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/2304.15010v1
LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Model
Peng Gao, Jiaming Han, Renrui Zhang, Ziyi Lin, Shijie Geng, Aojun Zhou, Wei Zhang, Pan Lu, Conghui He, Xiangyu Yue, Hongsheng Li, Yu Qiao
cs.CV, cs.AI, cs.CL, cs.LG, cs.MM
How to efficiently transform large language models (LLMs) into instruction followers is recently a popular research direction, while training LLM for multi-modal reasoning remains less explored. Although the recent LLaMA-Adapter demonstrates the potential to handle visual inputs with LLMs, it still cannot generalize we...
http://arxiv.org/pdf/2304.15010v1
2023-04-28T17:59:25
2023-04-28T17:59:25
{ "title": [ -0.031204224, 0.003660202, -0.030715942, -0.032806396, -0.007713318, -0.0038661957, -0.042633057, -0.038726807, -0.0045280457, 0.019851685, -0.025222778, 0.0026054382, -0.004398346, -0.032348633, -0.010025024, 0.011520386, 0.013755798, ...
http://arxiv.org/abs/2304.14999v1
Empirical Analysis of the Strengths and Weaknesses of PEFT Techniques for LLMs
George Pu, Anirudh Jain, Jihan Yin, Russell Kaplan
cs.CL, cs.AI
As foundation models continue to exponentially scale in size, efficient methods of adaptation become increasingly critical. Parameter-efficient fine-tuning (PEFT), a recent class of techniques that require only modifying a small percentage of the model parameters, is currently the most popular method for adapting large...
http://arxiv.org/pdf/2304.14999v1
2023-04-28T17:39:49
2023-04-28T17:39:49
{ "title": [ 0.01360321, 0.0014371872, -0.004360199, 0.031341553, -0.020614624, -0.028945923, 0.011367798, -0.0340271, -0.040161133, 0.050720215, -0.0041542053, 0.02885437, 0.032470703, -0.045654297, 0.023483276, 0.011131287, -0.012840271, 0.0356...
http://arxiv.org/abs/2304.14986v2
Interpreting Vision and Language Generative Models with Semantic Visual Priors
Michele Cafagna, Lina M. Rojas-Barahona, Kees van Deemter, Albert Gatt
cs.CV, cs.AI, cs.CL
When applied to Image-to-text models, interpretability methods often provide token-by-token explanations namely, they compute a visual explanation for each token of the generated sequence. Those explanations are expensive to compute and unable to comprehensively explain the model's output. Therefore, these models often...
http://arxiv.org/pdf/2304.14986v2
2023-05-04T15:25:00
2023-04-28T17:10:08
{ "title": [ -0.029907227, -0.010612488, -0.03100586, -0.05557251, -0.024108887, -0.0022201538, -0.0637207, -0.010017395, 0.013824463, 0.008483887, -0.027114868, -0.011360168, -0.010574341, -0.017990112, -0.039001465, -0.004928589, 0.003648758, 0...
http://arxiv.org/abs/2304.14953v2
CCpdf: Building a High Quality Corpus for Visually Rich Documents from Web Crawl Data
Michał Turski, Tomasz Stanisławek, Karol Kaczmarek, Paweł Dyda, Filip Graliński
cs.CL
In recent years, the field of document understanding has progressed a lot. A significant part of this progress has been possible thanks to the use of language models pretrained on large amounts of documents. However, pretraining corpora used in the domain of document understanding are single domain, monolingual, or non...
http://arxiv.org/pdf/2304.14953v2
2023-06-06T07:35:17
2023-04-28T16:12:18
{ "title": [ -0.015312195, -0.017044067, -0.049072266, -0.06317139, 0.005180359, -0.0146484375, -0.02420044, 0.0010604858, 0.036193848, 0.037597656, -0.039733887, 0.010475159, 0.024505615, 0.0023021698, -0.05633545, 0.007247925, -0.001660347, 0.0...
http://arxiv.org/abs/2304.14936v1
Information Redundancy and Biases in Public Document Information Extraction Benchmarks
Seif Laatiri, Pirashanth Ratnamogan, Joel Tang, Laurent Lam, William Vanhuffel, Fabien Caspani
cs.CL, cs.AI
Advances in the Visually-rich Document Understanding (VrDU) field and particularly the Key-Information Extraction (KIE) task are marked with the emergence of efficient Transformer-based approaches such as the LayoutLM models. Despite the good performance of KIE models when fine-tuned on public benchmarks, they still st...
http://arxiv.org/pdf/2304.14936v1
2023-04-28T15:48:26
2023-04-28T15:48:26
{ "title": [ -0.023498535, 0.019348145, -0.047668457, -0.03704834, 0.002729416, -0.013664246, -0.027786255, -0.03375244, -0.0078086853, 0.040283203, -0.048583984, 0.019470215, -0.012298584, 0.0030288696, -0.037506104, -0.024627686, -0.0033111572, ...
http://arxiv.org/abs/2304.14933v2
An Empirical Study of Multimodal Model Merging
Yi-Lin Sung, Linjie Li, Kevin Lin, Zhe Gan, Mohit Bansal, Lijuan Wang
cs.CV, cs.AI, cs.CL, cs.LG
Model merging (e.g., via interpolation or task arithmetic) fuses multiple models trained on different tasks to generate a multi-task solution. The technique has been proven successful in previous studies, where the models are trained on similar tasks and with the same initialization. In this paper, we expand on this co...
http://arxiv.org/pdf/2304.14933v2
2023-10-11T15:08:51
2023-04-28T15:43:21
{ "title": [ -0.028701782, 0.0051345825, -0.058807373, -0.072753906, 0.01928711, 0.0020446777, -0.07208252, -0.014564514, -0.00080156326, 0.035583496, -0.022750854, -0.0052375793, -0.0018415451, -0.015434265, -0.030822754, -0.0050735474, -0.012657165...
http://arxiv.org/abs/2304.14931v2
HQP: A Human-Annotated Dataset for Detecting Online Propaganda
Abdurahman Maarouf, Dominik Bär, Dominique Geissler, Stefan Feuerriegel
cs.CL
Online propaganda poses a severe threat to the integrity of societies. However, existing datasets for detecting online propaganda have a key limitation: they were annotated using weak labels that can be noisy and even incorrect. To address this limitation, our work makes the following contributions: (1) We present HQP:...
http://arxiv.org/pdf/2304.14931v2
2023-05-01T08:29:51
2023-04-28T15:42:55
{ "title": [ 0.029373169, 0.016174316, -0.014724731, -0.037841797, 0.032196045, -0.020248413, 0.016448975, -0.03237915, -0.036621094, 0.033447266, -0.0012712479, 0.030273438, -0.0032577515, -0.017105103, -0.012138367, -0.00409317, -0.024536133, 0...
http://arxiv.org/abs/2304.14827v2
ChatGPT Evaluation on Sentence Level Relations: A Focus on Temporal, Causal, and Discourse Relations
Chunkit Chan, Jiayang Cheng, Weiqi Wang, Yuxin Jiang, Tianqing Fang, Xin Liu, Yangqiu Song
cs.CL
This paper aims to quantitatively evaluate the performance of ChatGPT, an interactive large language model, on inter-sentential relations such as temporal relations, causal relations, and discourse relations. Given ChatGPT's promising performance across various tasks, we conduct extensive evaluations on the whole test ...
http://arxiv.org/pdf/2304.14827v2
2023-05-11T18:20:00
2023-04-28T13:14:36
{ "title": [ 0.011161804, -0.005973816, -0.028060913, -0.0748291, -0.004558563, -0.002626419, -0.05126953, 0.010894775, -0.010879517, 0.018951416, 0.013832092, 0.009254456, 0.012176514, -0.03643799, -0.010108948, 0.009185791, -0.04147339, 0.06628...
http://arxiv.org/abs/2304.14803v1
SemEval-2023 Task 11: Learning With Disagreements (LeWiDi)
Elisa Leonardelli, Alexandra Uma, Gavin Abercrombie, Dina Almanea, Valerio Basile, Tommaso Fornaciari, Barbara Plank, Verena Rieser, Massimo Poesio
cs.CL
NLP datasets annotated with human judgments are rife with disagreements between the judges. This is especially true for tasks depending on subjective judgments such as sentiment analysis or offensive language detection. Particularly in these latter cases, the NLP community has come to realize that the approach of 'reco...
http://arxiv.org/pdf/2304.14803v1
2023-04-28T12:20:35
2023-04-28T12:20:35
{ "title": [ -0.00566864, -0.016311646, -0.04083252, -0.029754639, -0.025558472, -0.022827148, -0.029312134, -0.0059814453, -0.036132812, 0.0011405945, -0.009254456, -0.0030021667, -0.038604736, -0.054534912, 0.027694702, -0.042144775, -0.010635376, ...
http://arxiv.org/abs/2304.14802v1
ResiDual: Transformer with Dual Residual Connections
Shufang Xie, Huishuai Zhang, Junliang Guo, Xu Tan, Jiang Bian, Hany Hassan Awadalla, Arul Menezes, Tao Qin, Rui Yan
cs.CL, cs.AI, cs.LG, cs.NE
Transformer networks have become the preferred architecture for many tasks due to their state-of-the-art performance. However, the optimal way to implement residual connections in Transformer, which are essential for effective training, is still debated. Two widely used variants are the Post-Layer-Normalization (Post-L...
http://arxiv.org/pdf/2304.14802v1
2023-04-28T12:19:47
2023-04-28T12:19:47
{ "title": [ 0.0657959, 0.008506775, -0.013259888, -0.04159546, -0.029144287, -0.019943237, -0.02142334, -0.029037476, -0.040222168, 0.034484863, 0.0015182495, 0.005554199, 0.05822754, -0.048858643, 0.046661377, -0.00040125847, 0.02293396, 0.0245...
http://arxiv.org/abs/2304.14796v1
Are the Best Multilingual Document Embeddings simply Based on Sentence Embeddings?
Sonal Sannigrahi, Josef van Genabith, Cristina Espana-Bonet
cs.CL, cs.IR
Dense vector representations for textual data are crucial in modern NLP. Word embeddings and sentence embeddings estimated from raw texts are key in achieving state-of-the-art results in various tasks requiring semantic understanding. However, obtaining embeddings at the document level is challenging due to computation...
http://arxiv.org/pdf/2304.14796v1
2023-04-28T12:11:21
2023-04-28T12:11:21
{ "title": [ -0.028167725, -0.021011353, -0.015434265, -0.019165039, -0.05581665, 0.007320404, -0.069885254, -0.013198853, -0.0022258759, 0.024917603, -0.0030326843, -0.017318726, 0.010696411, -0.03439331, 0.009391785, -0.0049972534, -0.015457153, ...
http://arxiv.org/abs/2304.14780v1
Training and Evaluation of a Multilingual Tokenizer for GPT-SW3
Felix Stollenwerk
cs.CL, cs.AI
This paper provides a detailed discussion of the multilingual tokenizer used for GPT-SW3. It was trained on the Nordic Pile using the SentencePiece library and the BPE algorithm. We outline the tokenizer's most important features and share details on its learned vocabulary. In addition, we systematically analyze the pr...
http://arxiv.org/pdf/2304.14780v1
2023-04-28T11:40:48
2023-04-28T11:40:48
{ "title": [ -0.00074386597, 0.015808105, -0.03173828, -0.050567627, 0.009796143, -0.007446289, -0.06365967, -0.032318115, -0.01776123, 0.028305054, -0.013290405, 0.013694763, -0.012565613, -0.02633667, -0.016235352, -0.001958847, -0.014190674, 0...
http://arxiv.org/abs/2304.14770v2
RexUIE: A Recursive Method with Explicit Schema Instructor for Universal Information Extraction
Chengyuan Liu, Fubang Zhao, Yangyang Kang, Jingyuan Zhang, Xiang Zhou, Changlong Sun, Kun Kuang, Fei Wu
cs.CL
Universal Information Extraction (UIE) is an area of interest due to the challenges posed by varying targets, heterogeneous structures, and demand-specific schemas. However, previous works have only achieved limited success by unifying a few tasks, such as Named Entity Recognition (NER) and Relation Extraction (RE), wh...
http://arxiv.org/pdf/2304.14770v2
2023-10-18T02:30:55
2023-04-28T11:28:56
{ "title": [ -0.044189453, -0.011894226, -0.06414795, -0.07922363, 0.0033664703, 0.015777588, -0.06512451, 0.009567261, 0.013999939, 0.027282715, -0.02255249, 0.01739502, -0.0013027191, -0.011329651, -0.025009155, 0.0084991455, -0.020141602, 0.05...
http://arxiv.org/abs/2304.14767v3
Dissecting Recall of Factual Associations in Auto-Regressive Language Models
Mor Geva, Jasmijn Bastings, Katja Filippova, Amir Globerson
cs.CL
Transformer-based language models (LMs) are known to capture factual knowledge in their parameters. While previous work looked into where factual associations are stored, only little is known about how they are retrieved internally during inference. We investigate this question through the lens of information flow. Giv...
http://arxiv.org/pdf/2304.14767v3
2023-10-13T19:01:20
2023-04-28T11:26:17
{ "title": [ -0.007129669, -0.0018758774, -0.04071045, -0.05026245, -0.011917114, -0.012557983, -0.034088135, 0.000014424324, 0.018920898, 0.04647827, -0.014961243, 0.006832123, 0.018432617, -0.00466156, -0.06311035, 0.020462036, -0.015136719, 0....
http://arxiv.org/abs/2304.14746v1
FlowTransformer: A Transformer Framework for Flow-based Network Intrusion Detection Systems
Liam Daly Manocchio, Siamak Layeghy, Wai Weng Lo, Gayan K. Kulatilleke, Mohanad Sarhan, Marius Portmann
cs.CR, cs.AI, cs.CL, cs.NE, cs.NI
This paper presents the FlowTransformer framework, a novel approach for implementing transformer-based Network Intrusion Detection Systems (NIDSs). FlowTransformer leverages the strengths of transformer models in identifying the long-term behaviour and characteristics of networks, which are often overlooked by most exi...
http://arxiv.org/pdf/2304.14746v1
2023-04-28T10:40:34
2023-04-28T10:40:34
{ "title": [ -0.011940002, -0.0024261475, -0.040618896, -0.05886841, 0.013694763, -0.009010315, -0.037750244, -0.005340576, -0.024169922, 0.03842163, -0.022521973, 0.019866943, -0.0053520203, -0.019195557, -0.031677246, -0.00573349, -0.015213013, ...
http://arxiv.org/abs/2304.14745v1
Made of Steel? Learning Plausible Materials for Components in the Vehicle Repair Domain
Annerose Eichel, Helena Schlipf, Sabine Schulte im Walde
cs.CL, cs.AI, cs.IR
We propose a novel approach to learn domain-specific plausible materials for components in the vehicle repair domain by probing Pretrained Language Models (PLMs) in a cloze task style setting to overcome the lack of annotated datasets. We devise a new method to aggregate salient predictions from a set of cloze query te...
http://arxiv.org/pdf/2304.14745v1
2023-04-28T10:39:44
2023-04-28T10:39:44
{ "title": [ -0.029830933, -0.0056533813, -0.07489014, -0.03692627, -0.011062622, -0.041107178, -0.03768921, 0.007896423, -0.04736328, -0.0069885254, -0.049316406, 0.036315918, -0.022735596, -0.029541016, 0.033172607, -0.037841797, -0.015655518, ...
http://arxiv.org/abs/2304.14738v1
Cost-Sensitive Self-Training for Optimizing Non-Decomposable Metrics
Harsh Rangwani, Shrinivas Ramasubramanian, Sho Takemori, Kato Takashi, Yuhei Umeda, Venkatesh Babu Radhakrishnan
cs.LG, cs.CL, cs.CV
Self-training based semi-supervised learning algorithms have enabled the learning of highly accurate deep neural networks, using only a fraction of labeled data. However, the majority of work on self-training has focused on the objective of improving accuracy, whereas practical machine learning systems can have complex...
http://arxiv.org/pdf/2304.14738v1
2023-04-28T10:31:12
2023-04-28T10:31:12
{ "title": [ -0.026000977, -0.005794525, 0.011764526, 0.012527466, -0.028305054, 0.009414673, -0.029876709, -0.06304932, -0.024032593, 0.024658203, 0.01033783, 0.02217102, -0.047088623, -0.030639648, -0.03579712, -0.0032196045, 0.022445679, 0.017...
http://arxiv.org/abs/2304.14732v6
Search-in-the-Chain: Towards Accurate, Credible and Traceable Large Language Models for Knowledge-intensive Tasks
Shicheng Xu, Liang Pang, Huawei Shen, Xueqi Cheng, Tat-Seng Chua
cs.CL
Making the contents generated by Large Language Model (LLM) such as ChatGPT, accurate, credible and traceable is crucial, especially in complex knowledge-intensive tasks that require multi-step reasoning and each of which needs knowledge to solve. Introducing Information Retrieval (IR) to provide LLM with external know...
http://arxiv.org/pdf/2304.14732v6
2023-09-22T08:15:25
2023-04-28T10:15:25
{ "title": [ -0.026916504, -0.031951904, -0.058044434, -0.07531738, 0.0011043549, -0.00093126297, -0.05883789, 0.016586304, 0.012207031, 0.035949707, -0.010467529, -0.0077285767, 0.008872986, -0.017089844, -0.057006836, 0.0022335052, -0.009056091, ...
http://arxiv.org/abs/2304.14721v4
Towards autonomous system: flexible modular production system enhanced with large language model agents
Yuchen Xia, Manthan Shenoy, Nasser Jazdi, Michael Weyrich
cs.RO, cs.CL, cs.SE, cs.SY, eess.SY
In this paper, we present a novel framework that combines large language models (LLMs), digital twins and industrial automation system to enable intelligent planning and control of production processes. We retrofit the automation system for a modular production facility and create executable control interfaces of fine-...
http://arxiv.org/pdf/2304.14721v4
2023-07-24T09:49:55
2023-04-28T09:42:18
{ "title": [ -0.04748535, 0.0069465637, -0.07952881, -0.076293945, 0.029815674, -0.0047683716, -0.051239014, 0.031982422, -0.041381836, 0.03869629, -0.022094727, 0.017593384, -0.014076233, -0.009109497, -0.022537231, -0.023727417, -0.017715454, 0...
http://arxiv.org/abs/2304.14662v1
CED: Catalog Extraction from Documents
Tong Zhu, Guoliang Zhang, Zechang Li, Zijian Yu, Junfei Ren, Mengsong Wu, Zhefeng Wang, Baoxing Huai, Pingfu Chao, Wenliang Chen
cs.CL
Sentence-by-sentence information extraction from long documents is an exhausting and error-prone task. As the indicator of document skeleton, catalogs naturally chunk documents into segments and provide informative cascade semantics, which can help to reduce the search space. Despite their usefulness, catalogs are hard...
http://arxiv.org/pdf/2304.14662v1
2023-04-28T07:32:00
2023-04-28T07:32:00
{ "title": [ -0.03152466, -0.0127334595, -0.064819336, -0.056884766, -0.0049209595, -0.019683838, -0.04071045, -0.009002686, 0.010040283, 0.025482178, -0.020248413, 0.006198883, 0.010971069, -0.032928467, -0.037078857, -0.0017166138, 0.0016298294, ...
http://arxiv.org/abs/2304.14599v1
Antisemitic Messages? A Guide to High-Quality Annotation and a Labeled Dataset of Tweets
Gunther Jikeli, Sameer Karali, Daniel Miehling, Katharina Soemer
cs.CL, cs.CY
One of the major challenges in automatic hate speech detection is the lack of datasets that cover a wide range of biased and unbiased messages and that are consistently labeled. We propose a labeling procedure that addresses some of the common weaknesses of labeled datasets. We focus on antisemitic speech on Twitter an...
http://arxiv.org/pdf/2304.14599v1
2023-04-28T02:52:38
2023-04-28T02:52:38
{ "title": [ 0.04763794, 0.010437012, -0.015899658, -0.033477783, -0.0003426075, -0.01802063, -0.0008416176, -0.013435364, -0.030303955, 0.031402588, 0.00598526, 0.04562378, 0.00030255318, -0.035736084, 0.007675171, -0.011642456, 0.021942139, 0.0...
http://arxiv.org/abs/2304.14590v2
A logical word embedding for learning grammar
Sean Deyo, Veit Elser
cs.CL
We introduce the logical grammar emdebbing (LGE), a model inspired by pregroup grammars and categorial grammars to enable unsupervised inference of lexical categories and syntactic rules from a corpus of text. LGE produces comprehensible output summarizing its inferences, has a completely transparent process for produc...
http://arxiv.org/pdf/2304.14590v2
2023-06-06T00:46:49
2023-04-28T01:53:54
{ "title": [ -0.021484375, -0.0085372925, -0.027542114, -0.050201416, -0.0024776459, -0.0037250519, -0.036987305, 0.01361084, -0.0085372925, 0.01574707, -0.04269409, 0.035461426, -0.020446777, -0.0023517609, -0.030319214, -0.006504059, -0.023620605, ...
http://arxiv.org/abs/2304.14585v1
Improving Knowledge Graph Entity Alignment with Graph Augmentation
Feng Xie, Xiang Zeng, Bin Zhou, Yusong Tan
cs.CL, cs.SI
Entity alignment (EA) which links equivalent entities across different knowledge graphs (KGs) plays a crucial role in knowledge fusion. In recent years, graph neural networks (GNNs) have been successfully applied in many embedding-based EA methods. However, existing GNN-based methods either suffer from the structural h...
http://arxiv.org/pdf/2304.14585v1
2023-04-28T01:22:47
2023-04-28T01:22:47
{ "title": [ -0.032348633, -0.009506226, -0.08569336, -0.06488037, -0.0010223389, -0.008384705, -0.047180176, 0.022735596, 0.01713562, 0.033081055, -0.0072135925, 0.012840271, 0.024230957, -0.018341064, -0.025527954, 0.021743774, -0.0052757263, 0...
http://arxiv.org/abs/2304.14553v1
Appropriateness is all you need!
Hendrik Kempt, Alon Lavie, Saskia K. Nagel
cs.AI, cs.CL, cs.HC
The strive to make AI applications "safe" has led to the development of safety-measures as the main or even sole normative requirement of their permissible use. Similar can be attested to the latest version of chatbots, such as chatGPT. In this view, if they are "safe", they are supposed to be permissible to deploy. Th...
http://arxiv.org/pdf/2304.14553v1
2023-04-27T22:21:52
2023-04-27T22:21:52
{ "title": [ 0.052734375, 0.0014247894, -0.008026123, -0.09716797, 0.014259338, 0.008865356, -0.03945923, 0.015686035, -0.002357483, 0.058654785, 0.0031032562, -0.007980347, 0.02116394, -0.027297974, -0.019210815, -0.018173218, 0.007217407, 0.008...
http://arxiv.org/abs/2304.14543v1
Discourse over Discourse: The Need for an Expanded Pragmatic Focus in Conversational AI
S. M. Seals, Valerie L. Shalin
cs.CL
The summarization of conversation, that is, discourse over discourse, elevates pragmatic considerations as a pervasive limitation of both summarization and other applications of contemporary conversational AI. Building on impressive progress in both semantics and syntax, pragmatics concerns meaning in the practical sen...
http://arxiv.org/pdf/2304.14543v1
2023-04-27T21:51:42
2023-04-27T21:51:42
{ "title": [ -0.005420685, -0.009284973, -0.04385376, -0.042419434, 0.005306244, -0.03338623, -0.0054855347, -0.0014019012, 0.011260986, 0.013511658, -0.026855469, -0.0006723404, -0.008621216, -0.018218994, -0.04046631, -0.008934021, -0.0236969, ...
http://arxiv.org/abs/2304.14535v2
Deep Transfer Learning for Automatic Speech Recognition: Towards Better Generalization
Hamza Kheddar, Yassine Himeur, Somaya Al-Maadeed, Abbes Amira, Faycal Bensaali
cs.SD, cs.CL, cs.LG, eess.AS
Automatic speech recognition (ASR) has recently become an important challenge when using deep learning (DL). It requires large-scale training datasets and high computational and storage resources. Moreover, DL techniques and machine learning (ML) approaches in general, hypothesize that training and testing data come fr...
http://arxiv.org/pdf/2304.14535v2
2023-07-31T11:58:18
2023-04-27T21:08:05
{ "title": [ -0.0059547424, 0.0026779175, -0.047851562, -0.07055664, 0.022567749, -0.0006146431, -0.03250122, 0.00843811, -0.0029258728, 0.038116455, -0.014434814, 0.018737793, 0.0074386597, -0.013328552, -0.028549194, 0.008125305, -0.008552551, ...
http://arxiv.org/abs/2304.14522v1
Multivariate Representation Learning for Information Retrieval
Hamed Zamani, Michael Bendersky
cs.IR, cs.CL, cs.LG
Dense retrieval models use bi-encoder network architectures for learning query and document representations. These representations are often in the form of a vector representation and their similarities are often computed using the dot product function. In this paper, we propose a new representation learning framework ...
http://arxiv.org/pdf/2304.14522v1
2023-04-27T20:30:46
2023-04-27T20:30:46
{ "title": [ 0.01209259, 0.0067367554, -0.031707764, -0.049926758, 0.017120361, -0.012573242, -0.019302368, -0.022521973, 0.016326904, 0.03503418, -0.029403687, 0.0066986084, 0.01008606, -0.014526367, -0.04055786, 0.004962921, -0.0075263977, 0.03...
http://arxiv.org/abs/2304.14514v1
Understanding Shared Speech-Text Representations
Gary Wang, Kyle Kastner, Ankur Bapna, Zhehuai Chen, Andrew Rosenberg, Bhuvana Ramabhadran, Yu Zhang
cs.CL, cs.LG, cs.SD, eess.AS
Recently, a number of approaches to train speech models by incorpo-rating text into end-to-end models have been developed, with Mae-stro advancing state-of-the-art automatic speech recognition (ASR)and Speech Translation (ST) performance. In this paper, we expandour understanding of the resulting shared speech-text rep...
http://arxiv.org/pdf/2304.14514v1
2023-04-27T20:05:36
2023-04-27T20:05:36
{ "title": [ -0.00036096573, 0.0013561249, -0.017562866, -0.073913574, 0.0014028549, 0.0075531006, -0.05999756, 0.004924774, -0.00002270937, 0.05026245, -0.023910522, 0.005744934, 0.00038337708, -0.00623703, -0.04736328, 0.01209259, -0.011131287, ...
http://arxiv.org/abs/2304.14511v1
Visual Referential Games Further the Emergence of Disentangled Representations
Kevin Denamganaï, Sondess Missaoui, James Alfred Walker
cs.CL
Natural languages are powerful tools wielded by human beings to communicate information. Among their desirable properties, compositionality has been the main focus in the context of referential games and variants, as it promises to enable greater systematicity to the agents which would wield it. The concept of disentan...
http://arxiv.org/pdf/2304.14511v1
2023-04-27T20:00:51
2023-04-27T20:00:51
{ "title": [ 0.04055786, 0.009246826, -0.0030097961, -0.013114929, -0.002166748, -0.040100098, 0.033172607, -0.03692627, -0.009506226, 0.032043457, -0.026184082, 0.0063095093, 0.0149383545, -0.023803711, -0.017913818, -0.0057296753, 0.008110046, ...
http://arxiv.org/abs/2304.14456v1
Framing the News:From Human Perception to Large Language Model Inferences
David Alonso del Barrio, Daniel Gatica-Perez
cs.CL, cs.HC
Identifying the frames of news is important to understand the articles' vision, intention, message to be conveyed, and which aspects of the news are emphasized. Framing is a widely studied concept in journalism, and has emerged as a new topic in computing, with the potential to automate processes and facilitate the wor...
http://arxiv.org/pdf/2304.14456v1
2023-04-27T18:30:18
2023-04-27T18:30:18
{ "title": [ -0.01625061, -0.011398315, -0.02671814, -0.05810547, -0.00027513504, -0.0018968582, -0.032043457, -0.01789856, -0.0206604, 0.03466797, -0.01876831, 0.00617218, -0.0029258728, -0.006504059, -0.04574585, -0.003271103, -0.015655518, 0.0...
http://arxiv.org/abs/2304.14454v3
PMC-LLaMA: Towards Building Open-source Language Models for Medicine
Chaoyi Wu, Weixiong Lin, Xiaoman Zhang, Ya Zhang, Yanfeng Wang, Weidi Xie
cs.CL
Recently, Large Language Models (LLMs) have showcased remarkable capabilities in natural language understanding. While demonstrating proficiency in everyday conversations and question-answering situations, these models frequently struggle in domains that require precision, such as medical applications, due to their lac...
http://arxiv.org/pdf/2304.14454v3
2023-08-25T14:08:38
2023-04-27T18:29:05
{ "title": [ -0.045135498, -0.014572144, -0.0368042, -0.06945801, -0.0011606216, 0.02684021, -0.077941895, -0.0082473755, -0.024536133, 0.037353516, 0.02609253, 0.005039215, -0.008857727, -0.035736084, -0.031433105, 0.016326904, -0.008399963, 0.0...
http://arxiv.org/abs/2304.14447v1
Analyzing Vietnamese Legal Questions Using Deep Neural Networks with Biaffine Classifiers
Nguyen Anh Tu, Hoang Thi Thu Uyen, Tu Minh Phuong, Ngo Xuan Bach
cs.CL
In this paper, we propose using deep neural networks to extract important information from Vietnamese legal questions, a fundamental task towards building a question answering system in the legal domain. Given a legal question in natural language, the goal is to extract all the segments that contain the needed informat...
http://arxiv.org/pdf/2304.14447v1
2023-04-27T18:19:24
2023-04-27T18:19:24
{ "title": [ -0.019241333, -0.014190674, -0.0231781, -0.02067566, -0.009468079, -0.010826111, -0.008415222, 0.00012779236, 0.036468506, 0.012840271, -0.035125732, -0.000094771385, -0.014289856, -0.021499634, -0.040802002, -0.005115509, -0.03100586, ...
http://arxiv.org/abs/2304.14405v1
ViMQ: A Vietnamese Medical Question Dataset for Healthcare Dialogue System Development
Ta Duc Huy, Nguyen Anh Tu, Tran Hoang Vu, Nguyen Phuc Minh, Nguyen Phan, Trung H. Bui, Steven Q. H. Truong
cs.CL
Existing medical text datasets usually take the form of ques- tion and answer pairs that support the task of natural language gener- ation, but lacking the composite annotations of the medical terms. In this study, we publish a Vietnamese dataset of medical questions from patients with sentence-level and entity-level a...
http://arxiv.org/pdf/2304.14405v1
2023-04-27T17:59:53
2023-04-27T17:59:53
{ "title": [ -0.012123108, -0.01007843, -0.048980713, -0.066467285, 0.016311646, 0.0004799366, -0.02545166, 0.01939392, 0.0009608269, 0.026885986, -0.014373779, -0.006706238, -0.039855957, -0.02319336, -0.072387695, -0.021331787, -0.0057792664, 0...
http://arxiv.org/abs/2304.14402v2
LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions
Minghao Wu, Abdul Waheed, Chiyu Zhang, Muhammad Abdul-Mageed, Alham Fikri Aji
cs.CL
Large language models (LLMs) with instruction fine-tuning demonstrate superior generative capabilities. However, these models are resource-intensive. To alleviate this issue, we explore distilling knowledge from instruction-tuned LLMs into much smaller ones. To this end, we carefully develop a large set of 2.58M instru...
http://arxiv.org/pdf/2304.14402v2
2023-05-24T07:15:26
2023-04-27T17:58:49
{ "title": [ -0.039611816, -0.019592285, -0.060028076, -0.046203613, -0.02645874, -0.0051879883, -0.088012695, -0.019485474, -0.002691269, 0.005493164, -0.025283813, -0.01436615, -0.0022850037, -0.051696777, -0.000093877316, -0.013366699, 0.017852783...
http://arxiv.org/abs/2304.14399v2
We're Afraid Language Models Aren't Modeling Ambiguity
Alisa Liu, Zhaofeng Wu, Julian Michael, Alane Suhr, Peter West, Alexander Koller, Swabha Swayamdipta, Noah A. Smith, Yejin Choi
cs.CL
Ambiguity is an intrinsic feature of natural language. Managing ambiguity is a key part of human language understanding, allowing us to anticipate misunderstanding as communicators and revise our interpretations as listeners. As language models (LMs) are increasingly employed as dialogue interfaces and writing aids, ha...
http://arxiv.org/pdf/2304.14399v2
2023-10-20T05:46:14
2023-04-27T17:57:58
{ "title": [ -0.025497437, -0.009353638, -0.0025100708, 0.033355713, -0.06964111, 0.004436493, -0.008361816, -0.030822754, -0.008224487, 0.05340576, -0.009063721, -0.044036865, -0.0012693405, -0.022415161, -0.045196533, -0.0024204254, -0.0033512115, ...
http://arxiv.org/abs/2304.14395v1
string2string: A Modern Python Library for String-to-String Algorithms
Mirac Suzgun, Stuart M. Shieber, Dan Jurafsky
cs.CL, cs.DL
We introduce string2string, an open-source library that offers a comprehensive suite of efficient algorithms for a broad range of string-to-string problems. It includes traditional algorithmic solutions as well as recent advanced neural approaches to tackle various problems in string alignment, distance measurement, le...
http://arxiv.org/pdf/2304.14395v1
2023-04-27T17:57:19
2023-04-27T17:57:19
{ "title": [ -0.03756714, -0.01965332, -0.051696777, -0.054351807, 0.013633728, -0.0026798248, -0.026794434, 0.009727478, 0.0118255615, 0.04260254, 0.0463562, -0.009651184, 0.023147583, -0.05517578, -0.04006958, 0.03491211, 0.0026741028, 0.031768...
http://arxiv.org/abs/2304.14391v3
Energy-based Models are Zero-Shot Planners for Compositional Scene Rearrangement
Nikolaos Gkanatsios, Ayush Jain, Zhou Xian, Yunchu Zhang, Christopher Atkeson, Katerina Fragkiadaki
cs.RO, cs.AI, cs.CL, cs.CV, cs.LG
Language is compositional; an instruction can express multiple relation constraints to hold among objects in a scene that a robot is tasked to rearrange. Our focus in this work is an instructable scene-rearranging framework that generalizes to longer instructions and to spatial concept compositions never seen at traini...
http://arxiv.org/pdf/2304.14391v3
2023-06-12T21:13:45
2023-04-27T17:55:13
{ "title": [ 0.034942627, -0.019439697, 0.0065345764, -0.005142212, -0.045288086, -0.015991211, -0.025283813, -0.026473999, -0.010055542, 0.03463745, -0.00065517426, 0.025512695, 0.0074157715, -0.042907715, 0.020339966, 0.0066452026, 0.01977539, ...
http://arxiv.org/abs/2304.14364v1
CONSCENDI: A Contrastive and Scenario-Guided Distillation Approach to Guardrail Models for Virtual Assistants
Albert Yu Sun, Varun Nair, Elliot Schumacher, Anitha Kannan
cs.CL, cs.AI, cs.LG
A wave of new task-based virtual assistants has been fueled by increasingly powerful large language models, such as GPT-4. These conversational agents can be customized to serve customer-specific use cases, but ensuring that agent-generated text conforms to designer-specified rules included in prompt instructions alone...
http://arxiv.org/pdf/2304.14364v1
2023-04-27T17:39:11
2023-04-27T17:39:11
{ "title": [ 0.007080078, -0.02746582, -0.04876709, -0.07159424, 0.007095337, -0.011528015, -0.033935547, 0.0181427, 0.008666992, 0.030654907, -0.024887085, -0.01663208, 0.024353027, -0.031555176, -0.035064697, -0.019119263, -0.020233154, 0.02134...
http://arxiv.org/abs/2304.14354v1
Industrial Engineering with Large Language Models: A case study of ChatGPT's performance on Oil & Gas problems
Oluwatosin Ogundare, Srinath Madasu, Nathanial Wiggins
cs.CL
Large Language Models (LLMs) have shown great potential in solving complex problems in various fields, including oil and gas engineering and other industrial engineering disciplines like factory automation, PLC programming etc. However, automatic identification of strong and weak solutions to fundamental physics equati...
http://arxiv.org/pdf/2304.14354v1
2023-04-27T17:33:49
2023-04-27T17:33:49
{ "title": [ -0.026290894, -0.023880005, -0.022018433, -0.004760742, -0.019699097, 0.0050239563, -0.032409668, -0.006965637, -0.021972656, 0.007221222, -0.023117065, 0.011383057, -0.0151901245, -0.022842407, 0.010017395, -0.021728516, -0.037506104, ...
http://arxiv.org/abs/2304.14333v1
Idioms, Probing and Dangerous Things: Towards Structural Probing for Idiomaticity in Vector Space
Filip Klubička, Vasudevan Nedumpozhimana, John D. Kelleher
cs.CL, cs.AI, cs.LG, 68T50
The goal of this paper is to learn more about how idiomatic information is structurally encoded in embeddings, using a structural probing method. We repurpose an existing English verbal multi-word expression (MWE) dataset to suit the probing framework and perform a comparative probing study of static (GloVe) and contex...
http://arxiv.org/pdf/2304.14333v1
2023-04-27T17:06:20
2023-04-27T17:06:20
{ "title": [ -0.0013093948, -0.0084991455, 0.009475708, -0.029327393, -0.0066566467, -0.008659363, 0.00025510788, -0.03793335, 0.0024204254, 0.0071525574, -0.031585693, 0.010810852, -0.014984131, -0.018051147, -0.026275635, -0.0046157837, -0.02523803...
http://arxiv.org/abs/2304.14318v1
q2d: Turning Questions into Dialogs to Teach Models How to Search
Yonatan Bitton, Shlomi Cohen-Ganor, Ido Hakimi, Yoad Lewenberg, Roee Aharoni, Enav Weinreb
cs.CL
One of the exciting capabilities of recent language models for dialog is their ability to independently search for relevant information to ground a given dialog response. However, obtaining training data to teach models how to issue search queries is time and resource consuming. In this work, we propose q2d: an automat...
http://arxiv.org/pdf/2304.14318v1
2023-04-27T16:39:15
2023-04-27T16:39:15
{ "title": [ 0.0022888184, -0.016433716, -0.01828003, -0.011558533, -0.011489868, -0.016983032, 0.003332138, -0.043884277, 0.021530151, 0.029312134, -0.03036499, 0.0066184998, -0.0076560974, -0.010643005, -0.053741455, -0.009246826, -0.0055885315, ...
http://arxiv.org/abs/2304.14317v1
Large Language Models Are State-of-the-Art Evaluators of Code Generation
Terry Yue Zhuo
cs.AI, cs.CL, cs.SE
Recent advancements in the field of natural language generation have facilitated the use of large language models to assess the quality of generated text. Although these models have shown promising results in tasks such as machine translation and summarization, their applicability in code generation tasks remains limit...
http://arxiv.org/pdf/2304.14317v1
2023-04-27T16:38:17
2023-04-27T16:38:17
{ "title": [ 0.006351471, -0.041290283, -0.002910614, -0.07861328, -0.06829834, 0.0005540848, -0.07232666, 0.027297974, 0.023010254, 0.021087646, -0.010482788, -0.010696411, 0.00504303, -0.046081543, 0.0068359375, -0.009773254, -0.0054855347, -0....
http://arxiv.org/abs/2304.14293v2
Controlled Text Generation with Natural Language Instructions
Wangchunshu Zhou, Yuchen Eleanor Jiang, Ethan Wilcox, Ryan Cotterell, Mrinmaya Sachan
cs.CL, cs.AI, cs.LG
Large language models generate fluent texts and can follow natural language instructions to solve a wide range of tasks without task-specific training. Nevertheless, it is notoriously difficult to control their generation to satisfy the various constraints required by different applications. In this work, we present In...
http://arxiv.org/pdf/2304.14293v2
2023-06-08T06:33:23
2023-04-27T15:56:34
{ "title": [ -0.03982544, -0.020324707, -0.033813477, -0.052093506, -0.021743774, 0.017837524, -0.052947998, -0.004096985, -0.0045166016, 0.03543091, -0.017410278, 0.017669678, 0.0036392212, -0.011001587, -0.030059814, 0.008644104, -0.011062622, ...
http://arxiv.org/abs/2304.14286v1
Semantic Frame Induction with Deep Metric Learning
Kosuke Yamada, Ryohei Sasano, Koichi Takeda
cs.CL
Recent studies have demonstrated the usefulness of contextualized word embeddings in unsupervised semantic frame induction. However, they have also revealed that generic contextualized embeddings are not always consistent with human intuitions about semantic frames, which causes unsatisfactory performance for frame ind...
http://arxiv.org/pdf/2304.14286v1
2023-04-27T15:46:09
2023-04-27T15:46:09
{ "title": [ -0.015731812, -0.022659302, -0.040649414, -0.059295654, -0.0025539398, -0.002702713, -0.034729004, 0.017120361, -0.00001835823, 0.04385376, -0.013130188, 0.014656067, 0.019744873, -0.014633179, -0.023010254, 0.018096924, -0.0041236877, ...
http://arxiv.org/abs/2304.14241v1
Entity-Level Sentiment Analysis (ELSA): An exploratory task survey
Egil Rønningstad, Erik Velldal, Lilja Øvrelid
cs.CL
This paper explores the task of identifying the overall sentiment expressed towards volitional entities (persons and organizations) in a document -- what we refer to as Entity-Level Sentiment Analysis (ELSA). While identifying sentiment conveyed towards an entity is well researched for shorter texts like tweets, we fin...
http://arxiv.org/pdf/2304.14241v1
2023-04-27T15:01:20
2023-04-27T15:01:20
{ "title": [ -0.008865356, -0.01889038, -0.031951904, -0.03161621, 0.015914917, -0.013748169, -0.007118225, -0.010467529, 0.02331543, 0.038208008, -0.02684021, 0.004875183, -0.0018615723, -0.012252808, -0.07171631, 0.008934021, -0.014389038, 0.04...
http://arxiv.org/abs/2304.14238v2
The Intended Uses of Automated Fact-Checking Artefacts: Why, How and Who
Michael Schlichtkrull, Nedjma Ousidhoum, Andreas Vlachos
cs.CL
Automated fact-checking is often presented as an epistemic tool that fact-checkers, social media consumers, and other stakeholders can use to fight misinformation. Nevertheless, few papers thoroughly discuss how. We document this by analysing 100 highly-cited papers, and annotating epistemic elements related to intende...
http://arxiv.org/pdf/2304.14238v2
2023-11-08T12:10:49
2023-04-27T14:55:23
{ "title": [ 0.019378662, -0.0011072159, -0.039215088, -0.039642334, -0.007888794, -0.028717041, -0.002210617, -0.026412964, -0.06451416, 0.04421997, -0.009521484, 0.017715454, 0.015174866, -0.02986145, -0.017593384, -0.01322937, -0.01927185, 0.0...
http://arxiv.org/abs/2304.14233v2
Large Language Models are Strong Zero-Shot Retriever
Tao Shen, Guodong Long, Xiubo Geng, Chongyang Tao, Tianyi Zhou, Daxin Jiang
cs.CL, cs.IR
In this work, we propose a simple method that applies a large language model (LLM) to large-scale retrieval in zero-shot scenarios. Our method, the Language language model as Retriever (LameR), is built upon no other neural models but an LLM, while breaking brute-force combinations of retrievers with LLMs and lifting t...
http://arxiv.org/pdf/2304.14233v2
2023-08-02T02:06:28
2023-04-27T14:45:55
{ "title": [ 0.04360962, -0.010101318, -0.023986816, -0.091552734, -0.033721924, -0.02168274, -0.062438965, 0.025894165, -0.0017127991, 0.045318604, 0.0016994476, -0.016281128, 0.028701782, -0.050445557, 0.0040740967, 0.021865845, 0.01209259, 0.0...
http://arxiv.org/abs/2304.14221v1
A Modular Approach for Multilingual Timex Detection and Normalization using Deep Learning and Grammar-based methods
Nayla Escribano, German Rigau, Rodrigo Agerri
cs.CL
Detecting and normalizing temporal expressions is an essential step for many NLP tasks. While a variety of methods have been proposed for detection, best normalization approaches rely on hand-crafted rules. Furthermore, most of them have been designed only for English. In this paper we present a modular multilingual te...
http://arxiv.org/pdf/2304.14221v1
2023-04-27T14:32:23
2023-04-27T14:32:23
{ "title": [ -0.017044067, 0.008140564, -0.06524658, -0.0496521, 0.016998291, -0.02432251, -0.03137207, -0.0076789856, -0.009849548, 0.03237915, -0.0061302185, 0.014427185, -0.007549286, -0.03677368, -0.02696228, -0.007686615, -0.0042877197, 0.03...
http://arxiv.org/abs/2304.14189v1
UIO at SemEval-2023 Task 12: Multilingual fine-tuning for sentiment classification in low-resource languages
Egil Rønningstad
cs.CL
Our contribution to the 2023 AfriSenti-SemEval shared task 12: Sentiment Analysis for African Languages, provides insight into how a multilingual large language model can be a resource for sentiment analysis in languages not seen during pretraining. The shared task provides datasets of a variety of African languages fr...
http://arxiv.org/pdf/2304.14189v1
2023-04-27T13:51:18
2023-04-27T13:51:18
{ "title": [ -0.010101318, -0.026107788, -0.0073127747, -0.052978516, -0.016906738, 0.0052719116, -0.05419922, -0.0032196045, 0.012886047, 0.0524292, -0.025527954, 0.0030269623, -0.0036506653, 0.0015211105, -0.0736084, -0.0015449524, -0.022064209, ...
http://arxiv.org/abs/2304.14179v1
NAP at SemEval-2023 Task 3: Is Less Really More? (Back-)Translation as Data Augmentation Strategies for Detecting Persuasion Techniques
Neele Falk, Annerose Eichel, Prisca Piccirilli
cs.CL
Persuasion techniques detection in news in a multi-lingual setup is non-trivial and comes with challenges, including little training data. Our system successfully leverages (back-)translation as data augmentation strategies with multi-lingual transformer models for the task of detecting persuasion techniques. The autom...
http://arxiv.org/pdf/2304.14179v1
2023-04-27T13:33:08
2023-04-27T13:33:08
{ "title": [ 0.034423828, 0.0052223206, 0.0066871643, 0.007858276, 0.007522583, -0.02961731, 0.020233154, -0.04550171, -0.06573486, 0.023757935, -0.006095886, 0.052093506, -0.019546509, -0.020980835, 0.017211914, -0.0154418945, -0.022354126, 0.05...
http://arxiv.org/abs/2304.14178v1
mPLUG-Owl: Modularization Empowers Large Language Models with Multimodality
Qinghao Ye, Haiyang Xu, Guohai Xu, Jiabo Ye, Ming Yan, Yiyang Zhou, Junyang Wang, Anwen Hu, Pengcheng Shi, Yaya Shi, Chenliang Li, Yuanhong Xu, Hehong Chen, Junfeng Tian, Qian Qi, Ji Zhang, Fei Huang
cs.CL, cs.CV, cs.LG
Large language models (LLMs) have demonstrated impressive zero-shot abilities on a variety of open-ended tasks, while recent research has also explored the use of LLMs for multi-modal generation. In this study, we introduce mPLUG-Owl, a novel training paradigm that equips LLMs with multi-modal abilities through modular...
http://arxiv.org/pdf/2304.14178v1
2023-04-27T13:27:01
2023-04-27T13:27:01
{ "title": [ -0.025756836, -0.017684937, -0.07647705, -0.0970459, -0.0028820038, 0.0014829636, -0.07849121, 0.029922485, -0.018798828, 0.031463623, -0.001159668, 0.019119263, 0.0053710938, -0.017669678, -0.023849487, 0.012992859, -0.013809204, 0....
http://arxiv.org/abs/2304.14177v2
ChatGPT vs State-of-the-Art Models: A Benchmarking Study in Keyphrase Generation Task
Roberto Martínez-Cruz, Alvaro J. López-López, José Portela
cs.CL, cs.AI
Transformer-based language models, including ChatGPT, have demonstrated exceptional performance in various natural language generation tasks. However, there has been limited research evaluating ChatGPT's keyphrase generation ability, which involves identifying informative phrases that accurately reflect a document's co...
http://arxiv.org/pdf/2304.14177v2
2023-06-29T13:40:42
2023-04-27T13:25:43
{ "title": [ -0.03427124, -0.022979736, -0.004421234, 0.00013780594, -0.029006958, 0.017349243, -0.03543091, -0.010307312, -0.011367798, 0.05117798, 0.00086069107, 0.030319214, -0.005508423, -0.005973816, -0.048309326, 0.014175415, -0.038848877, ...
http://arxiv.org/abs/2304.14108v5
DataComp: In search of the next generation of multimodal datasets
Samir Yitzhak Gadre, Gabriel Ilharco, Alex Fang, Jonathan Hayase, Georgios Smyrnis, Thao Nguyen, Ryan Marten, Mitchell Wortsman, Dhruba Ghosh, Jieyu Zhang, Eyal Orgad, Rahim Entezari, Giannis Daras, Sarah Pratt, Vivek Ramanujan, Yonatan Bitton, Kalyani Marathe, Stephen Mussmann, Richard Vencu, Mehdi Cherti, Ranjay Kris...
cs.CV, cs.CL, cs.LG
Multimodal datasets are a critical component in recent breakthroughs such as Stable Diffusion and GPT-4, yet their design does not receive the same research attention as model architectures or training algorithms. To address this shortcoming in the ML ecosystem, we introduce DataComp, a testbed for dataset experiments ...
http://arxiv.org/pdf/2304.14108v5
2023-10-20T17:01:44
2023-04-27T11:37:18
{ "title": [ -0.040161133, -0.015365601, -0.0690918, -0.07824707, 0.0062675476, 0.0070648193, -0.06378174, 0.0014162064, 0.024017334, 0.020751953, -0.03326416, -0.005722046, 0.00038290024, -0.017303467, -0.053619385, -0.0034008026, -0.011054993, ...
http://arxiv.org/abs/2304.14106v1
ChatLog: Recording and Analyzing ChatGPT Across Time
Shangqing Tu, Chunyang Li, Jifan Yu, Xiaozhi Wang, Lei Hou, Juanzi Li
cs.CL, cs.AI
While there are abundant researches about evaluating ChatGPT on natural language understanding and generation tasks, few studies have investigated how ChatGPT's behavior changes over time. In this paper, we collect a coarse-to-fine temporal dataset called ChatLog, consisting of two parts that update monthly and daily: ...
http://arxiv.org/pdf/2304.14106v1
2023-04-27T11:33:48
2023-04-27T11:33:48
{ "title": [ 0.017578125, -0.03225708, -0.034942627, -0.033203125, -0.028137207, -0.015434265, -0.022735596, 0.00016212463, -0.0054969788, 0.02406311, -0.027893066, 0.029281616, -0.0041618347, -0.024383545, -0.014877319, 0.002319336, -0.0061531067, ...
http://arxiv.org/abs/2304.14104v4
Learning Human-Human Interactions in Images from Weak Textual Supervision
Morris Alper, Hadar Averbuch-Elor
cs.CV, cs.CL, cs.LG
Interactions between humans are diverse and context-dependent, but previous works have treated them as categorical, disregarding the heavy tail of possible interactions. We propose a new paradigm of learning human-human interactions as free text from a single still image, allowing for flexibility in modeling the unlimi...
http://arxiv.org/pdf/2304.14104v4
2023-09-18T17:50:31
2023-04-27T11:32:48
{ "title": [ 0.015548706, 0.0061569214, -0.013244629, -0.058288574, 0.0074653625, -0.0066986084, -0.024383545, -0.016433716, 0.0077667236, 0.01600647, -0.049987793, 0.019348145, -0.011398315, 0.0016813278, -0.035095215, -0.027923584, 0.0006790161, ...
http://arxiv.org/abs/2304.14072v1
Origin Tracing and Detecting of LLMs
Linyang Li, Pengyu Wang, Ke Ren, Tianxiang Sun, Xipeng Qiu
cs.CL
The extraordinary performance of large language models (LLMs) heightens the importance of detecting whether the context is generated by an AI system. More importantly, while more and more companies and institutions release their LLMs, the origin can be hard to trace. Since LLMs are heading towards the time of AGI, simi...
http://arxiv.org/pdf/2304.14072v1
2023-04-27T10:05:57
2023-04-27T10:05:57
{ "title": [ -0.0013084412, 0.005794525, -0.027832031, -0.045196533, -0.039764404, 0.0061302185, -0.07989502, -0.030975342, -0.052886963, 0.01991272, 0.01776123, 0.002506256, 0.011276245, -0.071899414, 0.05065918, 0.002040863, 0.015274048, 0.0115...
http://arxiv.org/abs/2304.13994v3
SweCTRL-Mini: a data-transparent Transformer-based large language model for controllable text generation in Swedish
Dmytro Kalpakchi, Johan Boye
cs.CL
We present SweCTRL-Mini, a large Swedish language model that can be used for inference and fine-tuning on a single consumer-grade GPU. The model is based on the CTRL architecture by Keskar, McCann, Varshney, Xiong, and Socher (2019), which means that users of the SweCTRL-Mini model can control the genre of the generate...
http://arxiv.org/pdf/2304.13994v3
2023-06-22T13:02:23
2023-04-27T07:32:37
{ "title": [ -0.011108398, -0.010955811, -0.02961731, -0.05239868, 0.0021018982, 0.009536743, -0.0546875, 0.0043258667, -0.009933472, 0.044067383, -0.0045204163, 0.019226074, 0.05947876, -0.03111267, -0.015487671, 0.010856628, -0.021209717, 0.049...
http://arxiv.org/abs/2304.13989v1
Cross-Domain Evaluation of POS Taggers: From Wall Street Journal to Fandom Wiki
Kia Kirstein Hansen, Rob van der Goot
cs.CL
The Wall Street Journal section of the Penn Treebank has been the de-facto standard for evaluating POS taggers for a long time, and accuracies over 97\% have been reported. However, less is known about out-of-domain tagger performance, especially with fine-grained label sets. Using data from Elder Scrolls Fandom, a wik...
http://arxiv.org/pdf/2304.13989v1
2023-04-27T07:24:29
2023-04-27T07:24:29
{ "title": [ 0.030960083, 0.028427124, -0.008644104, -0.025817871, 0.018035889, -0.03213501, 0.007129669, -0.042663574, -0.019104004, 0.01864624, -0.029846191, 0.013511658, 0.010765076, -0.019607544, -0.012840271, -0.0206604, -0.017593384, 0.0331...
http://arxiv.org/abs/2305.04836v1
Understanding the Impact of Culture in Assessing Helpfulness of Online Reviews
Khaled Alanezi, Nuha Albadi, Omar Hammad, Maram Kurdi, Shivakant Mishra
cs.IR, cs.CL, cs.CY
Online reviews have become essential for users to make informed decisions in everyday tasks ranging from planning summer vacations to purchasing groceries and making financial investments. A key problem in using online reviews is the overabundance of online that overwhelms the users. As a result, recommendation systems...
http://arxiv.org/pdf/2305.04836v1
2023-04-27T07:20:55
2023-04-27T07:20:55
{ "title": [ 0.025894165, -0.01260376, -0.00057935715, -0.035461426, -0.012046814, -0.032104492, 0.010444641, -0.016937256, -0.025009155, 0.028533936, -0.050720215, 0.008506775, 0.011047363, -0.009590149, -0.020858765, -0.0032539368, 0.02696228, ...
http://arxiv.org/abs/2305.07468v1
BactInt: A domain driven transfer learning approach and a corpus for extracting inter-bacterial interactions from biomedical text
Krishanu Das Baksi, Vatsala Pokhrel, Kuntal Kumar Bhusan, Sharmila Mande
cs.IR, cs.AI, cs.CL, cs.LG, q-bio.QM
The community of different types of microbes present in a biological niche plays a very important role in functioning of the system. The crosstalk or interactions among the different microbes contributes to the building blocks of such microbial community structures. Evidence reported in biomedical text serves as a reli...
http://arxiv.org/pdf/2305.07468v1
2023-04-27T06:14:25
2023-04-27T06:14:25
{ "title": [ -0.01777649, -0.004055023, -0.05316162, -0.074645996, 0.009010315, 0.004371643, -0.05496216, -0.0034503937, 0.024536133, 0.016433716, -0.028076172, 0.008880615, -0.0044555664, -0.007194519, -0.0524292, -0.01979065, -0.015777588, 0.03...
http://arxiv.org/abs/2304.13958v1
Learning and Reasoning Multifaceted and Longitudinal Data for Poverty Estimates and Livelihood Capabilities of Lagged Regions in Rural India
Atharva Kulkarni, Raya Das, Ravi S. Srivastava, Tanmoy Chakraborty
cs.CL, cs.CY, cs.SI
Poverty is a multifaceted phenomenon linked to the lack of capabilities of households to earn a sustainable livelihood, increasingly being assessed using multidimensional indicators. Its spatial pattern depends on social, economic, political, and regional variables. Artificial intelligence has shown immense scope in an...
http://arxiv.org/pdf/2304.13958v1
2023-04-27T05:33:08
2023-04-27T05:33:08
{ "title": [ 0.030059814, 0.00056791306, -0.04724121, -0.048980713, 0.007129669, -0.031585693, -0.02330017, 0.021316528, -0.011192322, -0.022369385, -0.04647827, 0.00075626373, -0.050354004, -0.030197144, 0.02017212, -0.06512451, -0.013908386, 0....
http://arxiv.org/abs/2304.13923v2
Retrieval-based Knowledge Augmented Vision Language Pre-training
Jiahua Rao, Zifei Shan, Longpo Liu, Yao Zhou, Yuedong Yang
cs.CV, cs.CL, cs.MM
With the recent progress in large-scale vision and language representation learning, Vision Language Pre-training (VLP) models have achieved promising improvements on various multi-modal downstream tasks. Albeit powerful, these models have not fully leveraged world knowledge to their advantage. A key challenge of knowl...
http://arxiv.org/pdf/2304.13923v2
2023-08-06T08:06:43
2023-04-27T02:23:47
{ "title": [ 0.0046310425, -0.02420044, -0.04675293, -0.039276123, -0.024246216, -0.021087646, -0.041778564, 0.013298035, -0.020385742, 0.04949951, -0.01828003, -0.013633728, 0.04852295, -0.02848816, 0.0030269623, -0.0178833, 0.049316406, -0.0282...
http://arxiv.org/abs/2304.13902v2
Controllable Data Augmentation for Context-Dependent Text-to-SQL
Dingzirui Wang, Longxu Dou, Wanxiang Che
cs.CL
The limited scale of annotated data constraints existing context-dependent text-to-SQL models because of the complexity of labeling. The data augmentation method is a commonly used method to solve this problem. However, the data generated by current augmentation methods often lack diversity. In this paper, we introduce...
http://arxiv.org/pdf/2304.13902v2
2023-04-28T02:45:31
2023-04-27T01:00:10
{ "title": [ 0.013908386, -0.0029582977, -0.06555176, -0.09893799, 0.038726807, -0.0011081696, -0.014709473, 0.039215088, 0.018371582, 0.054992676, -0.048095703, 0.022750854, 0.023742676, 0.008491516, -0.04748535, 0.0013046265, -0.0033340454, 0.0...
http://arxiv.org/abs/2304.13883v1
Neural Keyphrase Generation: Analysis and Evaluation
Tuhin Kundu, Jishnu Ray Chowdhury, Cornelia Caragea
cs.CL, cs.IR
Keyphrase generation aims at generating topical phrases from a given text either by copying from the original text (present keyphrases) or by producing new keyphrases (absent keyphrases) that capture the semantic meaning of the text. Encoder-decoder models are most widely used for this task because of their capabilitie...
http://arxiv.org/pdf/2304.13883v1
2023-04-27T00:10:21
2023-04-27T00:10:21
{ "title": [ -0.043945312, -0.0055274963, -0.017089844, -0.01008606, -0.013847351, 0.0104522705, -0.03656006, -0.011009216, -0.004966736, 0.06414795, 0.010002136, 0.008514404, 0.013000488, -0.009437561, -0.061767578, 0.020370483, -0.025497437, 0....
http://arxiv.org/abs/2304.13875v1
MasonNLP+ at SemEval-2023 Task 8: Extracting Medical Questions, Experiences and Claims from Social Media using Knowledge-Augmented Pre-trained Language Models
Giridhar Kaushik Ramachandran, Haritha Gangavarapu, Kevin Lybarger, Ozlem Uzuner
cs.CL, cs.LG
In online forums like Reddit, users share their experiences with medical conditions and treatments, including making claims, asking questions, and discussing the effects of treatments on their health. Building systems to understand this information can effectively monitor the spread of misinformation and verify user cl...
http://arxiv.org/pdf/2304.13875v1
2023-04-26T23:45:11
2023-04-26T23:45:11
{ "title": [ -0.0015764236, -0.0070724487, -0.02796936, -0.018936157, -0.027893066, -0.0022945404, -0.035583496, -0.014335632, -0.023117065, 0.026229858, 0.017242432, 0.048065186, 0.0029201508, -0.03427124, 0.034240723, 0.018829346, 0.0007824898, ...
http://arxiv.org/abs/2304.13867v2
Transferring Procedural Knowledge across Commonsense Tasks
Yifan Jiang, Filip Ilievski, Kaixin Ma
cs.CL
Stories about everyday situations are an essential part of human communication, motivating the need to develop AI agents that can reliably understand these stories. Despite the long list of supervised methods for story completion and procedural understanding, current AI has no mechanisms to automatically track and expl...
http://arxiv.org/pdf/2304.13867v2
2023-07-27T21:56:30
2023-04-26T23:24:50
{ "title": [ -0.014854431, -0.037963867, -0.029006958, -0.0463562, -0.030395508, -0.010116577, -0.022033691, -0.0053977966, 0.020095825, 0.020889282, -0.008903503, -0.010276794, -0.021209717, -0.031433105, -0.037841797, 0.004169464, 0.005088806, ...
http://arxiv.org/abs/2304.13861v1
Is a prompt and a few samples all you need? Using GPT-4 for data augmentation in low-resource classification tasks
Anders Giovanni Møller, Jacob Aarup Dalsgaard, Arianna Pera, Luca Maria Aiello
cs.CL, cs.CY, physics.soc-ph
Obtaining and annotating data can be expensive and time-consuming, especially in complex, low-resource domains. We use GPT-4 and ChatGPT to augment small labeled datasets with synthetic data via simple prompts, in three different classification tasks with varying complexity. For each task, we randomly select a base sam...
http://arxiv.org/pdf/2304.13861v1
2023-04-26T23:09:02
2023-04-26T23:09:02
{ "title": [ 0.0099105835, 0.01914978, -0.035980225, -0.033996582, 0.013938904, -0.033569336, -0.011817932, -0.02330017, 0.004550934, 0.002609253, -0.051757812, 0.0033912659, -0.02532959, -0.025024414, -0.028244019, -0.027420044, 0.0029182434, 0....
http://arxiv.org/abs/2304.13846v1
Extracting Structured Seed-Mediated Gold Nanorod Growth Procedures from Literature with GPT-3
Nicholas Walker, John Dagdelen, Kevin Cruse, Sanghoon Lee, Samuel Gleason, Alexander Dunn, Gerbrand Ceder, A. Paul Alivisatos, Kristin A. Persson, Anubhav Jain
physics.app-ph, cs.CL, cs.IR
Although gold nanorods have been the subject of much research, the pathways for controlling their shape and thereby their optical properties remain largely heuristically understood. Although it is apparent that the simultaneous presence of and interaction between various reagents during synthesis control these properti...
http://arxiv.org/pdf/2304.13846v1
2023-04-26T22:21:33
2023-04-26T22:21:33
{ "title": [ -0.015319824, 0.026794434, -0.037841797, -0.029312134, 0.015716553, -0.013305664, -0.041656494, -0.046142578, -0.039031982, 0.018692017, -0.016601562, 0.051696777, -0.021987915, -0.010665894, -0.0008993149, 0.023544312, -0.0005068779, ...
http://arxiv.org/abs/2304.13835v3
Multi-Party Chat: Conversational Agents in Group Settings with Humans and Models
Jimmy Wei, Kurt Shuster, Arthur Szlam, Jason Weston, Jack Urbanek, Mojtaba Komeili
cs.CL, cs.LG
Current dialogue research primarily studies pairwise (two-party) conversations, and does not address the everyday setting where more than two speakers converse together. In this work, we both collect and evaluate multi-party conversations to study this more general case. We use the LIGHT environment to construct ground...
http://arxiv.org/pdf/2304.13835v3
2023-06-08T21:45:46
2023-04-26T21:41:17
{ "title": [ -0.008743286, -0.0053482056, -0.041748047, -0.05178833, 0.017700195, -0.0121536255, -0.032806396, -0.008842468, 0.005558014, 0.04232788, -0.027191162, 0.011177063, 0.0077934265, -0.008262634, -0.044555664, 0.0037822723, -0.022857666, ...
http://arxiv.org/abs/2304.13803v1
Translate to Disambiguate: Zero-shot Multilingual Word Sense Disambiguation with Pretrained Language Models
Haoqiang Kang, Terra Blevins, Luke Zettlemoyer
cs.CL
Pretrained Language Models (PLMs) learn rich cross-lingual knowledge and can be finetuned to perform well on diverse tasks such as translation and multilingual word sense disambiguation (WSD). However, they often struggle at disambiguating word sense in a zero-shot setting. To better understand this contrast, we presen...
http://arxiv.org/pdf/2304.13803v1
2023-04-26T19:55:52
2023-04-26T19:55:52
{ "title": [ -0.010009766, -0.017593384, -0.020431519, -0.027160645, -0.027709961, -0.014831543, -0.041046143, -0.007144928, 0.011436462, 0.053466797, -0.00806427, -0.0037326813, 0.041137695, -0.023330688, -0.041259766, 0.007511139, 0.002248764, ...
http://arxiv.org/abs/2304.13783v1
Fine Tuning with Abnormal Examples
Will Rieger
cs.CL
Given the prevalence of crowd sourced labor in creating Natural Language processing datasets, these aforementioned sets have become increasingly large. For instance, the SQUAD dataset currently sits at over 80,000 records. However, because the English language is rather repetitive in structure, the distribution of word...
http://arxiv.org/pdf/2304.13783v1
2023-04-26T18:59:48
2023-04-26T18:59:48
{ "title": [ 0.07208252, -0.013908386, 0.03894043, -0.04736328, -0.029174805, 0.0033512115, 0.004737854, -0.011505127, -0.032165527, 0.012359619, -0.030380249, 0.010910034, -0.024871826, -0.026687622, 0.035614014, -0.027313232, -0.005405426, 0.05...
http://arxiv.org/abs/2304.13714v3
Evaluation of GPT-3.5 and GPT-4 for supporting real-world information needs in healthcare delivery
Debadutta Dash, Rahul Thapa, Juan M. Banda, Akshay Swaminathan, Morgan Cheatham, Mehr Kashyap, Nikesh Kotecha, Jonathan H. Chen, Saurabh Gombar, Lance Downing, Rachel Pedreira, Ethan Goh, Angel Arnaout, Garret Kenn Morris, Honor Magon, Matthew P Lungren, Eric Horvitz, Nigam H. Shah
cs.AI, cs.CL, cs.IR
Despite growing interest in using large language models (LLMs) in healthcare, current explorations do not assess the real-world utility and safety of LLMs in clinical settings. Our objective was to determine whether two LLMs can serve information needs submitted by physicians as questions to an informatics consultation...
http://arxiv.org/pdf/2304.13714v3
2023-05-01T00:41:37
2023-04-26T17:54:28
{ "title": [ 0.0060577393, 0.011787415, -0.026229858, -0.05581665, 0.011558533, -0.01474762, -0.042022705, -0.017730713, -0.023757935, 0.014022827, -0.022888184, 0.004169464, -0.014099121, -0.03640747, 0.0014209747, -0.010284424, 0.006526947, -0....
http://arxiv.org/abs/2304.13712v2
Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyond
Jingfeng Yang, Hongye Jin, Ruixiang Tang, Xiaotian Han, Qizhang Feng, Haoming Jiang, Bing Yin, Xia Hu
cs.CL, cs.AI, cs.LG
This paper presents a comprehensive and practical guide for practitioners and end-users working with Large Language Models (LLMs) in their downstream natural language processing (NLP) tasks. We provide discussions and insights into the usage of LLMs from the perspectives of models, data, and downstream tasks. Firstly, ...
http://arxiv.org/pdf/2304.13712v2
2023-04-27T17:56:11
2023-04-26T17:52:30
{ "title": [ -0.05001831, -0.036987305, -0.040618896, -0.026428223, -0.034301758, 0.01210022, -0.059936523, 0.0018415451, -0.0018081665, 0.0070152283, -0.048309326, 0.039398193, -0.023208618, -0.016448975, -0.008132935, 0.0019626617, -0.02960205, ...
http://arxiv.org/abs/2304.13689v1
HeySQuAD: A Spoken Question Answering Dataset
Yijing Wu, SaiKrishna Rallabandi, Ravisutha Srinivasamurthy, Parag Pravin Dakle, Alolika Gon, Preethi Raghavan
cs.CL, cs.AI
Human-spoken questions are critical to evaluating the performance of spoken question answering (SQA) systems that serve several real-world use cases including digital assistants. We present a new large-scale community-shared SQA dataset, HeySQuAD that consists of 76k human-spoken questions and 97k machine-generated que...
http://arxiv.org/pdf/2304.13689v1
2023-04-26T17:15:39
2023-04-26T17:15:39
{ "title": [ -0.012413025, -0.01020813, -0.045013428, -0.072387695, 0.012817383, 0.0015573502, -0.04714966, 0.024261475, 0.022369385, 0.032196045, -0.024993896, -0.0077705383, -0.0043525696, 0.006275177, -0.062683105, 0.0018358231, 0.0012102127, ...
http://arxiv.org/abs/2304.13664v1
Using Implicit Feedback to Improve Question Generation
Hugo Rodrigues, Eric Nyberg, Luisa Coheur
cs.CL, cs.AI
Question Generation (QG) is a task of Natural Language Processing (NLP) that aims at automatically generating questions from text. Many applications can benefit from automatically generated questions, but often it is necessary to curate those questions, either by selecting or editing them. This task is informative on i...
http://arxiv.org/pdf/2304.13664v1
2023-04-26T16:37:47
2023-04-26T16:37:47
{ "title": [ -0.0054473877, 0.0001604557, -0.01576233, -0.031921387, -0.0020103455, 0.004016876, -0.018981934, -0.031280518, 0.036132812, 0.004398346, -0.05026245, 0.0039367676, -0.03591919, -0.0015249252, -0.04547119, -0.0137786865, 0.00047969818, ...
http://arxiv.org/abs/2304.13649v1
A Symmetric Dual Encoding Dense Retrieval Framework for Knowledge-Intensive Visual Question Answering
Alireza Salemi, Juan Altmayer Pizzorno, Hamed Zamani
cs.CV, cs.CL, cs.IR
Knowledge-Intensive Visual Question Answering (KI-VQA) refers to answering a question about an image whose answer does not lie in the image. This paper presents a new pipeline for KI-VQA tasks, consisting of a retriever and a reader. First, we introduce DEDR, a symmetric dual encoding dense retrieval framework in which...
http://arxiv.org/pdf/2304.13649v1
2023-04-26T16:14:39
2023-04-26T16:14:39
{ "title": [ 0.006713867, -0.0051345825, -0.015449524, -0.0597229, -0.008468628, -0.00030755997, -0.024551392, 0.003604889, 0.04220581, 0.04827881, -0.028549194, 0.0018110275, 0.010192871, -0.0035476685, -0.07623291, 0.0077667236, 0.03704834, -0....
http://arxiv.org/abs/2304.13634v1
HausaNLP at SemEval-2023 Task 12: Leveraging African Low Resource TweetData for Sentiment Analysis
Saheed Abdullahi Salahudeen, Falalu Ibrahim Lawan, Ahmad Mustapha Wali, Amina Abubakar Imam, Aliyu Rabiu Shuaibu, Aliyu Yusuf, Nur Bala Rabiu, Musa Bello, Shamsuddeen Umaru Adamu, Saminu Mohammad Aliyu, Murja Sani Gadanya, Sanah Abdullahi Muaz, Mahmoud Said Ahmad, Abdulkadir Abdullahi, Abdulmalik Yusuf Jamoh
cs.CL
We present the findings of SemEval-2023 Task 12, a shared task on sentiment analysis for low-resource African languages using Twitter dataset. The task featured three subtasks; subtask A is monolingual sentiment classification with 12 tracks which are all monolingual languages, subtask B is multilingual sentiment class...
http://arxiv.org/pdf/2304.13634v1
2023-04-26T15:47:50
2023-04-26T15:47:50
{ "title": [ -0.0070343018, -0.023986816, -0.015670776, -0.05078125, -0.022003174, 0.015701294, -0.042938232, -0.0016298294, 0.00340271, 0.04321289, -0.04574585, 0.020614624, 0.0025138855, 0.0013017654, -0.03778076, 0.0003027916, 0.0016717911, 0....
http://arxiv.org/abs/2304.13620v3
ChartSumm: A Comprehensive Benchmark for Automatic Chart Summarization of Long and Short Summaries
Raian Rahman, Rizvi Hasan, Abdullah Al Farhad, Md Tahmid Rahman Laskar, Md. Hamjajul Ashmafee, Abu Raihan Mostofa Kamal
cs.CL, cs.LG
Automatic chart to text summarization is an effective tool for the visually impaired people along with providing precise insights of tabular data in natural language to the user. A large and well-structured dataset is always a key part for data driven models. In this paper, we propose ChartSumm: a large-scale benchmark...
http://arxiv.org/pdf/2304.13620v3
2023-06-11T04:07:27
2023-04-26T15:25:24
{ "title": [ -0.0107040405, -0.0022201538, -0.037750244, -0.080566406, -0.008544922, 0.0031013489, -0.059326172, -0.022201538, 0.021942139, 0.04650879, -0.008583069, 0.029266357, 0.0019702911, -0.009750366, -0.047088623, 0.029785156, 0.014732361, ...
http://arxiv.org/abs/2304.13597v1
Shades of meaning: Uncovering the geometry of ambiguous word representations through contextualised language models
Benedetta Cevoli, Chris Watkins, Yang Gao, Kathleen Rastle
cs.CL
Lexical ambiguity presents a profound and enduring challenge to the language sciences. Researchers for decades have grappled with the problem of how language users learn, represent and process words with more than one meaning. Our work offers new insight into psychological understanding of lexical ambiguity through a s...
http://arxiv.org/pdf/2304.13597v1
2023-04-26T14:47:38
2023-04-26T14:47:38
{ "title": [ -0.026626587, 0.0042686462, -0.038269043, -0.03781128, -0.0024547577, 0.0006971359, -0.049835205, -0.019760132, -0.00013792515, 0.043273926, -0.009681702, 0.0011224747, 0.016616821, -0.009529114, -0.053344727, 0.0087509155, -0.011451721,...
http://arxiv.org/abs/2304.13568v1
Toxic comments reduce the activity of volunteer editors on Wikipedia
Ivan Smirnov, Camelia Oprea, Markus Strohmaier
cs.CY, cs.CL
Wikipedia is one of the most successful collaborative projects in history. It is the largest encyclopedia ever created, with millions of users worldwide relying on it as the first source of information as well as for fact-checking and in-depth research. As Wikipedia relies solely on the efforts of its volunteer-editors...
http://arxiv.org/pdf/2304.13568v1
2023-04-26T13:59:25
2023-04-26T13:59:25
{ "title": [ 0.057373047, 0.014961243, -0.044708252, -0.036499023, 0.014007568, -0.032043457, 0.021972656, -0.006866455, -0.0395813, 0.055114746, -0.0009417534, -0.00289917, 0.034698486, -0.02670288, 0.008346558, -0.017837524, 0.005783081, 0.0124...
http://arxiv.org/abs/2304.13567v2
Technical Report on Token Position Bias in Transformers
Mehdi Ben Amor, Michael Granitzer, Jelena Mitrović
cs.CL, cs.AI
Language Models (LMs) have shown state-of-the-art performance in Natural Language Processing (NLP) tasks. Downstream tasks such as Named Entity Recognition (NER) or Part-of-Speech (POS) tagging are known to suffer from data imbalance issues, specifically in terms of the ratio of positive to negative examples, and class...
http://arxiv.org/pdf/2304.13567v2
2023-06-04T09:11:03
2023-04-26T13:57:25
{ "title": [ 0.043823242, 0.01436615, -0.017593384, -0.016525269, 0.007762909, -0.045135498, 0.022232056, -0.033477783, -0.026412964, 0.05633545, -0.05041504, -0.024383545, 0.05267334, -0.03475952, 0.0022563934, -0.009849548, -0.021102905, 0.0582...
http://arxiv.org/abs/2304.13559v2
Towards Multi-Modal DBMSs for Seamless Querying of Texts and Tables
Matthias Urban, Carsten Binnig
cs.DB, cs.CL
In this paper, we propose Multi-Modal Databases (MMDBs), which is a new class of database systems that can seamlessly query text and tables using SQL. To enable seamless querying of textual data using SQL in an MMDB, we propose to extend relational databases with so-called multi-modal operators (MMOps) which are based ...
http://arxiv.org/pdf/2304.13559v2
2023-04-28T14:55:48
2023-04-26T13:31:04
{ "title": [ -0.01828003, 0.006454468, -0.06549072, -0.07208252, 0.036590576, -0.01108551, -0.04248047, 0.004558563, 0.032409668, 0.066101074, -0.031799316, -0.0050315857, 0.036010742, 0.008285522, -0.07086182, 0.007270813, 0.001162529, 0.0111694...
http://arxiv.org/abs/2304.13462v1
Multidimensional Evaluation for Text Style Transfer Using ChatGPT
Huiyuan Lai, Antonio Toral, Malvina Nissim
cs.CL
We investigate the potential of ChatGPT as a multidimensional evaluator for the task of \emph{Text Style Transfer}, alongside, and in comparison to, existing automatic metrics as well as human judgements. We focus on a zero-shot setting, i.e. prompting ChatGPT with specific task instructions, and test its performance o...
http://arxiv.org/pdf/2304.13462v1
2023-04-26T11:33:35
2023-04-26T11:33:35
{ "title": [ -0.026657104, -0.007381439, -0.036468506, 0.0017337799, -0.012435913, -0.007205963, -0.0060195923, -0.0067367554, -0.015014648, 0.04928589, 0.00944519, 0.029525757, 0.025024414, -0.01260376, -0.013130188, 0.026687622, -0.023223877, 0...
http://arxiv.org/abs/2304.13343v1
Unleashing Infinite-Length Input Capacity for Large-scale Language Models with Self-Controlled Memory System
Xinnian Liang, Bing Wang, Hui Huang, Shuangzhi Wu, Peihao Wu, Lu Lu, Zejun Ma, Zhoujun Li
cs.CL
Large-scale Language Models (LLMs) are constrained by their inability to process lengthy inputs. To address this limitation, we propose the Self-Controlled Memory (SCM) system to unleash infinite-length input capacity for large-scale language models. Our SCM system is composed of three key modules: the language model a...
http://arxiv.org/pdf/2304.13343v1
2023-04-26T07:25:31
2023-04-26T07:25:31
{ "title": [ -0.02420044, 0.0028896332, -0.038848877, -0.061523438, -0.0061950684, 0.002483368, -0.053009033, 0.013244629, -0.0024337769, 0.040740967, -0.015129089, 0.003490448, -0.02204895, -0.0046081543, -0.061828613, 0.003528595, 0.0013465881, ...
http://arxiv.org/abs/2304.13337v2
Nominal Topology for Data Languages
Fabian Birkmann, Stefan Milius, Henning Urbat
cs.CL, F.4.3
We propose a novel topological perspective on data languages recognizable by orbit-finite nominal monoids. For this purpose, we introduce pro-orbit-finite nominal topological spaces. Assuming globally bounded support sizes, they coincide with nominal Stone spaces and are shown to be dually equivalent to a subcategory o...
http://arxiv.org/pdf/2304.13337v2
2023-05-10T06:50:37
2023-04-26T07:11:44
{ "title": [ -0.038116455, -0.015159607, -0.047729492, -0.087646484, -0.0020618439, 0.011978149, -0.060577393, 0.031677246, 0.017242432, 0.024978638, -0.039123535, -0.028625488, -0.015090942, -0.022125244, -0.039031982, -0.003929138, -0.004386902, ...
http://arxiv.org/abs/2304.13301v2
Prompting GPT-3.5 for Text-to-SQL with De-semanticization and Skeleton Retrieval
Chunxi Guo, Zhiliang Tian, Jintao Tang, Pancheng Wang, Zhihua Wen, Kang Yang, Ting Wang
cs.CL, cs.AI
Text-to-SQL is a task that converts a natural language question into a structured query language (SQL) to retrieve information from a database. Large language models (LLMs) work well in natural language generation tasks, but they are not specifically pre-trained to understand the syntax and semantics of SQL commands. I...
http://arxiv.org/pdf/2304.13301v2
2023-08-31T15:24:36
2023-04-26T06:02:01
{ "title": [ 0.01146698, 0.011451721, -0.03378296, -0.056549072, 0.013298035, -0.005748749, -0.036010742, -0.01360321, 0.035095215, 0.031921387, -0.05340576, -0.018341064, 0.022491455, -0.02154541, -0.040771484, -0.0056762695, 0.0044441223, 0.024...
http://arxiv.org/abs/2304.13292v1
Zero-Shot Slot and Intent Detection in Low-Resource Languages
Sang Yun Kwon, Gagan Bhatia, El Moatez Billah Nagoudi, Alcides Alcoba Inciarte, Muhammad Abdul-Mageed
cs.CL
Intent detection and slot filling are critical tasks in spoken and natural language understanding for task-oriented dialog systems. In this work we describe our participation in the slot and intent detection for low-resource language varieties (SID4LR; Aepli et al. (2023)). We investigate the slot and intent detection ...
http://arxiv.org/pdf/2304.13292v1
2023-04-26T05:10:12
2023-04-26T05:10:12
{ "title": [ -0.0051116943, -0.017578125, -0.030899048, -0.04373169, -0.014312744, -0.018966675, -0.016830444, 0.007358551, 0.027008057, 0.023727417, -0.019042969, -0.013153076, -0.0038833618, -0.028671265, -0.043548584, 0.00039577484, -0.013595581, ...
http://arxiv.org/abs/2304.13276v1
The Closeness of In-Context Learning and Weight Shifting for Softmax Regression
Shuai Li, Zhao Song, Yu Xia, Tong Yu, Tianyi Zhou
cs.CL, cs.LG
Large language models (LLMs) are known for their exceptional performance in natural language processing, making them highly effective in many human life-related or even job-related tasks. The attention mechanism in the Transformer architecture is a critical component of LLMs, as it allows the model to selectively focus...
http://arxiv.org/pdf/2304.13276v1
2023-04-26T04:33:41
2023-04-26T04:33:41
{ "title": [ -0.013221741, -0.0056114197, -0.029693604, -0.034362793, 0.012763977, -0.013664246, -0.0038394928, -0.020950317, -0.03817749, 0.02267456, -0.028121948, 0.022064209, -0.04144287, -0.018203735, -0.022109985, -0.0287323, 0.012176514, 0....
http://arxiv.org/abs/2304.13273v3
From Association to Generation: Text-only Captioning by Unsupervised Cross-modal Mapping
Junyang Wang, Ming Yan, Yi Zhang, Jitao Sang
cs.CV, cs.CL, cs.LG
With the development of Vision-Language Pre-training Models (VLPMs) represented by CLIP and ALIGN, significant breakthroughs have been achieved for association-based visual tasks such as image classification and image-text retrieval by the zero-shot capability of CLIP without fine-tuning. However, CLIP is hard to apply...
http://arxiv.org/pdf/2304.13273v3
2023-05-08T02:29:28
2023-04-26T04:06:20
{ "title": [ 0.0026874542, 0.017059326, -0.005214691, -0.011642456, -0.011062622, -0.00630188, -0.007911682, -0.049591064, 0.016357422, 0.057403564, -0.028762817, 0.015159607, 0.022155762, -0.009414673, -0.037109375, 0.010902405, -0.008644104, 0....
http://arxiv.org/abs/2304.13734v2
The Internal State of an LLM Knows When It's Lying
Amos Azaria, Tom Mitchell
cs.CL, cs.AI, cs.LG
While Large Language Models (LLMs) have shown exceptional performance in various tasks, one of their most prominent drawbacks is generating inaccurate or false information with a confident tone. In this paper, we provide evidence that the LLM's internal state can be used to reveal the truthfulness of statements. This i...
http://arxiv.org/pdf/2304.13734v2
2023-10-17T09:34:30
2023-04-26T02:49:38
{ "title": [ -0.008041382, -0.01687622, 0.010322571, 0.0231781, -0.07128906, 0.010749817, -0.013465881, -0.019805908, -0.019073486, 0.014282227, -0.06210327, -0.0014877319, -0.021606445, -0.025146484, 0.0256958, -0.042114258, 0.0109939575, 0.0252...
http://arxiv.org/abs/2304.13250v1
Exploring the Curious Case of Code Prompts
Li Zhang, Liam Dugan, Hainiu Xu, Chris Callison-Burch
cs.CL
Recent work has shown that prompting language models with code-like representations of natural language leads to performance improvements on structured reasoning tasks. However, such tasks comprise only a small subset of all natural language tasks. In our work, we seek to answer whether or not code-prompting is the pre...
http://arxiv.org/pdf/2304.13250v1
2023-04-26T02:37:52
2023-04-26T02:37:52
{ "title": [ 0.011016846, -0.0602417, 0.017944336, 0.033111572, -0.08074951, -0.023666382, 0.0259552, 0.014328003, -0.0073623657, 0.003929138, -0.060455322, -0.0013189316, -0.038604736, -0.036834717, 0.03111267, -0.045043945, -0.0017681122, 0.042...
http://arxiv.org/abs/2304.13191v1
Towards Explainable and Safe Conversational Agents for Mental Health: A Survey
Surjodeep Sarkar, Manas Gaur, L. Chen, Muskan Garg, Biplav Srivastava, Bhaktee Dongaonkar
cs.AI, cs.CL
Virtual Mental Health Assistants (VMHAs) are seeing continual advancements to support the overburdened global healthcare system that gets 60 million primary care visits, and 6 million Emergency Room (ER) visits annually. These systems are built by clinical psychologists, psychiatrists, and Artificial Intelligence (AI) ...
http://arxiv.org/pdf/2304.13191v1
2023-04-25T23:12:13
2023-04-25T23:12:13
{ "title": [ 0.0211792, -0.00017857552, -0.020446777, -0.08166504, 0.02230835, 0.0038986206, -0.037750244, -0.006942749, -0.0079956055, 0.040771484, -0.0064926147, 0.0033626556, 0.024505615, -0.029632568, -0.032562256, 0.003364563, -0.020507812, ...
http://arxiv.org/abs/2304.13188v1
TABLET: Learning From Instructions For Tabular Data
Dylan Slack, Sameer Singh
cs.LG, cs.CL
Acquiring high-quality data is often a significant challenge in training machine learning (ML) models for tabular prediction, particularly in privacy-sensitive and costly domains like medicine and finance. Providing natural language instructions to large language models (LLMs) offers an alternative solution. However, i...
http://arxiv.org/pdf/2304.13188v1
2023-04-25T23:07:20
2023-04-25T23:07:20
{ "title": [ -0.033843994, -0.013900757, -0.048065186, -0.049591064, -0.016296387, -0.006526947, -0.046691895, -0.003112793, 0.002538681, 0.023117065, -0.046966553, 0.016677856, -0.00724411, -0.010231018, -0.025619507, -0.014968872, 0.015266418, ...
http://arxiv.org/abs/2304.13180v2
Sebis at SemEval-2023 Task 7: A Joint System for Natural Language Inference and Evidence Retrieval from Clinical Trial Reports
Juraj Vladika, Florian Matthes
cs.CL
With the increasing number of clinical trial reports generated every day, it is becoming hard to keep up with novel discoveries that inform evidence-based healthcare recommendations. To help automate this process and assist medical experts, NLP solutions are being developed. This motivated the SemEval-2023 Task 7, wher...
http://arxiv.org/pdf/2304.13180v2
2023-05-02T16:46:33
2023-04-25T22:22:42
{ "title": [ -0.03414917, -0.030593872, -0.044677734, -0.04525757, -0.02166748, 0.012420654, -0.06439209, -0.001783371, -0.0037899017, 0.024353027, -0.030578613, 0.014297485, -0.007003784, -0.008132935, -0.033355713, -0.0015544891, -0.0236969, 0....
http://arxiv.org/abs/2304.13149v1
Modeling Spoken Information Queries for Virtual Assistants: Open Problems, Challenges and Opportunities
Christophe Van Gysel
cs.IR, cs.CL
Virtual assistants are becoming increasingly important speech-driven Information Retrieval platforms that assist users with various tasks. We discuss open problems and challenges with respect to modeling spoken information queries for virtual assistants, and list opportunities where Information Retrieval methods and ...
http://arxiv.org/pdf/2304.13149v1
2023-04-25T20:52:40
2023-04-25T20:52:40
{ "title": [ -0.010215759, -0.009384155, -0.037872314, -0.05618286, 0.007709503, -0.0063095093, -0.028884888, 0.013252258, 0.0027046204, 0.059143066, -0.018707275, -0.025100708, 0.026443481, 0.0011606216, -0.0770874, -0.0016489029, -0.009063721, ...
http://arxiv.org/abs/2304.13134v1
LAST: Scalable Lattice-Based Speech Modelling in JAX
Ke Wu, Ehsan Variani, Tom Bagby, Michael Riley
cs.CL
We introduce LAST, a LAttice-based Speech Transducer library in JAX. With an emphasis on flexibility, ease-of-use, and scalability, LAST implements differentiable weighted finite state automaton (WFSA) algorithms needed for training \& inference that scale to a large WFSA such as a recognition lattice over the entire u...
http://arxiv.org/pdf/2304.13134v1
2023-04-25T20:25:37
2023-04-25T20:25:37
{ "title": [ -0.036895752, -0.008880615, -0.074157715, -0.10253906, -0.008384705, 0.008369446, -0.10046387, 0.036254883, -0.03503418, 0.044128418, -0.007133484, -0.002155304, 0.010070801, -0.0284729, -0.043060303, 0.0057029724, -0.01689148, 0.040...
http://arxiv.org/abs/2304.13060v2
Injecting structural hints: Using language models to study inductive biases in language learning
Isabel Papadimitriou, Dan Jurafsky
cs.CL
Both humans and large language models are able to learn language without explicit structural supervision. What inductive biases make this learning possible? We address this fundamental cognitive question by leveraging transformer language models: we inject inductive bias into language models by pretraining on formally-...
http://arxiv.org/pdf/2304.13060v2
2023-10-29T17:14:06
2023-04-25T18:00:08
{ "title": [ -0.0206604, -0.012557983, -0.015052795, -0.016418457, -0.021408081, -0.01461792, -0.015777588, -0.023391724, -0.02255249, 0.006061554, -0.04626465, 0.017959595, -0.012016296, -0.013877869, -0.022094727, -0.025787354, -0.03152466, 0.0...
http://arxiv.org/abs/2304.13010v2
Unstructured and structured data: Can we have the best of both worlds with large language models?
Wang-Chiew Tan
cs.DB, cs.CL
This paper presents an opinion on the potential of using large language models to query on both unstructured and structured data. It also outlines some research challenges related to the topic of building question-answering systems for both types of data.
http://arxiv.org/pdf/2304.13010v2
2023-07-05T21:33:33
2023-04-25T17:30:05
{ "title": [ -0.04788208, -0.0035514832, -0.055908203, -0.09246826, 0.005622864, 0.0068473816, -0.08081055, 0.013130188, 0.0013561249, 0.025299072, -0.017669678, -0.002254486, -0.012557983, -0.019958496, -0.03894043, -0.014709473, 0.0005950928, 0...
http://arxiv.org/abs/2304.13007v3
Answering Questions by Meta-Reasoning over Multiple Chains of Thought
Ori Yoran, Tomer Wolfson, Ben Bogin, Uri Katz, Daniel Deutch, Jonathan Berant
cs.CL, cs.AI
Modern systems for multi-hop question answering (QA) typically break questions into a sequence of reasoning steps, termed chain-of-thought (CoT), before arriving at a final answer. Often, multiple chains are sampled and aggregated through a voting mechanism over the final answers, but the intermediate steps themselves ...
http://arxiv.org/pdf/2304.13007v3
2023-10-17T19:18:05
2023-04-25T17:27:37
{ "title": [ -0.010643005, -0.019638062, -0.038970947, -0.058929443, 0.00050354004, -0.0015525818, -0.018554688, -0.010726929, 0.011528015, 0.020553589, -0.006225586, -0.010810852, -0.026626587, -0.03363037, -0.04156494, 0.0042686462, 0.014816284, ...
http://arxiv.org/abs/2304.13005v2
Evaluating Inter-Bilingual Semantic Parsing for Indian Languages
Divyanshu Aggarwal, Vivek Gupta, Anoop Kunchukuttan
cs.CL
Despite significant progress in Natural Language Generation for Indian languages (IndicNLP), there is a lack of datasets around complex structured tasks such as semantic parsing. One reason for this imminent gap is the complexity of the logical form, which makes English to multilingual translation difficult. The proces...
http://arxiv.org/pdf/2304.13005v2
2023-06-05T15:55:47
2023-04-25T17:24:32
{ "title": [ -0.03869629, 0.0007805824, -0.030075073, -0.060943604, -0.007045746, 0.012893677, -0.076171875, 0.00021016598, -0.035003662, 0.031402588, -0.02722168, 0.011833191, -0.030441284, -0.0022258759, -0.054260254, -0.026473999, -0.027282715, ...