxiv.org/abs/2310.13800v1
Evaluation Metrics in the Era of GPT-4: Reliably Evaluating Large Language Models on Sequence to Sequence Tasks
Andrea Sottana, Bin Liang, Kai Zou, Zheng Yuan
cs.CL
Large Language Models (LLMs) evaluation is a patchy and inconsistent landscape, and it is becoming clear that the quality of automatic evaluation metrics is not keeping up with the pace of development of generative models. We aim to improve the understanding of current models' performance by providing a preliminary and...
http://arxiv.org/pdf/2310.13800v1
2023-10-20T20:17:09
2023-10-20T20:17:09
{ "title": [ -0.030929565, 0.0005469322, -0.02319336, -0.041809082, 0.0020866394, -0.0033473969, -0.04296875, -0.015960693, 0.005004883, 0.008216858, -0.002943039, 0.0038528442, -0.009292603, -0.037200928, -0.026916504, 0.0019874573, -0.020385742, ...
http://arxiv.org/abs/2310.13798v1
Specific versus General Principles for Constitutional AI
Sandipan Kundu, Yuntao Bai, Saurav Kadavath, Amanda Askell, Andrew Callahan, Anna Chen, Anna Goldie, Avital Balwit, Azalia Mirhoseini, Brayden McLean, Catherine Olsson, Cassie Evraets, Eli Tran-Johnson, Esin Durmus, Ethan Perez, Jackson Kernion, Jamie Kerr, Kamal Ndousse, Karina Nguyen, Nelson Elhage, Newton Cheng, Nic...
cs.CL, cs.AI
Human feedback can prevent overtly harmful utterances in conversational models, but may not automatically mitigate subtle problematic behaviors such as a stated desire for self-preservation or power. Constitutional AI offers an alternative, replacing human feedback with feedback from AI models conditioned only on a lis...
http://arxiv.org/pdf/2310.13798v1
2023-10-20T20:12:45
2023-10-20T20:12:45
{ "title": [ 0.012428284, 0.0013999939, 0.00037431717, -0.012252808, -0.025985718, -0.019073486, 0.0059661865, -0.02859497, -0.022216797, 0.024093628, -0.044128418, 0.014350891, -0.005908966, -0.010498047, -0.0062561035, 0.0005083084, -0.015220642, ...
http://arxiv.org/abs/2310.13793v1
A Unified View of Evaluation Metrics for Structured Prediction
Yunmo Chen, William Gantt, Tongfei Chen, Aaron Steven White, Benjamin Van Durme
cs.CL, cs.LG
We present a conceptual framework that unifies a variety of evaluation metrics for different structured prediction tasks (e.g. event and relation extraction, syntactic and semantic parsing). Our framework requires representing the outputs of these tasks as objects of certain data types, and derives metrics through matc...
http://arxiv.org/pdf/2310.13793v1
2023-10-20T20:02:02
2023-10-20T20:02:02
{ "title": [ -0.028671265, -0.008804321, -0.032958984, -0.08557129, 0.000477314, 0.008125305, -0.055267334, -0.0024318695, 0.011054993, 0.006706238, -0.017181396, -0.002527237, -0.029174805, -0.024490356, -0.043548584, -0.0054016113, 0.0045204163, ...
http://arxiv.org/abs/2310.13781v1
How Much Consistency Is Your Accuracy Worth?
Jacob K. Johnson, Ana Marasović
cs.CL
Contrast set consistency is a robustness measurement that evaluates the rate at which a model correctly responds to all instances in a bundle of minimally different examples relying on the same knowledge. To draw additional insights, we propose to complement consistency with relative consistency -- the probability that...
http://arxiv.org/pdf/2310.13781v1
2023-10-20T19:28:06
2023-10-20T19:28:06
{ "title": [ 0.0044555664, 0.00057792664, -0.052886963, -0.009666443, -0.02279663, -0.034606934, -0.02130127, -0.003232956, -0.040985107, 0.034729004, -0.02810669, -0.026947021, -0.008773804, -0.01852417, -0.013442993, -0.033050537, 0.010757446, ...
http://arxiv.org/abs/2310.13774v1
Seq2seq is All You Need for Coreference Resolution
Wenzheng Zhang, Sam Wiseman, Karl Stratos
cs.CL
Existing works on coreference resolution suggest that task-specific models are necessary to achieve state-of-the-art performance. In this work, we present compelling evidence that such models are not necessary. We finetune a pretrained seq2seq transformer to map an input document to a tagged sequence encoding the coref...
http://arxiv.org/pdf/2310.13774v1
2023-10-20T19:17:22
2023-10-20T19:17:22
{ "title": [ 0.02468872, -0.022644043, -0.040283203, -0.072753906, -0.019226074, -0.024841309, -0.059783936, 0.025436401, -0.0060539246, 0.06573486, 0.009353638, -0.006881714, 0.08123779, -0.02696228, -0.00024819374, 0.03778076, -0.0036621094, 0....
http://arxiv.org/abs/2310.13771v1
Copyright Violations and Large Language Models
Antonia Karamolegkou, Jiaang Li, Li Zhou, Anders Søgaard
cs.CL, cs.AI
Language models may memorize more than just facts, including entire chunks of texts seen during training. Fair use exemptions to copyright laws typically allow for limited use of copyrighted material without permission from the copyright holder, but typically for extraction of information from copyrighted materials, ra...
http://arxiv.org/pdf/2310.13771v1
2023-10-20T19:14:59
2023-10-20T19:14:59
{ "title": [ -0.008468628, 0.00014913082, -0.023788452, -0.053131104, -0.0062789917, -0.006095886, -0.035125732, 0.002073288, -0.0022449493, 0.04815674, -0.035369873, -0.0050849915, 0.017791748, -0.011054993, -0.037628174, 0.0030727386, -0.01966858, ...
http://arxiv.org/abs/2310.13760v1
Enhancing Abstractiveness of Summarization Models through Calibrated Distillation
Hwanjun Song, Igor Shalyminov, Hang Su, Siffi Singh, Kaisheng Yao, Saab Mansour
cs.CL
Sequence-level knowledge distillation reduces the size of Seq2Seq models for more efficient abstractive summarization. However, it often leads to a loss of abstractiveness in summarization. In this paper, we propose a novel approach named DisCal to enhance the level of abstractiveness (measured by n-gram overlap) witho...
http://arxiv.org/pdf/2310.13760v1
2023-10-20T18:43:49
2023-10-20T18:43:49
{ "title": [ 0.009536743, 0.02960205, -0.046722412, -0.10638428, -0.0028648376, 0.011062622, -0.1071167, -0.02017212, 0.05496216, 0.055236816, -0.030761719, -0.0065078735, -0.010421753, -0.0014390945, -0.08355713, 0.009963989, -0.0035629272, 0.03...
http://arxiv.org/abs/2310.13747v1
ALDi: Quantifying the Arabic Level of Dialectness of Text
Amr Keleg, Sharon Goldwater, Walid Magdy
cs.CL
Transcribed speech and user-generated text in Arabic typically contain a mixture of Modern Standard Arabic (MSA), the standardized language taught in schools, and Dialectal Arabic (DA), used in daily communications. To handle this variation, previous work in Arabic NLP has focused on Dialect Identification (DI) on the ...
http://arxiv.org/pdf/2310.13747v1
2023-10-20T18:07:39
2023-10-20T18:07:39
{ "title": [ -0.02684021, -0.019515991, -0.06008911, -0.019241333, -0.04751587, -0.0038661957, -0.055419922, 0.0020065308, -0.0012178421, 0.03778076, -0.001666069, 0.0009946823, 0.049713135, -0.06549072, 0.009719849, 0.0075302124, 0.02229309, 0.0...
http://arxiv.org/abs/2310.13686v1
Exploring Linguistic Probes for Morphological Generalization
Jordan Kodner, Salam Khalifa, Sarah Payne
cs.CL
Modern work on the cross-linguistic computational modeling of morphological inflection has typically employed language-independent data splitting algorithms. In this paper, we supplement that approach with language-specific probes designed to test aspects of morphological generalization. Testing these probes on three m...
http://arxiv.org/pdf/2310.13686v1
2023-10-20T17:45:30
2023-10-20T17:45:30
{ "title": [ -0.013542175, -0.009902954, 0.0012483597, -0.0491333, -0.0178833, 0.0006866455, -0.032287598, -0.0033302307, -0.019622803, 0.03149414, -0.018447876, 0.017974854, -0.015434265, -0.019241333, -0.030059814, -0.003862381, -0.01939392, 0....
http://arxiv.org/abs/2310.13682v2
Optimizing Retrieval-augmented Reader Models via Token Elimination
Moshe Berchansky, Peter Izsak, Avi Caciularu, Ido Dagan, Moshe Wasserblat
cs.CL, cs.AI, cs.LG
Fusion-in-Decoder (FiD) is an effective retrieval-augmented language model applied across a variety of open-domain tasks, such as question answering, fact checking, etc. In FiD, supporting passages are first retrieved and then processed using a generative model (Reader), which can cause a significant bottleneck in deco...
http://arxiv.org/pdf/2310.13682v2
2023-11-05T06:31:55
2023-10-20T17:41:36
{ "title": [ -0.01285553, 0.0012083054, -0.016662598, -0.031799316, -0.010101318, -0.00045657158, -0.03945923, -0.038146973, 0.034210205, 0.044158936, -0.0107803345, -0.014945984, 0.013084412, -0.0066108704, -0.07446289, 0.009109497, -0.019607544, ...
http://arxiv.org/abs/2310.13678v2
Long-Form Speech Translation through Segmentation with Finite-State Decoding Constraints on Large Language Models
Arya D. McCarthy, Hao Zhang, Shankar Kumar, Felix Stahlberg, Ke Wu
cs.CL, cs.AI, cs.LG
One challenge in speech translation is that plenty of spoken content is long-form, but short units are necessary for obtaining high-quality translations. To address this mismatch, we adapt large language models (LLMs) to split long ASR transcripts into segments that can be independently translated so as to maximize the...
http://arxiv.org/pdf/2310.13678v2
2023-10-23T15:25:55
2023-10-20T17:31:39
{ "title": [ -0.03918457, 0.0033340454, -0.022064209, -0.058532715, -0.017364502, 0.009651184, -0.057922363, 0.016540527, -0.024337769, 0.030258179, -0.004558563, 0.0035037994, 0.007888794, -0.027908325, -0.024353027, -0.0063209534, 0.0021591187, ...
http://arxiv.org/abs/2310.13676v1
Information Value: Measuring Utterance Predictability as Distance from Plausible Alternatives
Mario Giulianelli, Sarenne Wallbridge, Raquel Fernández
cs.CL
We present information value, a measure which quantifies the predictability of an utterance relative to a set of plausible alternatives. We introduce a method to obtain interpretable estimates of information value using neural text generators, and exploit their psychometric predictive power to investigate the dimension...
http://arxiv.org/pdf/2310.13676v1
2023-10-20T17:25:36
2023-10-20T17:25:36
{ "title": [ -0.004322052, 0.0019350052, -0.032989502, -0.035736084, -0.006931305, -0.02558899, -0.020812988, -0.03152466, -0.0022182465, -0.0063095093, -0.035705566, 0.00046944618, -0.0072402954, -0.038726807, -0.0044136047, -0.020507812, -0.0065307...
http://arxiv.org/abs/2310.13675v1
On Synthetic Data for Back Translation
Jiahao Xu, Yubin Ruan, Wei Bi, Guoping Huang, Shuming Shi, Lihui Chen, Lemao Liu
cs.CL
Back translation (BT) is one of the most significant technologies in NMT research fields. Existing attempts on BT share a common characteristic: they employ either beam search or random sampling to generate synthetic data with a backward model but seldom work studies the role of synthetic data in the performance of BT....
http://arxiv.org/pdf/2310.13675v1
2023-10-20T17:24:12
2023-10-20T17:24:12
{ "title": [ -0.02154541, 0.0022201538, -0.008201599, -0.060577393, -0.032287598, 0.012321472, -0.068725586, -0.011482239, -0.010627747, 0.03878784, -0.0154800415, -0.0051460266, -0.029953003, -0.028442383, -0.025131226, 0.0046081543, 0.009422302, ...
http://arxiv.org/abs/2310.13673v2
StereoMap: Quantifying the Awareness of Human-like Stereotypes in Large Language Models
Sullam Jeoung, Yubin Ge, Jana Diesner
cs.CL
Large Language Models (LLMs) have been observed to encode and perpetuate harmful associations present in the training data. We propose a theoretically grounded framework called StereoMap to gain insights into their perceptions of how demographic groups have been viewed by society. The framework is grounded in the Stere...
http://arxiv.org/pdf/2310.13673v2
2023-10-31T16:41:31
2023-10-20T17:22:30
{ "title": [ -0.014076233, 0.010932922, -0.036987305, -0.030349731, -0.0026855469, -0.01928711, -0.02909851, -0.020950317, 0.014968872, 0.02722168, -0.04135132, 0.016647339, 0.02406311, -0.005378723, -0.0395813, 0.01361084, -0.027999878, 0.040588...
http://arxiv.org/abs/2310.13671v1
Let's Synthesize Step by Step: Iterative Dataset Synthesis with Large Language Models by Extrapolating Errors from Small Models
Ruida Wang, Wangchunshu Zhou, Mrinmaya Sachan
cs.CL, cs.AI
*Data Synthesis* is a promising way to train a small model with very little labeled data. One approach for data synthesis is to leverage the rich knowledge from large language models to synthesize pseudo training examples for small models, making it possible to achieve both data and compute efficiency at the same time....
http://arxiv.org/pdf/2310.13671v1
2023-10-20T17:14:25
2023-10-20T17:14:25
{ "title": [ -0.039031982, -0.0024375916, -0.059143066, -0.06237793, -0.039978027, 0.0024814606, -0.10479736, 0.0056114197, -0.0050468445, 0.03781128, -0.08526611, 0.03756714, -0.003955841, 0.010498047, 0.0037002563, -0.044036865, 0.020187378, 0....
http://arxiv.org/abs/2310.13669v1
Automatic Unit Test Data Generation and Actor-Critic Reinforcement Learning for Code Synthesis
Philip John Gorinski, Matthieu Zimmer, Gerasimos Lampouras, Derrick Goh Xin Deik, Ignacio Iacobacci
cs.LG, cs.AI, cs.CL, cs.PL
The advent of large pre-trained language models in the domain of Code Synthesis has shown remarkable performance on various benchmarks, treating the problem of Code Generation in a fashion similar to Natural Language Generation, trained with a Language Modelling (LM) objective. In addition, the property of programming ...
http://arxiv.org/pdf/2310.13669v1
2023-10-20T17:13:16
2023-10-20T17:13:16
{ "title": [ -0.04928589, -0.019180298, -0.027145386, -0.04220581, -0.02279663, 0.012817383, -0.05609131, 0.014289856, -0.03250122, 0.0093688965, -0.035003662, 0.046081543, -0.05102539, -0.010665894, -0.0063934326, -0.028167725, -0.01574707, 0.00...
http://arxiv.org/abs/2310.13664v2
Explainable Depression Symptom Detection in Social Media
Eliseo Bao Souto, Anxo Pérez, Javier Parapar
cs.CL
Users of social platforms often perceive these sites as supportive spaces to post about their mental health issues. Those conversations contain important traces about individuals' health risks. Recently, researchers have exploited this online information to construct mental health detection models, which aim to identif...
http://arxiv.org/pdf/2310.13664v2
2023-10-23T08:31:50
2023-10-20T17:05:27
{ "title": [ 0.024017334, 0.012481689, 0.001953125, -0.046722412, 0.022949219, -0.0006451607, 0.0008277893, -0.02619934, -0.008682251, 0.023468018, -0.017471313, -0.0066223145, 0.009429932, -0.027893066, -0.022598267, -0.016525269, -0.011802673, ...
http://arxiv.org/abs/2310.13661v1
Arabic Dialect Identification under Scrutiny: Limitations of Single-label Classification
Amr Keleg, Walid Magdy
cs.CL
Automatic Arabic Dialect Identification (ADI) of text has gained great popularity since it was introduced in the early 2010s. Multiple datasets were developed, and yearly shared tasks have been running since 2018. However, ADI systems are reported to fail in distinguishing between the micro-dialects of Arabic. We argue...
http://arxiv.org/pdf/2310.13661v1
2023-10-20T17:04:22
2023-10-20T17:04:22
{ "title": [ -0.0068626404, 0.00248909, -0.021453857, -0.03491211, -0.019119263, -0.0037651062, -0.015975952, -0.006881714, -0.039276123, 0.019836426, 0.01864624, -0.016479492, -0.021148682, -0.06756592, -0.012390137, -0.0043411255, 0.016342163, ...
http://arxiv.org/abs/2310.13659v1
Benchmarking and Improving Text-to-SQL Generation under Ambiguity
Adithya Bhaskar, Tushar Tomar, Ashutosh Sathe, Sunita Sarawagi
cs.CL
Research in Text-to-SQL conversion has been largely benchmarked against datasets where each text query corresponds to one correct SQL. However, natural language queries over real-life databases frequently involve significant ambiguity about the intended SQL due to overlapping schema names and multiple confusing relatio...
http://arxiv.org/pdf/2310.13659v1
2023-10-20T17:00:53
2023-10-20T17:00:53
{ "title": [ -0.042663574, -0.009735107, -0.048187256, -0.066223145, 0.0054397583, 0.016906738, -0.0619812, 0.01259613, 0.023788452, 0.044189453, -0.027404785, 0.011795044, -0.0027236938, -0.002374649, -0.07165527, -0.012123108, -0.008644104, 0.0...
http://arxiv.org/abs/2310.13650v1
BotChat: Evaluating LLMs' Capabilities of Having Multi-Turn Dialogues
Haodong Duan, Jueqi Wei, Chonghua Wang, Hongwei Liu, Yixiao Fang, Songyang Zhang, Dahua Lin, Kai Chen
cs.CL
Interacting with human via high-quality multi-turn dialogues is a key feature of large language models (LLMs). However, human-based evaluation of such capability involves intensive manual labor. This report provides a preliminary evaluation of existing large language models for human-style multi-turn chatting, through ...
http://arxiv.org/pdf/2310.13650v1
2023-10-20T16:53:51
2023-10-20T16:53:51
{ "title": [ 0.004173279, -0.019943237, -0.025787354, -0.01889038, -0.037078857, -0.019638062, -0.021972656, 0.0011959076, -0.013877869, 0.032989502, -0.0546875, 0.03237915, -0.013786316, -0.017089844, -0.011245728, -0.032104492, -0.03967285, 0.0...
http://arxiv.org/abs/2310.13620v2
Bridging Information-Theoretic and Geometric Compression in Language Models
Emily Cheng, Corentin Kervadec, Marco Baroni
cs.CL
For a language model (LM) to faithfully model human language, it must compress vast, potentially infinite information into relatively few dimensions. We propose analyzing compression in (pre-trained) LMs from two points of view: geometric and information-theoretic. We demonstrate that the two views are highly correlate...
http://arxiv.org/pdf/2310.13620v2
2023-11-09T14:03:46
2023-10-20T16:12:13
{ "title": [ -0.062683105, -0.018417358, -0.038879395, -0.05883789, -0.01109314, 0.022918701, -0.06945801, 0.021759033, 0.012161255, 0.039123535, 0.006565094, -0.0027427673, -0.003276825, -0.014526367, -0.05444336, 0.020339966, 0.012519836, 0.014...
http://arxiv.org/abs/2310.13619v1
Semi-supervised multimodal coreference resolution in image narrations
Arushi Goel, Basura Fernando, Frank Keller, Hakan Bilen
cs.CL, cs.CV
In this paper, we study multimodal coreference resolution, specifically where a longer descriptive text, i.e., a narration is paired with an image. This poses significant challenges due to fine-grained image-text alignment, inherent ambiguity present in narrative language, and unavailability of large annotated training...
http://arxiv.org/pdf/2310.13619v1
2023-10-20T16:10:14
2023-10-20T16:10:14
{ "title": [ 0.0037670135, 0.0042915344, -0.025161743, -0.07305908, -0.0067596436, -0.018051147, -0.03463745, 0.019882202, -0.0029525757, 0.054260254, -0.0070724487, 0.017959595, 0.039794922, -0.017333984, -0.022109985, 0.01058197, -0.019805908, ...
http://arxiv.org/abs/2310.13615v1
Three Questions Concerning the Use of Large Language Models to Facilitate Mathematics Learning
An-Zi Yen, Wei-Ling Hsu
cs.CL
Due to the remarkable language understanding and generation abilities of large language models (LLMs), their use in educational applications has been explored. However, little work has been done on investigating the pedagogical ability of LLMs in helping students to learn mathematics. In this position paper, we discuss...
http://arxiv.org/pdf/2310.13615v1
2023-10-20T16:05:35
2023-10-20T16:05:35
{ "title": [ -0.008621216, -0.01121521, -0.027557373, -0.008125305, -0.031158447, -0.015899658, -0.0064353943, 0.0118637085, -0.012550354, 0.020065308, -0.021850586, 0.0047302246, -0.006866455, -0.039276123, 0.018081665, -0.023651123, 0.017684937, ...
http://arxiv.org/abs/2310.13613v2
Hunayn: Elevating Translation Beyond the Literal
Nasser Almousa, Nasser Alzamil, Abdullah Alshehri, Ahmad Sait
cs.CL, cs.AI
This project introduces an advanced English-to-Arabic translator surpassing conventional tools. Leveraging the Helsinki transformer (MarianMT), our approach involves fine-tuning on a self-scraped, purely literary Arabic dataset. Evaluations against Google Translate show consistent outperformance in qualitative assessme...
http://arxiv.org/pdf/2310.13613v2
2023-10-25T17:41:39
2023-10-20T16:03:33
{ "title": [ -0.02418518, 0.0051612854, -0.029144287, 0.0060768127, -0.020950317, -0.018051147, -0.022979736, -0.046905518, -0.030258179, 0.04055786, 0.010803223, 0.0013198853, 0.0519104, -0.068481445, -0.0048217773, 0.0087509155, 0.0052757263, 0...
http://arxiv.org/abs/2310.13610v1
Make Your Decision Convincing! A Unified Two-Stage Framework: Self-Attribution and Decision-Making
Yanrui Du, Sendong Zhao, Haochun Wang, Yuhan Chen, Rui Bai, Zewen Qiang, Muzhen Cai, Bing Qin
cs.CL, cs.AI
Explaining black-box model behavior with natural language has achieved impressive results in various NLP tasks. Recent research has explored the utilization of subsequences from the input text as a rationale, providing users with evidence to support the model decision. Although existing frameworks excel in generating h...
http://arxiv.org/pdf/2310.13610v1
2023-10-20T15:59:57
2023-10-20T15:59:57
{ "title": [ 0.019958496, -0.012908936, -0.032287598, -0.05722046, 0.016098022, -0.031921387, 0.004470825, -0.009590149, -0.06665039, 0.013084412, 0.004096985, 0.01914978, -0.057006836, -0.042877197, 0.0042419434, -0.023651123, 0.0011358261, 0.03...
http://arxiv.org/abs/2310.13606v1
MULTITuDE: Large-Scale Multilingual Machine-Generated Text Detection Benchmark
Dominik Macko, Robert Moro, Adaku Uchendu, Jason Samuel Lucas, Michiharu Yamashita, Matúš Pikuliak, Ivan Srba, Thai Le, Dongwon Lee, Jakub Simko, Maria Bielikova
cs.CL, cs.AI
There is a lack of research into capabilities of recent LLMs to generate convincing text in languages other than English and into performance of detectors of machine-generated text in multilingual settings. This is also reflected in the available benchmarks which lack authentic texts in languages other than English and...
http://arxiv.org/pdf/2310.13606v1
2023-10-20T15:57:17
2023-10-20T15:57:17
{ "title": [ -0.030700684, -0.003232956, -0.022491455, -0.035888672, 0.005908966, -0.0024299622, -0.027008057, -0.015579224, 0.0028305054, 0.042999268, -0.03100586, 0.022628784, 0.003818512, 0.0020179749, -0.04916382, 0.0029010773, -0.019836426, ...
http://arxiv.org/abs/2310.13596v1
MarineGPT: Unlocking Secrets of Ocean to the Public
Ziqiang Zheng, Jipeng Zhang, Tuan-Anh Vu, Shizhe Diao, Yue Him Wong Tim, Sai-Kit Yeung
cs.CL, cs.AI
Large language models (LLMs), such as ChatGPT/GPT-4, have proven to be powerful tools in promoting the user experience as an AI assistant. The continuous works are proposing multi-modal large language models (MLLM), empowering LLMs with the ability to sense multiple modality inputs through constructing a joint semantic...
http://arxiv.org/pdf/2310.13596v1
2023-10-20T15:45:39
2023-10-20T15:45:39
{ "title": [ -0.004940033, -0.02885437, -0.03955078, 0.025115967, -0.042266846, -0.034362793, 0.0053520203, -0.03515625, -0.014228821, 0.009552002, -0.046783447, -0.0054397583, 0.001914978, -0.039367676, -0.0010194778, -0.036193848, -0.0039043427, ...
http://arxiv.org/abs/2310.13588v2
Simultaneous Machine Translation with Tailored Reference
Shoutao Guo, Shaolei Zhang, Yang Feng
cs.CL
Simultaneous machine translation (SiMT) generates translation while reading the whole source sentence. However, existing SiMT models are typically trained using the same reference disregarding the varying amounts of available source information at different latency. Training the model with ground-truth at low latency m...
http://arxiv.org/pdf/2310.13588v2
2023-10-26T03:17:37
2023-10-20T15:32:26
{ "title": [ -0.041259766, -0.015304565, -0.01474762, -0.07879639, -0.016159058, 0.023391724, -0.05480957, 0.013504028, -0.012306213, 0.068115234, 0.003578186, -0.00064754486, 0.035339355, -0.015129089, -0.047698975, 0.02619934, -0.0034885406, 0....
http://arxiv.org/abs/2310.13583v1
Improving Cross-Lingual Transfer through Subtree-Aware Word Reordering
Ofir Arviv, Dmitry Nikolaev, Taelin Karidi, Omri Abend
cs.CL, cs.LG
Despite the impressive growth of the abilities of multilingual language models, such as XLM-R and mT5, it has been shown that they still face difficulties when tackling typologically-distant languages, particularly in the low-resource setting. One obstacle for effective cross-lingual transfer is variability in word-ord...
http://arxiv.org/pdf/2310.13583v1
2023-10-20T15:25:53
2023-10-20T15:25:53
{ "title": [ -0.020843506, -0.003255844, -0.015419006, -0.036712646, -0.0072784424, 0.007904053, -0.04840088, -0.018630981, 0.008140564, 0.054534912, -0.0077438354, 0.013206482, 0.04437256, -0.016799927, -0.042816162, 0.018203735, -0.011566162, 0...
http://arxiv.org/abs/2310.13575v1
Semantic Decomposition of Question and SQL for Text-to-SQL Parsing
Ben Eyal, Amir Bachar, Ophir Haroche, Moran Mahabi, Michael Elhadad
cs.CL
Text-to-SQL semantic parsing faces challenges in generalizing to cross-domain and complex queries. Recent research has employed a question decomposition strategy to enhance the parsing of complex SQL queries. However, this strategy encounters two major obstacles: (1) existing datasets lack question decomposition; (2) d...
http://arxiv.org/pdf/2310.13575v1
2023-10-20T15:13:34
2023-10-20T15:13:34
{ "title": [ -0.019073486, -0.007965088, -0.040527344, -0.085876465, 0.01928711, 0.017166138, -0.03668213, 0.025604248, 0.0151901245, 0.03564453, -0.042236328, 0.014328003, 0.011222839, -0.0039253235, -0.04763794, -0.012580872, 0.0009775162, 0.02...
http://arxiv.org/abs/2310.13571v2
Why Can Large Language Models Generate Correct Chain-of-Thoughts?
Rasul Tutunov, Antoine Grosnit, Juliusz Ziomek, Jun Wang, Haitham Bou-Ammar
cs.CL
This paper delves into the capabilities of large language models (LLMs), specifically focusing on advancing the theoretical comprehension of chain-of-thought prompting. We investigate how LLMs can be effectively induced to generate a coherent chain of thoughts. To achieve this, we introduce a two-level hierarchical gra...
http://arxiv.org/pdf/2310.13571v2
2023-10-30T09:47:39
2023-10-20T15:09:46
{ "title": [ 0.018997192, -0.025375366, 0.011222839, -0.03213501, -0.05444336, 0.006454468, -0.05227661, -0.012260437, -0.0021858215, 0.026184082, -0.021636963, -0.029205322, 0.008857727, -0.03668213, -0.010879517, -0.016983032, -0.01474762, 0.03...
http://arxiv.org/abs/2310.13566v1
Retrieval-Augmented Neural Response Generation Using Logical Reasoning and Relevance Scoring
Nicholas Thomas Walker, Stefan Ultes, Pierre Lison
cs.CL, cs.AI
Constructing responses in task-oriented dialogue systems typically relies on information sources such the current dialogue state or external databases. This paper presents a novel approach to knowledge-grounded response generation that combines retrieval-augmented language models with logical reasoning. The approach re...
http://arxiv.org/pdf/2310.13566v1
2023-10-20T15:05:18
2023-10-20T15:05:18
{ "title": [ -0.0069999695, -0.0010919571, -0.031097412, -0.029876709, -0.010437012, -0.009902954, -0.023712158, -0.008163452, -0.005039215, 0.039093018, -0.000022530556, -0.007820129, -0.027374268, -0.023498535, -0.07293701, -0.0029754639, 0.0040168...
http://arxiv.org/abs/2310.13561v1
Cache & Distil: Optimising API Calls to Large Language Models
Guillem Ramírez, Matthias Lindemann, Alexandra Birch, Ivan Titov
cs.CL, cs.LG
Large-scale deployment of generative AI tools often depends on costly API calls to a Large Language Model (LLM) to fulfil user queries. To curtail the frequency of these calls, one can employ a smaller language model -- a student -- which is continuously trained on the responses of the LLM. This student gradually gains...
http://arxiv.org/pdf/2310.13561v1
2023-10-20T15:01:55
2023-10-20T15:01:55
{ "title": [ -0.026901245, -0.018981934, -0.057891846, -0.09442139, 0.01751709, 0.0039749146, -0.06213379, 0.016860962, -0.0016775131, 0.028961182, -0.021438599, -0.019699097, 0.018112183, -0.022216797, -0.035217285, -0.0030727386, 0.0011806488, ...
http://arxiv.org/abs/2310.13552v2
Self-prompted Chain-of-Thought on Large Language Models for Open-domain Multi-hop Reasoning
Jinyuan Wang, Junlong Li, Hai Zhao
cs.CL, cs.AI
In open-domain question-answering (ODQA), most existing questions require single-hop reasoning on commonsense. To further extend this task, we officially introduce open-domain multi-hop reasoning (ODMR) by answering multi-hop questions with explicit reasoning steps in open-domain setting. Recently, large language model...
http://arxiv.org/pdf/2310.13552v2
2023-10-23T05:42:42
2023-10-20T14:51:10
{ "title": [ -0.008087158, -0.020889282, -0.028411865, -0.061584473, -0.02003479, 0.007980347, -0.04864502, 0.018707275, 0.014915466, 0.011238098, -0.027130127, 0.0068473816, 0.00023448467, -0.025024414, -0.017318726, 0.0020256042, 0.0047454834, ...
http://arxiv.org/abs/2310.13549v1
The Perils & Promises of Fact-checking with Large Language Models
Dorian Quelle, Alexandre Bovet
cs.CL, cs.CY, cs.HC
Autonomous fact-checking, using machine learning to verify claims, has grown vital as misinformation spreads beyond human fact-checking capacity. Large Language Models (LLMs) like GPT-4 are increasingly trusted to verify information and write academic papers, lawsuits, and news articles, emphasizing their role in disce...
http://arxiv.org/pdf/2310.13549v1
2023-10-20T14:49:47
2023-10-20T14:49:47
{ "title": [ -0.012565613, -0.009735107, -0.034942627, -0.08685303, 0.0069351196, -0.00043821335, -0.040252686, -0.0027179718, -0.0340271, 0.02609253, -0.03604126, 0.008369446, 0.009712219, -0.016555786, -0.03062439, -0.011657715, -0.021530151, 0...
http://arxiv.org/abs/2310.13548v3
Towards Understanding Sycophancy in Language Models
Mrinank Sharma, Meg Tong, Tomasz Korbak, David Duvenaud, Amanda Askell, Samuel R. Bowman, Newton Cheng, Esin Durmus, Zac Hatfield-Dodds, Scott R. Johnston, Shauna Kravec, Timothy Maxwell, Sam McCandlish, Kamal Ndousse, Oliver Rausch, Nicholas Schiefer, Da Yan, Miranda Zhang, Ethan Perez
cs.CL, cs.AI, cs.LG, stat.ML, I.2.6
Human feedback is commonly utilized to finetune AI assistants. But human feedback may also encourage model responses that match user beliefs over truthful ones, a behaviour known as sycophancy. We investigate the prevalence of sycophancy in models whose finetuning procedure made use of human feedback, and the potential...
http://arxiv.org/pdf/2310.13548v3
2023-10-27T17:45:26
2023-10-20T14:46:48
{ "title": [ -0.027893066, -0.020339966, -0.005859375, -0.06762695, -0.026870728, 0.011444092, -0.03955078, 0.017745972, -0.011924744, 0.04046631, -0.02923584, -0.012214661, 0.017242432, -0.013824463, -0.05545044, -0.010009766, -0.036956787, 0.05...
http://arxiv.org/abs/2310.13544v1
A Diachronic Perspective on User Trust in AI under Uncertainty
Shehzaad Dhuliawala, Vilém Zouhar, Mennatallah El-Assady, Mrinmaya Sachan
cs.CL, cs.HC
In a human-AI collaboration, users build a mental model of the AI system based on its reliability and how it presents its decision, e.g. its presentation of system confidence and an explanation of the output. Modern NLP systems are often uncalibrated, resulting in confidently incorrect predictions that undermine user t...
http://arxiv.org/pdf/2310.13544v1
2023-10-20T14:41:46
2023-10-20T14:41:46
{ "title": [ -0.018844604, 0.006149292, -0.057037354, -0.0078086853, 0.019088745, -0.042938232, 0.033203125, -0.010749817, 0.008377075, 0.0036354065, -0.050811768, 0.009483337, -0.020431519, -0.014404297, -0.045959473, -0.013244629, -0.0184021, -...
http://arxiv.org/abs/2310.18333v1
Design-Inclusive Language Models for Responsible Information Access
Veronica Chatrath, Oluwanifemi Bamgbose, Shaina Raza
cs.CL, cs.AI
As the use of large language models (LLMs) increases for everyday tasks, appropriate safeguards must be in place to ensure unbiased and safe output. Recent events highlight ethical concerns around conventionally trained LLMs, leading to overall unsafe user experiences. This motivates the need for responsible LLMs that ...
http://arxiv.org/pdf/2310.18333v1
2023-10-20T14:18:40
2023-10-20T14:18:40
{ "title": [ -0.0262146, -0.012550354, -0.055389404, -0.06689453, 0.0049324036, -0.0025959015, -0.04888916, 0.0158844, -0.00037312508, 0.04586792, -0.023712158, -0.006290436, -0.0005879402, -0.0093688965, -0.05532837, -0.002008438, -0.012542725, ...
http://arxiv.org/abs/2310.13526v1
Controlled Randomness Improves the Performance of Transformer Models
Tobias Deußer, Cong Zhao, Wolfgang Krämer, David Leonhard, Christian Bauckhage, Rafet Sifa
cs.CL, cs.LG
During the pre-training step of natural language models, the main objective is to learn a general representation of the pre-training dataset, usually requiring large amounts of textual data to capture the complexity and diversity of natural language. Contrasting this, in most cases, the size of the data available to so...
http://arxiv.org/pdf/2310.13526v1
2023-10-20T14:12:55
2023-10-20T14:12:55
{ "title": [ 0.0637207, 0.022338867, -0.008529663, -0.027893066, 0.00031661987, -0.03527832, -0.0031051636, -0.039489746, -0.053131104, 0.041809082, 0.005268097, 0.012001038, 0.051513672, -0.05230713, 0.06173706, 0.022872925, 0.009239197, 0.02864...
http://arxiv.org/abs/2310.13522v1
Teaching Language Models to Self-Improve through Interactive Demonstrations
Xiao Yu, Baolin Peng, Michel Galley, Jianfeng Gao, Zhou Yu
cs.CL
The self-improving ability of large language models (LLMs), enabled by prompting them to analyze and revise their own outputs, has garnered significant interest in recent research. However, this ability has been shown to be absent and difficult to learn for smaller models, thus widening the performance gap between stat...
http://arxiv.org/pdf/2310.13522v1
2023-10-20T14:11:04
2023-10-20T14:11:04
{ "title": [ -0.0063552856, 0.0012645721, -0.02168274, -0.050933838, -0.010551453, 0.00097465515, -0.048431396, -0.006187439, -0.0184021, 0.02709961, -0.0018720627, 0.00015199184, -0.032409668, -0.02029419, -0.043914795, -0.008613586, 0.0032234192, ...
http://arxiv.org/abs/2310.13512v2
Improving Question Generation with Multi-level Content Planning
Zehua Xia, Qi Gou, Bowen Yu, Haiyang Yu, Fei Huang, Yongbin Li, Cam-Tu Nguyen
cs.CL
This paper addresses the problem of generating questions from a given context and an answer, specifically focusing on questions that require multi-hop reasoning across an extended context. Previous studies have suggested that key phrase selection is essential for question generation (QG), yet it is still challenging to...
http://arxiv.org/pdf/2310.13512v2
2023-10-23T03:37:22
2023-10-20T13:57:01
{ "title": [ -0.0061149597, -0.010429382, -0.035064697, -0.04940796, -0.00945282, -0.0018377304, -0.017990112, 0.010269165, 0.008148193, -0.002445221, -0.006038666, 0.03302002, -0.03857422, -0.020767212, -0.014137268, 0.005569458, -0.005794525, 0...
http://arxiv.org/abs/2310.13506v1
Explaining Interactions Between Text Spans
Sagnik Ray Choudhury, Pepa Atanasova, Isabelle Augenstein
cs.CL, cs.AI, I.2.7
Reasoning over spans of tokens from different parts of the input is essential for natural language understanding (NLU) tasks such as fact-checking (FC), machine reading comprehension (MRC) or natural language inference (NLI). However, existing highlight-based explanations primarily focus on identifying individual impor...
http://arxiv.org/pdf/2310.13506v1
2023-10-20T13:52:37
2023-10-20T13:52:37
{ "title": [ 0.008003235, -0.0056381226, -0.0015668869, -0.043182373, -0.038726807, -0.010574341, -0.03475952, -0.014099121, -0.014549255, 0.033416748, -0.034088135, 0.0071487427, -0.009559631, -0.018600464, -0.019104004, -0.026824951, 0.0030155182, ...
http://arxiv.org/abs/2310.13505v2
Robust Training for Conversational Question Answering Models with Reinforced Reformulation Generation
Magdalena Kaiser, Rishiraj Saha Roy, Gerhard Weikum
cs.CL, cs.AI, cs.IR
Models for conversational question answering (ConvQA) over knowledge graphs (KGs) are usually trained and tested on benchmarks of gold QA pairs. This implies that training is limited to surface forms seen in the respective datasets, and evaluation is on a small set of held-out questions. Through our proposed framework ...
http://arxiv.org/pdf/2310.13505v2
2023-11-06T08:09:12
2023-10-20T13:51:08
{ "title": [ -0.017288208, 0.01878357, -0.036590576, -0.06677246, -0.0023059845, 0.015174866, -0.07727051, -0.011627197, 0.008102417, 0.036468506, -0.021408081, 0.016616821, -0.006427765, -0.006591797, -0.05709839, 0.0014257431, -0.019226074, 0.0...
http://arxiv.org/abs/2310.13500v1
Analogical Proportions and Creativity: A Preliminary Study
Stergos Afantenos, Henri Prade, Leonardo Cortez Bernardes
cs.CL, cs.AI
Analogical proportions are statements of the form "$a$ is to $b$ as $c$ is to $d$", which expresses that the comparisons of the elements in pair $(a, b)$ and in pair $(c, d)$ yield similar results. Analogical proportions are creative in the sense that given 3 distinct items, the representation of a 4th item $d$, distin...
http://arxiv.org/pdf/2310.13500v1
2023-10-20T13:46:04
2023-10-20T13:46:04
{ "title": [ 0.005016327, -0.0063934326, -0.02104187, -0.02154541, 0.016494751, -0.025726318, 0.023223877, -0.022659302, -0.018707275, 0.03768921, -0.0035800934, -0.000005722046, 0.0037822723, -0.019210815, -0.0362854, 0.0027599335, -0.0103302, 0...
http://arxiv.org/abs/2310.13499v2
DistillCSE: Distilled Contrastive Learning for Sentence Embeddings
Jiahao Xu, Wei Shao, Lihui Chen, Lemao Liu
cs.CL
This paper proposes the DistillCSE framework, which performs contrastive learning under the self-training paradigm with knowledge distillation. The potential advantage of DistillCSE is its self-enhancing feature: using a base model to provide additional supervision signals, a stronger model may be learned through knowl...
http://arxiv.org/pdf/2310.13499v2
2023-10-30T09:23:36
2023-10-20T13:45:59
{ "title": [ 0.009643555, 0.003112793, -0.02180481, -0.04736328, 0.0041656494, -0.007282257, -0.025985718, -0.028274536, 0.039764404, 0.0107421875, -0.036468506, 0.0206604, -0.004447937, -0.013175964, -0.033813477, 0.004047394, -0.023620605, 0.06...
http://arxiv.org/abs/2310.13486v1
Mind the instructions: a holistic evaluation of consistency and interactions in prompt-based learning
Lucas Weber, Elia Bruni, Dieuwke Hupkes
cs.CL, cs.AI
Finding the best way of adapting pre-trained language models to a task is a big challenge in current NLP. Just like the previous generation of task-tuned models (TT), models that are adapted to tasks via in-context-learning (ICL) are robust in some setups but not in others. Here, we present a detailed analysis of which...
http://arxiv.org/pdf/2310.13486v1
2023-10-20T13:25:24
2023-10-20T13:25:24
{ "title": [ -0.014442444, -0.0011692047, -0.043182373, -0.02619934, 0.011856079, -0.022415161, -0.018508911, -0.020645142, -0.008842468, 0.006477356, -0.027191162, -0.000726223, -0.023010254, -0.03112793, -0.030944824, -0.023910522, -0.0074768066, ...
http://arxiv.org/abs/2310.13469v3
Ask Language Model to Clean Your Noisy Translation Data
Quinten Bolding, Baohao Liao, Brandon James Denis, Jun Luo, Christof Monz
cs.CL, cs.AI
Transformer models have demonstrated remarkable performance in neural machine translation (NMT). However, their vulnerability to noisy input poses a significant challenge in practical implementation, where generating clean output from noisy input is crucial. The MTNT dataset is widely used as a benchmark for evaluating...
http://arxiv.org/pdf/2310.13469v3
2023-10-24T16:14:55
2023-10-20T13:05:32
{ "title": [ -0.010345459, 0.011817932, -0.03616333, -0.017120361, -0.045013428, -0.011405945, -0.041290283, -0.00982666, -0.0001938343, 0.06237793, -0.042175293, -0.022766113, 0.042999268, -0.018539429, -0.022109985, -0.0011482239, 0.015640259, ...
http://arxiv.org/abs/2310.18332v1
WordArt Designer: User-Driven Artistic Typography Synthesis using Large Language Models
Jun-Yan He, Zhi-Qi Cheng, Chenyang Li, Jingdong Sun, Wangmeng Xiang, Xianhui Lin, Xiaoyang Kang, Zengke Jin, Yusen Hu, Bin Luo, Yifeng Geng, Xuansong Xie, Jingren Zhou
cs.CL, cs.AI, cs.CV, cs.GR
This paper introduces "WordArt Designer", a user-driven framework for artistic typography synthesis, relying on Large Language Models (LLM). The system incorporates four key modules: the "LLM Engine", "SemTypo", "StyTypo", and "TexTypo" modules. 1) The "LLM Engine", empowered by LLM (e.g., GPT-3.5-turbo), interprets us...
http://arxiv.org/pdf/2310.18332v1
2023-10-20T12:44:44
2023-10-20T12:44:44
{ "title": [ -0.054840088, -0.011413574, -0.043273926, -0.00920105, -0.025115967, -0.0025520325, -0.041381836, -0.018707275, 0.0151901245, 0.045196533, -0.021759033, -0.009811401, 0.041931152, -0.02558899, -0.032226562, -0.004852295, -0.0002837181, ...
http://arxiv.org/abs/2310.13448v1
Steering Large Language Models for Machine Translation with Finetuning and In-Context Learning
Duarte M. Alves, Nuno M. Guerreiro, João Alves, José Pombal, Ricardo Rei, José G. C. de Souza, Pierre Colombo, André F. T. Martins
cs.CL
Large language models (LLMs) are a promising avenue for machine translation (MT). However, current LLM-based MT systems are brittle: their effectiveness highly depends on the choice of few-shot examples and they often require extra post-processing due to overgeneration. Alternatives such as finetuning on translation in...
http://arxiv.org/pdf/2310.13448v1
2023-10-20T12:29:51
2023-10-20T12:29:51
{ "title": [ -0.018875122, -0.014846802, -0.022491455, -0.06921387, -0.005382538, 0.006828308, -0.033691406, 0.017349243, -0.0070495605, 0.049865723, -0.032226562, 0.021697998, 0.019500732, -0.006542206, -0.043151855, 0.009559631, -0.002319336, 0...
http://arxiv.org/abs/2310.13447v2
Multiscale Superpixel Structured Difference Graph Convolutional Network for VL Representation
Siyu Zhang, Yeming Chen, Sirui Cheng, Yaoru Sun, Jun Yang, Lizhi Bai
cs.CV, cs.AI, cs.CL
Within the multimodal field, the key to integrating vision and language lies in establishing a good alignment strategy. Recently, benefiting from the success of self-supervised learning, significant progress has been made in multimodal semantic representation based on pre-trained models for vision and language. However...
http://arxiv.org/pdf/2310.13447v2
2023-10-25T13:14:40
2023-10-20T12:26:04
{ "title": [ 0.0014600754, 0.0090789795, -0.0637207, -0.078308105, -0.016830444, -0.010246277, -0.07446289, 0.017593384, 0.01890564, 0.013381958, -0.022644043, -0.012527466, 0.016067505, -0.051116943, -0.0010623932, -0.016708374, 0.029403687, 0.0...
http://arxiv.org/abs/2310.13440v1
The Past, Present, and Future of Typological Databases in NLP
Emi Baylor, Esther Ploeger, Johannes Bjerva
cs.CL
Typological information has the potential to be beneficial in the development of NLP models, particularly for low-resource languages. Unfortunately, current large-scale typological databases, notably WALS and Grambank, are inconsistent both with each other and with other sources of typological information, such as ling...
http://arxiv.org/pdf/2310.13440v1
2023-10-20T12:01:42
2023-10-20T12:01:42
{ "title": [ -0.01713562, -0.02859497, -0.0005302429, -0.07678223, -0.010810852, 0.025802612, -0.04647827, 0.010345459, 0.017562866, 0.014053345, -0.028747559, 0.017532349, -0.009162903, -0.018051147, -0.036193848, 0.016403198, -0.02053833, 0.051...
http://arxiv.org/abs/2310.13439v1
Self-Consistency of Large Language Models under Ambiguity
Henning Bartsch, Ole Jorgensen, Domenic Rosati, Jason Hoelscher-Obermaier, Jacob Pfau
cs.CL, cs.AI
Large language models (LLMs) that do not give consistent answers across contexts are problematic when used for tasks with expectations of consistency, e.g., question-answering, explanations, etc. Our work presents an evaluation benchmark for self-consistency in cases of under-specification where two or more answers can...
http://arxiv.org/pdf/2310.13439v1
2023-10-20T11:57:56
2023-10-20T11:57:56
{ "title": [ -0.06652832, 0.0057029724, -0.07397461, -0.0637207, 0.011070251, 0.010322571, -0.0826416, 0.018417358, 0.005760193, 0.036193848, 0.0026054382, -0.029205322, 0.0012836456, -0.01687622, -0.06518555, 0.0034980774, -0.009468079, 0.020156...
http://arxiv.org/abs/2310.18331v2
AllTogether: Investigating the Efficacy of Spliced Prompt for Web Navigation using Large Language Models
Jiarun Liu, Wentao Hu, Chunhong Zhang
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have emerged as promising agents for web navigation tasks, interpreting objectives and interacting with web pages. However, the efficiency of spliced prompts for such tasks remains underexplored. We introduces AllTogether, a standardized prompt template that enhances task context representa...
http://arxiv.org/pdf/2310.18331v2
2023-10-31T06:25:59
2023-10-20T11:10:14
{ "title": [ 0.0035152435, -0.02357483, -0.029220581, -0.026947021, -0.01576233, -0.022079468, -0.033813477, -0.018096924, -0.0034618378, 0.03189087, -0.009796143, 0.017303467, 0.008895874, -0.025787354, -0.004196167, 0.018188477, -0.0093307495, ...
http://arxiv.org/abs/2310.13420v1
Conversation Chronicles: Towards Diverse Temporal and Relational Dynamics in Multi-Session Conversations
Jihyoung Jang, Minseong Boo, Hyounghun Kim
cs.CL
In the field of natural language processing, open-domain chatbots have emerged as an important research topic. However, a major limitation of existing open-domain chatbot research is its singular focus on short single-session dialogue, neglecting the potential need for understanding contextual information in multiple c...
http://arxiv.org/pdf/2310.13420v1
2023-10-20T11:06:21
2023-10-20T11:06:21
{ "title": [ 0.005935669, -0.0057258606, -0.047790527, -0.08148193, 0.009399414, -0.012207031, -0.05227661, 0.008758545, 0.0044822693, 0.021560669, -0.0022964478, 0.015945435, -0.021621704, -0.024978638, -0.04714966, -0.009109497, -0.018005371, 0...
http://arxiv.org/abs/2310.13411v1
Towards Enhancing Relational Rules for Knowledge Graph Link Prediction
Shuhan Wu, Huaiyu Wan, Wei Chen, Yuting Wu, Junfeng Shen, Youfang Lin
cs.CL
Graph neural networks (GNNs) have shown promising performance for knowledge graph reasoning. A recent variant of GNN called progressive relational graph neural network (PRGNN), utilizes relational rules to infer missing knowledge in relational digraphs and achieves notable results. However, during reasoning with PRGNN,...
http://arxiv.org/pdf/2310.13411v1
2023-10-20T10:38:28
2023-10-20T10:38:28
{ "title": [ -0.018875122, -0.019348145, -0.062194824, -0.0814209, -0.014556885, 0.00064229965, -0.05532837, 0.030654907, -0.02407837, 0.033172607, -0.00035238266, 0.005493164, 0.0024719238, -0.019302368, -0.02696228, -0.003194809, -0.0072402954, ...
http://arxiv.org/abs/2310.13409v1
Explicit Alignment and Many-to-many Entailment Based Reasoning for Conversational Machine Reading
Yangyang Luo, Shiyu Tian, Caixia Yuan, Xiaojie Wang
cs.CL
Conversational Machine Reading (CMR) requires answering a user's initial question through multi-turn dialogue interactions based on a given document. Although there exist many effective methods, they largely neglected the alignment between the document and the user-provided information, which significantly affects the ...
http://arxiv.org/pdf/2310.13409v1
2023-10-20T10:27:24
2023-10-20T10:27:24
{ "title": [ -0.0287323, -0.018753052, -0.034851074, -0.08178711, -0.019348145, 0.0234375, -0.08111572, 0.027572632, 0.026473999, 0.03817749, -0.0024318695, -0.0066871643, 0.027252197, -0.014465332, -0.049835205, 0.01146698, -0.02079773, 0.007431...
http://arxiv.org/abs/2310.13395v1
Cache me if you Can: an Online Cost-aware Teacher-Student framework to Reduce the Calls to Large Language Models
Ilias Stogiannidis, Stavros Vassos, Prodromos Malakasiotis, Ion Androutsopoulos
cs.CL
Prompting Large Language Models (LLMs) performs impressively in zero- and few-shot settings. Hence, small and medium-sized enterprises (SMEs) that cannot afford the cost of creating large task-specific training datasets, but also the cost of pretraining their own LLMs, are increasingly turning to third-party services t...
http://arxiv.org/pdf/2310.13395v1
2023-10-20T10:05:07
2023-10-20T10:05:07
{ "title": [ -0.0003786087, -0.017089844, -0.025802612, -0.056488037, -0.005126953, -0.011787415, -0.039611816, 0.0041885376, -0.019378662, 0.03010559, -0.021224976, 0.0009083748, 0.022918701, -0.018707275, -0.031082153, -0.019332886, -0.0022392273, ...
http://arxiv.org/abs/2310.13394v1
POSQA: Probe the World Models of LLMs with Size Comparisons
Chang Shu, Jiuzhou Han, Fangyu Liu, Ehsan Shareghi, Nigel Collier
cs.CL, cs.AI, cs.CY
Embodied language comprehension emphasizes that language understanding is not solely a matter of mental processing in the brain but also involves interactions with the physical and social environment. With the explosive growth of Large Language Models (LLMs) and their already ubiquitous presence in our daily lives, it ...
http://arxiv.org/pdf/2310.13394v1
2023-10-20T10:05:01
2023-10-20T10:05:01
{ "title": [ -0.06982422, -0.019851685, -0.062164307, 0.020492554, -0.058380127, -0.023284912, -0.04550171, -0.030944824, -0.034301758, -0.016082764, -0.05090332, 0.009246826, -0.00894165, -0.05178833, 0.04348755, -0.035491943, 0.0105896, 0.00978...
http://arxiv.org/abs/2310.13385v1
Tuna: Instruction Tuning using Feedback from Large Language Models
Haoran Li, Yiran Liu, Xingxing Zhang, Wei Lu, Furu Wei
cs.CL, cs.LG
Instruction tuning of open-source large language models (LLMs) like LLaMA, using direct outputs from more powerful LLMs such as Instruct-GPT and GPT-4, has proven to be a cost-effective way to align model behaviors with human preferences. However, the instruction-tuned model has only seen one response per instruction, ...
http://arxiv.org/pdf/2310.13385v1
2023-10-20T09:55:06
2023-10-20T09:55:06
{ "title": [ -0.0692749, -0.02557373, -0.022872925, -0.046905518, -0.03366089, 0.016098022, -0.07281494, 0.0016326904, 0.0050582886, 0.022476196, -0.03765869, 0.012741089, -0.002658844, -0.0110321045, -0.028076172, -0.0005302429, -0.0074272156, 0...
http://arxiv.org/abs/2310.13380v1
APP: Adaptive Prototypical Pseudo-Labeling for Few-shot OOD Detection
Pei Wang, Keqing He, Yutao Mou, Xiaoshuai Song, Yanan Wu, Jingang Wang, Yunsen Xian, Xunliang Cai, Weiran Xu
cs.CL
Detecting out-of-domain (OOD) intents from user queries is essential for a task-oriented dialogue system. Previous OOD detection studies generally work on the assumption that plenty of labeled IND intents exist. In this paper, we focus on a more practical few-shot OOD setting where there are only a few labeled IND data...
http://arxiv.org/pdf/2310.13380v1
2023-10-20T09:48:52
2023-10-20T09:48:52
{ "title": [ 0.029846191, 0.006111145, 0.0051651, -0.053344727, 0.0016031265, -0.0012626648, 0.0015878677, -0.027862549, -0.019210815, 0.000010371208, -0.017288208, 0.017913818, -0.008598328, -0.03955078, 0.015899658, -0.02053833, 0.025268555, 0....
http://arxiv.org/abs/2310.13377v1
A Human-Robot Mutual Learning System with Affect-Grounded Language Acquisition and Differential Outcomes Training
Alva Markelius, Sofia Sjöberg, Zakaria Lemhauori, Laura Cohen, Martin Bergström, Robert Lowe, Lola Cañamero
cs.RO, cs.AI, cs.CL, cs.HC, I.2.9; I.2.6; I.2.10
This paper presents a novel human-robot interaction setup for robot and human learning of symbolic language for identifying robot homeostatic needs. The robot and human learn to use and respond to the same language symbols that convey homeostatic needs and the stimuli that satisfy the homeostatic needs, respectively. W...
http://arxiv.org/pdf/2310.13377v1
2023-10-20T09:41:31
2023-10-20T09:41:31
{ "title": [ -0.028549194, -0.0025024414, -0.03463745, -0.008529663, -0.019119263, -0.024368286, -0.013328552, -0.0061798096, -0.023956299, 0.028869629, -0.034210205, 0.01739502, -0.009780884, -0.007713318, -0.024627686, -0.029678345, -0.008987427, ...
http://arxiv.org/abs/2310.13362v1
Towards General Error Diagnosis via Behavioral Testing in Machine Translation
Junjie Wu, Lemao Liu, Dit-Yan Yeung
cs.CL, cs.AI, cs.LG, cs.SE
Behavioral testing offers a crucial means of diagnosing linguistic errors and assessing capabilities of NLP models. However, applying behavioral testing to machine translation (MT) systems is challenging as it generally requires human efforts to craft references for evaluating the translation quality of such systems on...
http://arxiv.org/pdf/2310.13362v1
2023-10-20T09:06:41
2023-10-20T09:06:41
{ "title": [ -0.008651733, -0.020721436, -0.0052375793, -0.093566895, -0.013877869, 0.015602112, -0.05734253, 0.034484863, 0.013214111, 0.036193848, -0.03125, 0.0075645447, 0.023010254, -0.012779236, -0.04537964, 0.0037288666, -0.012756348, 0.015...
http://arxiv.org/abs/2310.13361v1
Bridging the Gap between Synthetic and Authentic Images for Multimodal Machine Translation
Wenyu Guo, Qingkai Fang, Dong Yu, Yang Feng
cs.CV, cs.AI, cs.CL
Multimodal machine translation (MMT) simultaneously takes the source sentence and a relevant image as input for translation. Since there is no paired image available for the input sentence in most cases, recent studies suggest utilizing powerful text-to-image generation models to provide image inputs. Nevertheless, syn...
http://arxiv.org/pdf/2310.13361v1
2023-10-20T09:06:30
2023-10-20T09:06:30
{ "title": [ -0.023223877, 0.010551453, -0.006702423, -0.027267456, -0.0039138794, 0.0010433197, -0.019897461, -0.02885437, -0.006958008, 0.04397583, -0.0068855286, 0.015350342, 0.011260986, -0.02520752, -0.029800415, 0.011138916, 0.0005979538, 0...
http://arxiv.org/abs/2310.13348v1
Analyzing Cognitive Plausibility of Subword Tokenization
Lisa Beinborn, Yuval Pinter
cs.CL
Subword tokenization has become the de-facto standard for tokenization, although comparative evaluations of subword vocabulary quality across languages are scarce. Existing evaluation studies focus on the effect of a tokenization algorithm on the performance in downstream tasks, or on engineering criteria such as the c...
http://arxiv.org/pdf/2310.13348v1
2023-10-20T08:25:37
2023-10-20T08:25:37
{ "title": [ -0.017288208, -0.0069274902, -0.013404846, -0.034057617, -0.014450073, -0.0119018555, -0.022888184, -0.036376953, -0.024032593, 0.019882202, -0.033325195, 0.011238098, -0.0018796921, -0.00868988, -0.023345947, -0.016113281, -0.024398804,...
http://arxiv.org/abs/2310.13343v1
Challenges and Contributing Factors in the Utilization of Large Language Models (LLMs)
Xiaoliang Chen, Liangbin Li, Le Chang, Yunhe Huang, Yuxuan Zhao, Yuxiao Zhang, Dinuo Li
cs.CL, cs.AI
With the development of large language models (LLMs) like the GPT series, their widespread use across various application scenarios presents a myriad of challenges. This review initially explores the issue of domain specificity, where LLMs may struggle to provide precise answers to specialized questions within niche fi...
http://arxiv.org/pdf/2310.13343v1
2023-10-20T08:13:36
2023-10-20T08:13:36
{ "title": [ -0.034240723, 0.002210617, -0.057861328, -0.051452637, -0.016143799, -0.024261475, -0.061584473, 0.012557983, -0.013320923, 0.03414917, -0.019363403, 0.0018873215, 0.0134887695, -0.039764404, -0.0035552979, 0.015899658, 0.0023994446, ...
http://arxiv.org/abs/2310.13340v1
Large-Scale and Multi-Perspective Opinion Summarization with Diverse Review Subsets
Han Jiang, Rui Wang, Zhihua Wei, Yu Li, Xinpeng Wang
cs.CL
Opinion summarization is expected to digest larger review sets and provide summaries from different perspectives. However, most existing solutions are deficient in epitomizing extensive reviews and offering opinion summaries from various angles due to the lack of designs for information selection. To this end, we propo...
http://arxiv.org/pdf/2310.13340v1
2023-10-20T08:08:13
2023-10-20T08:08:13
{ "title": [ -0.0050239563, 0.006164551, -0.02746582, -0.06451416, 0.007003784, 0.0006866455, -0.04559326, -0.009666443, 0.01033783, 0.042755127, -0.034423828, 0.013580322, -0.017120361, 0.014343262, -0.066467285, -0.006336212, -0.008888245, 0.01...
http://arxiv.org/abs/2310.13332v1
Democratizing Reasoning Ability: Tailored Learning from Large Language Model
Zhaoyang Wang, Shaohan Huang, Yuxuan Liu, Jiahai Wang, Minghui Song, Zihan Zhang, Haizhen Huang, Furu Wei, Weiwei Deng, Feng Sun, Qi Zhang
cs.CL, cs.AI
Large language models (LLMs) exhibit impressive emergent abilities in natural language processing, but their democratization is hindered due to huge computation requirements and closed-source nature. Recent research on advancing open-source smaller LMs by distilling knowledge from black-box LLMs has obtained promising ...
http://arxiv.org/pdf/2310.13332v1
2023-10-20T07:50:10
2023-10-20T07:50:10
{ "title": [ -0.02720642, -0.018249512, -0.046661377, -0.069885254, -0.01928711, -0.0041122437, -0.050079346, 0.014434814, -0.019256592, 0.00957489, -0.011619568, -0.008132935, -0.029159546, -0.021072388, -0.031799316, -0.023666382, -0.022201538, ...
http://arxiv.org/abs/2310.13321v2
Beyond Hard Samples: Robust and Effective Grammatical Error Correction with Cycle Self-Augmenting
Zecheng Tang, Kaifeng Qi, Juntao Li, Min Zhang
cs.CL
Recent studies have revealed that grammatical error correction methods in the sequence-to-sequence paradigm are vulnerable to adversarial attack, and simply utilizing adversarial examples in the pre-training or post-training process can significantly enhance the robustness of GEC models to certain types of attack witho...
http://arxiv.org/pdf/2310.13321v2
2023-10-23T07:41:09
2023-10-20T07:31:23
{ "title": [ 0.008407593, -0.0032367706, -0.017059326, -0.06060791, -0.0029411316, 0.0032234192, -0.03781128, 0.006958008, 0.004463196, 0.03842163, -0.020492554, 0.022323608, -0.011489868, -0.009277344, -0.06604004, 0.0038089752, -0.022232056, 0....
http://arxiv.org/abs/2310.13316v1
Coarse-to-Fine Dual Encoders are Better Frame Identification Learners
Kaikai An, Ce Zheng, Bofei Gao, Haozhe Zhao, Baobao Chang
cs.CL, cs.AI
Frame identification aims to find semantic frames associated with target words in a sentence. Recent researches measure the similarity or matching score between targets and candidate frames by modeling frame definitions. However, they either lack sufficient representation learning of the definitions or face challenges ...
http://arxiv.org/pdf/2310.13316v1
2023-10-20T07:11:23
2023-10-20T07:11:23
{ "title": [ 0.024169922, 0.016555786, -0.03390503, -0.048583984, 0.010154724, -0.038879395, 0.010765076, -0.016601562, -0.02961731, 0.04421997, -0.03643799, 0.019546509, -0.0017690659, -0.02633667, -0.00957489, -0.007472992, -0.0016670227, 0.041...
http://arxiv.org/abs/2310.13315v1
Zero-Shot Sharpness-Aware Quantization for Pre-trained Language Models
Miaoxi Zhu, Qihuang Zhong, Li Shen, Liang Ding, Juhua Liu, Bo Du, Dacheng Tao
cs.CL
Quantization is a promising approach for reducing memory overhead and accelerating inference, especially in large pre-trained language model (PLM) scenarios. While having no access to original training data due to security and privacy concerns has emerged the demand for zero-shot quantization. Most of the cutting-edge ...
http://arxiv.org/pdf/2310.13315v1
2023-10-20T07:09:56
2023-10-20T07:09:56
{ "title": [ -0.020080566, 0.008460999, -0.03552246, -0.05505371, -0.008277893, -0.0093307495, -0.05706787, -0.0045280457, 0.007129669, 0.06689453, 0.02281189, -0.014717102, 0.018936157, -0.020050049, -0.07525635, 0.025634766, 0.00957489, 0.02067...
http://arxiv.org/abs/2310.13312v1
Exploring the Impact of Corpus Diversity on Financial Pretrained Language Models
Jaeyoung Choe, Keonwoong Noh, Nayeon Kim, Seyun Ahn, Woohwan Jung
cs.CL
Over the past few years, various domain-specific pretrained language models (PLMs) have been proposed and have outperformed general-domain PLMs in specialized areas such as biomedical, scientific, and clinical domains. In addition, financial PLMs have been studied because of the high economic impact of financial data a...
http://arxiv.org/pdf/2310.13312v1
2023-10-20T07:04:08
2023-10-20T07:04:08
{ "title": [ -0.01763916, 0.006828308, -0.062561035, -0.044036865, 0.004749298, -0.02508545, -0.048950195, 0.002067566, -0.013504028, 0.041290283, -0.022628784, -0.0034542084, 0.021057129, -0.009742737, -0.05529785, -0.008659363, -0.016616821, 0....
http://arxiv.org/abs/2310.13307v1
Test-Time Self-Adaptive Small Language Models for Question Answering
Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, Jong C. Park
cs.CL, cs.LG
Recent instruction-finetuned large language models (LMs) have achieved notable performances in various tasks, such as question-answering (QA). However, despite their ability to memorize a vast amount of general knowledge across diverse tasks, they might be suboptimal on specific tasks due to their limited capacity to t...
http://arxiv.org/pdf/2310.13307v1
2023-10-20T06:49:32
2023-10-20T06:49:32
{ "title": [ -0.042999268, -0.020065308, -0.032440186, -0.0524292, -0.01184082, 0.0035381317, -0.044311523, 0.013214111, 0.004283905, 0.033416748, -0.0027370453, -0.004016876, -0.019454956, -0.007881165, -0.06939697, 0.0049209595, 0.0025730133, 0...
http://arxiv.org/abs/2310.13297v1
Decoding the Silent Majority: Inducing Belief Augmented Social Graph with Large Language Model for Response Forecasting
Chenkai Sun, Jinning Li, Yi R. Fung, Hou Pong Chan, Tarek Abdelzaher, ChengXiang Zhai, Heng Ji
cs.CL, cs.AI, cs.LG
Automatic response forecasting for news media plays a crucial role in enabling content producers to efficiently predict the impact of news releases and prevent unexpected negative outcomes such as social conflict and moral injury. To effectively forecast responses, it is essential to develop measures that leverage the ...
http://arxiv.org/pdf/2310.13297v1
2023-10-20T06:17:02
2023-10-20T06:17:02
{ "title": [ 0.0006380081, -0.0055160522, -0.029724121, -0.05871582, -0.016296387, 0.0042152405, -0.053009033, 0.00630188, -0.017807007, 0.028640747, -0.005252838, 0.02027893, -0.026657104, -0.016067505, -0.041870117, -0.019088745, -0.005466461, ...
http://arxiv.org/abs/2310.13291v1
Assessing Privacy Risks in Language Models: A Case Study on Summarization Tasks
Ruixiang Tang, Gord Lueck, Rodolfo Quispe, Huseyin A Inan, Janardhan Kulkarni, Xia Hu
cs.CL, cs.AI, cs.LG
Large language models have revolutionized the field of NLP by achieving state-of-the-art performance on various tasks. However, there is a concern that these models may disclose information in the training data. In this study, we focus on the summarization task and investigate the membership inference (MI) attack: give...
http://arxiv.org/pdf/2310.13291v1
2023-10-20T05:44:39
2023-10-20T05:44:39
{ "title": [ -0.015357971, -0.016098022, -0.03137207, -0.057525635, -0.011810303, -0.0012712479, -0.056488037, 0.0062713623, -0.013801575, 0.03866577, -0.016021729, 0.025527954, 0.005077362, -0.0024433136, -0.03753662, 0.0070877075, -0.010154724, ...
http://arxiv.org/abs/2310.13290v1
Interpreting Indirect Answers to Yes-No Questions in Multiple Languages
Zijie Wang, Md Mosharaf Hossain, Shivam Mathur, Terry Cruz Melo, Kadir Bulut Ozler, Keun Hee Park, Jacob Quintero, MohammadHossein Rezaei, Shreya Nupur Shakya, Md Nayem Uddin, Eduardo Blanco
cs.CL
Yes-no questions expect a yes or no for an answer, but people often skip polar keywords. Instead, they answer with long explanations that must be interpreted. In this paper, we focus on this challenging problem and release new benchmarks in eight languages. We present a distant supervision approach to collect training ...
http://arxiv.org/pdf/2310.13290v1
2023-10-20T05:43:33
2023-10-20T05:43:33
{ "title": [ -0.027694702, -0.0071258545, -0.0032615662, -0.051971436, -0.017105103, -0.00076293945, -0.023803711, -0.015853882, -0.026321411, 0.032104492, -0.011161804, 0.0038013458, -0.025253296, -0.024108887, -0.040649414, -0.02456665, 0.0104599, ...
http://arxiv.org/abs/2310.13289v1
SALMONN: Towards Generic Hearing Abilities for Large Language Models
Changli Tang, Wenyi Yu, Guangzhi Sun, Xianzhao Chen, Tian Tan, Wei Li, Lu Lu, Zejun Ma, Chao Zhang
cs.SD, cs.CL, eess.AS
Hearing is arguably an essential ability of artificial intelligence (AI) agents in the physical world, which refers to the perception and understanding of general auditory information consisting of at least three types of sounds: speech, audio events, and music. In this paper, we propose SALMONN, a speech audio languag...
http://arxiv.org/pdf/2310.13289v1
2023-10-20T05:41:57
2023-10-20T05:41:57
{ "title": [ -0.031921387, -0.0026836395, -0.045074463, -0.07128906, -0.013458252, 0.011100769, -0.056610107, 0.02494812, 0.027816772, 0.03564453, -0.015426636, -0.025482178, 0.034606934, -0.020980835, -0.049987793, 0.025238037, -0.019485474, 0.0...
http://arxiv.org/abs/2310.13276v2
InvGC: Robust Cross-Modal Retrieval by Inverse Graph Convolution
Xiangru Jian, Yimu Wang
cs.CV, cs.CL, cs.LG, cs.MM
Over recent decades, significant advancements in cross-modal retrieval are mainly driven by breakthroughs in visual and linguistic modeling. However, a recent study shows that multi-modal data representations tend to cluster within a limited convex cone (as representation degeneration problem), which hinders retrieval ...
http://arxiv.org/pdf/2310.13276v2
2023-10-25T00:46:42
2023-10-20T04:45:44
{ "title": [ 0.0061950684, 0.0029411316, -0.03439331, -0.040649414, 0.017684937, -0.016357422, -0.020431519, -0.035491943, 0.023849487, 0.042144775, -0.032989502, 0.024810791, 0.017745972, -0.011512756, -0.037475586, 0.014884949, -0.0076141357, 0...
http://arxiv.org/abs/2310.13267v1
On the Language Encoder of Contrastive Cross-modal Models
Mengjie Zhao, Junya Ono, Zhi Zhong, Chieh-Hsin Lai, Yuhta Takida, Naoki Murata, Wei-Hsiang Liao, Takashi Shibuya, Hiromi Wakaki, Yuki Mitsufuji
cs.CL, cs.CV, cs.LG, cs.SD, eess.AS
Contrastive cross-modal models such as CLIP and CLAP aid various vision-language (VL) and audio-language (AL) tasks. However, there has been limited investigation of and improvement in their language encoder, which is the central component of encoding natural language descriptions of image/audio into vector representat...
http://arxiv.org/pdf/2310.13267v1
2023-10-20T04:21:09
2023-10-20T04:21:09
{ "title": [ -0.0012016296, 0.005542755, -0.0053749084, -0.048339844, -0.00018632412, 0.00072288513, -0.041259766, -0.032440186, 0.0418396, 0.043792725, -0.023742676, 0.008338928, 0.02243042, -0.0082092285, -0.04598999, 0.024917603, -0.015945435, ...
http://arxiv.org/abs/2310.13265v1
MoqaGPT : Zero-Shot Multi-modal Open-domain Question Answering with Large Language Model
Le Zhang, Yihong Wu, Fengran Mo, Jian-Yun Nie, Aishwarya Agrawal
cs.CL
Multi-modal open-domain question answering typically requires evidence retrieval from databases across diverse modalities, such as images, tables, passages, etc. Even Large Language Models (LLMs) like GPT-4 fall short in this task. To enable LLMs to tackle the task in a zero-shot manner, we introduce MoqaGPT, a straigh...
http://arxiv.org/pdf/2310.13265v1
2023-10-20T04:09:36
2023-10-20T04:09:36
{ "title": [ -0.0014867783, -0.0046920776, -0.02645874, -0.064819336, -0.0009551048, 0.003921509, -0.04473877, -0.0065231323, 0.017166138, 0.033050537, -0.0041160583, 0.0033855438, 0.0059814453, -0.012840271, -0.050720215, 0.014541626, -0.0034656525,...
http://arxiv.org/abs/2310.13262v1
A Quality-based Syntactic Template Retriever for Syntactically-controlled Paraphrase Generation
Xue Zhang, Songming Zhang, Yunlong Liang, Yufeng Chen, Jian Liu, Wenjuan Han, Jinan Xu
cs.CL
Existing syntactically-controlled paraphrase generation (SPG) models perform promisingly with human-annotated or well-chosen syntactic templates. However, the difficulty of obtaining such templates actually hinders the practical application of SPG models. For one thing, the prohibitive cost makes it unfeasible to manua...
http://arxiv.org/pdf/2310.13262v1
2023-10-20T03:55:39
2023-10-20T03:55:39
{ "title": [ -0.012191772, -0.010688782, 0.0007596016, -0.07861328, -0.022003174, 0.034942627, -0.09094238, -0.0054626465, -0.02166748, 0.06213379, -0.0024642944, 0.023345947, 0.027938843, -0.0068588257, -0.035217285, 0.010871887, -0.025009155, 0...
http://arxiv.org/abs/2310.13257v1
Visual Grounding Helps Learn Word Meanings in Low-Data Regimes
Chengxu Zhuang, Evelina Fedorenko, Jacob Andreas
cs.CL, cs.AI
Modern neural language models (LMs) are powerful tools for modeling human sentence production and comprehension, and their internal representations are remarkably well-aligned with representations of language in the human brain. But to achieve these results, LMs must be trained in distinctly un-human-like ways -- requi...
http://arxiv.org/pdf/2310.13257v1
2023-10-20T03:33:36
2023-10-20T03:33:36
{ "title": [ 0.033081055, 0.002363205, -0.044891357, -0.0057525635, -0.032348633, -0.059173584, 0.013267517, -0.020431519, 0.024810791, 0.039794922, -0.039245605, -0.0066566467, 0.039489746, -0.03161621, -0.028945923, -0.0050849915, 0.012649536, ...