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/2309.07689v1
Detecting ChatGPT: A Survey of the State of Detecting ChatGPT-Generated Text
Mahdi Dhaini, Wessel Poelman, Ege Erdogan
cs.CL, cs.AI
While recent advancements in the capabilities and widespread accessibility of generative language models, such as ChatGPT (OpenAI, 2022), have brought about various benefits by generating fluent human-like text, the task of distinguishing between human- and large language model (LLM) generated text has emerged as a cru...
http://arxiv.org/pdf/2309.07689v1
2023-09-14T13:05:20
2023-09-14T13:05:20
{ "title": [ -0.0037708282, 0.0015068054, -0.008270264, -0.018310547, 0.0033054352, 0.00554657, -0.009391785, -0.032928467, -0.032104492, 0.02116394, -0.011116028, 0.032562256, -0.022583008, -0.013977051, -0.027908325, 0.00356102, -0.037506104, 0...
http://arxiv.org/abs/2309.07683v1
Assessing the nature of large language models: A caution against anthropocentrism
Ann Speed
cs.AI, cs.CL, cs.CY, cs.HC
Generative AI models garnered a large amount of public attention and speculation with the release of OpenAIs chatbot, ChatGPT. At least two opinion camps exist: one excited about possibilities these models offer for fundamental changes to human tasks, and another highly concerned about power these models seem to have. ...
http://arxiv.org/pdf/2309.07683v1
2023-09-14T12:58:30
2023-09-14T12:58:30
{ "title": [ -0.039978027, -0.009994507, -0.048675537, -0.042388916, -0.005519867, -0.015609741, -0.04550171, -0.0025691986, -0.016021729, 0.027008057, -0.012634277, -0.0016832352, -0.011619568, -0.016571045, -0.05114746, -0.0010890961, -0.021224976,...
http://arxiv.org/abs/2309.07682v1
A Conversation is Worth A Thousand Recommendations: A Survey of Holistic Conversational Recommender Systems
Chuang Li, Hengchang Hu, Yan Zhang, Min-Yen Kan, Haizhou Li
cs.CL, cs.IR
Conversational recommender systems (CRS) generate recommendations through an interactive process. However, not all CRS approaches use human conversations as their source of interaction data; the majority of prior CRS work simulates interactions by exchanging entity-level information. As a result, claims of prior CRS wo...
http://arxiv.org/pdf/2309.07682v1
2023-09-14T12:55:23
2023-09-14T12:55:23
{ "title": [ 0.0050315857, -0.004322052, -0.045288086, -0.05154419, 0.0084991455, -0.008087158, -0.032440186, -0.0039901733, 0.00724411, 0.023208618, -0.02168274, 0.02494812, -0.009162903, -0.015914917, -0.038482666, 0.0077819824, -0.016479492, 0...
http://arxiv.org/abs/2309.07677v1
Aligning Speakers: Evaluating and Visualizing Text-based Diarization Using Efficient Multiple Sequence Alignment (Extended Version)
Chen Gong, Peilin Wu, Jinho D. Choi
cs.CL
This paper presents a novel evaluation approach to text-based speaker diarization (SD), tackling the limitations of traditional metrics that do not account for any contextual information in text. Two new metrics are proposed, Text-based Diarization Error Rate and Diarization F1, which perform utterance- and word-level ...
http://arxiv.org/pdf/2309.07677v1
2023-09-14T12:43:26
2023-09-14T12:43:26
{ "title": [ -0.013061523, 0.018814087, -0.03363037, -0.04953003, -0.006336212, 0.0016937256, -0.04876709, -0.022842407, -0.0069732666, 0.04925537, -0.030395508, 0.014953613, 0.014381409, -0.007335663, -0.035858154, -0.0056533813, -0.004787445, 0...
http://arxiv.org/abs/2309.07650v1
Automatic Data Visualization Generation from Chinese Natural Language Questions
Yan Ge, Victor Junqiu Wei, Yuanfeng Song, Jason Chen Zhang, Raymond Chi-Wing Wong
cs.CL
Data visualization has emerged as an effective tool for getting insights from massive datasets. Due to the hardness of manipulating the programming languages of data visualization, automatic data visualization generation from natural languages (Text-to-Vis) is becoming increasingly popular. Despite the plethora of rese...
http://arxiv.org/pdf/2309.07650v1
2023-09-14T12:16:21
2023-09-14T12:16:21
{ "title": [ -0.029800415, -0.008621216, -0.015602112, -0.036010742, -0.015304565, 0.009651184, -0.020477295, -0.016113281, 0.026748657, 0.016952515, -0.02468872, -0.0024776459, -0.0390625, -0.0236969, -0.056610107, -0.0137786865, 0.0014467239, 0...
http://arxiv.org/abs/2309.07648v1
Incorporating Class-based Language Model for Named Entity Recognition in Factorized Neural Transducer
Peng Wang, Yifan Yang, Zheng Liang, Tian Tan, Shiliang Zhang, Xie Chen
eess.AS, cs.CL, cs.SD
In spite of the excellent strides made by end-to-end (E2E) models in speech recognition in recent years, named entity recognition is still challenging but critical for semantic understanding. In order to enhance the ability to recognize named entities in E2E models, previous studies mainly focus on various rule-based o...
http://arxiv.org/pdf/2309.07648v1
2023-09-14T12:14:49
2023-09-14T12:14:49
{ "title": [ 0.0030174255, 0.0049972534, -0.038604736, -0.053497314, 0.006038666, -0.007106781, -0.03829956, -0.0028419495, -0.023864746, 0.03036499, -0.0004286766, 0.00554657, 0.011787415, -0.024505615, -0.026062012, 0.0052986145, -0.011650085, ...
http://arxiv.org/abs/2309.07624v1
Dynamic MOdularized Reasoning for Compositional Structured Explanation Generation
Xiyan Fu, Anette Frank
cs.CL
Despite the success of neural models in solving reasoning tasks, their compositional generalization capabilities remain unclear. In this work, we propose a new setting of the structured explanation generation task to facilitate compositional reasoning research. Previous works found that symbolic methods achieve superio...
http://arxiv.org/pdf/2309.07624v1
2023-09-14T11:40:30
2023-09-14T11:40:30
{ "title": [ -0.028396606, -0.021026611, -0.054534912, -0.13574219, -0.0017518997, 0.03955078, -0.09082031, 0.04067993, -0.019607544, 0.016296387, 0.000096678734, 0.019424438, -0.036254883, -0.017150879, -0.017868042, 0.0002565384, -0.0032176971, ...
http://arxiv.org/abs/2309.07606v1
Zero-shot Audio Topic Reranking using Large Language Models
Mengjie Qian, Rao Ma, Adian Liusie, Erfan Loweimi, Kate M. Knill, Mark J. F. Gales
cs.CL, cs.IR
The Multimodal Video Search by Examples (MVSE) project investigates using video clips as the query term for information retrieval, rather than the more traditional text query. This enables far richer search modalities such as images, speaker, content, topic, and emotion. A key element for this process is highly rapid, ...
http://arxiv.org/pdf/2309.07606v1
2023-09-14T11:13:36
2023-09-14T11:13:36
{ "title": [ -0.0072517395, 0.00084781647, -0.024017334, -0.046936035, -0.01222229, 0.008102417, -0.050933838, -0.007774353, 0.0236969, 0.0541687, -0.007686615, 0.007637024, 0.026428223, -0.013458252, -0.042022705, 0.012420654, -0.008178711, 0.00...
http://arxiv.org/abs/2309.07601v1
Detecting Misinformation with LLM-Predicted Credibility Signals and Weak Supervision
João A. Leite, Olesya Razuvayevskaya, Kalina Bontcheva, Carolina Scarton
cs.CL, cs.AI, cs.LG
Credibility signals represent a wide range of heuristics that are typically used by journalists and fact-checkers to assess the veracity of online content. Automating the task of credibility signal extraction, however, is very challenging as it requires high-accuracy signal-specific extractors to be trained, while ther...
http://arxiv.org/pdf/2309.07601v1
2023-09-14T11:06:51
2023-09-14T11:06:51
{ "title": [ -0.015609741, 0.0066184998, -0.04324341, -0.035949707, -0.007461548, -0.018432617, -0.018341064, -0.013534546, -0.05279541, 0.018051147, -0.0029678345, 0.027236938, -0.026153564, -0.04071045, 0.0014038086, -0.016540527, -0.029205322, ...
http://arxiv.org/abs/2309.07597v1
C-Pack: Packaged Resources To Advance General Chinese Embedding
Shitao Xiao, Zheng Liu, Peitian Zhang, Niklas Muennighof
cs.CL, cs.AI, cs.IR
We introduce C-Pack, a package of resources that significantly advance the field of general Chinese embeddings. C-Pack includes three critical resources. 1) C-MTEB is a comprehensive benchmark for Chinese text embeddings covering 6 tasks and 35 datasets. 2) C-MTP is a massive text embedding dataset curated from labeled...
http://arxiv.org/pdf/2309.07597v1
2023-09-14T10:57:50
2023-09-14T10:57:50
{ "title": [ -0.0098724365, -0.008987427, -0.033233643, -0.033996582, -0.020523071, -0.016921997, -0.020248413, -0.002840042, -0.02331543, 0.03451538, -0.02633667, 0.014785767, 0.035095215, -0.055633545, 0.015403748, 0.007980347, 0.0065841675, 0....
http://arxiv.org/abs/2309.07590v1
Revisiting Supertagging for HPSG
Olga Zamaraeva, Carlos Gómez-Rodríguez
cs.CL
We present new supertaggers trained on HPSG-based treebanks. These treebanks feature high-quality annotation based on a well-developed linguistic theory and include diverse and challenging test datasets, beyond the usual WSJ section 23 and Wikipedia data. HPSG supertagging has previously relied on MaxEnt-based models. ...
http://arxiv.org/pdf/2309.07590v1
2023-09-14T10:49:16
2023-09-14T10:49:16
{ "title": [ 0.008865356, 0.0046577454, -0.053833008, -0.06072998, -0.0060043335, -0.008110046, -0.023757935, 0.020187378, -0.01751709, 0.056915283, 0.0013866425, -0.016021729, 0.037719727, -0.04095459, -0.01184845, 0.007698059, 0.028213501, -0.0...
http://arxiv.org/abs/2309.07561v1
Adaptive Prompt Learning with Distilled Connective Knowledge for Implicit Discourse Relation Recognition
Bang Wang, Zhenglin Wang, Wei Xiang, Yijun Mo
cs.CL
Implicit discourse relation recognition (IDRR) aims at recognizing the discourse relation between two text segments without an explicit connective. Recently, the prompt learning has just been applied to the IDRR task with great performance improvements over various neural network-based approaches. However, the discrete...
http://arxiv.org/pdf/2309.07561v1
2023-09-14T09:44:46
2023-09-14T09:44:46
{ "title": [ 0.012374878, -0.0018320084, -0.04260254, -0.08380127, -0.013694763, 0.00111866, -0.06994629, 0.007911682, 0.057128906, 0.0018892288, -0.035308838, -0.024749756, -0.023483276, -0.023086548, -0.05114746, -0.029327393, -0.010215759, 0.0...
http://arxiv.org/abs/2309.07545v2
DBLPLink: An Entity Linker for the DBLP Scholarly Knowledge Graph
Debayan Banerjee, Arefa, Ricardo Usbeck, Chris Biemann
cs.CL
In this work, we present a web application named DBLPLink, which performs entity linking over the DBLP scholarly knowledge graph. DBLPLink uses text-to-text pre-trained language models, such as T5, to produce entity label spans from an input text question. Entity candidates are fetched from a database based on the labe...
http://arxiv.org/pdf/2309.07545v2
2023-09-25T08:44:19
2023-09-14T09:15:36
{ "title": [ -0.010284424, 0.0124435425, -0.0847168, -0.068115234, 0.031921387, -0.035980225, -0.037231445, -0.01096344, -0.002105713, 0.037963867, 0.0006284714, -0.007820129, 0.012992859, -0.033172607, -0.040252686, 0.0054969788, 0.0013256073, 0...
http://arxiv.org/abs/2309.07478v1
Direct Text to Speech Translation System using Acoustic Units
Victoria Mingote, Pablo Gimeno, Luis Vicente, Sameer Khurana, Antoine Laurent, Jarod Duret
cs.CL, cs.LG, cs.SD, eess.AS
This paper proposes a direct text to speech translation system using discrete acoustic units. This framework employs text in different source languages as input to generate speech in the target language without the need for text transcriptions in this language. Motivated by the success of acoustic units in previous wor...
http://arxiv.org/pdf/2309.07478v1
2023-09-14T07:35:14
2023-09-14T07:35:14
{ "title": [ -0.014816284, -0.008743286, -0.035583496, -0.058502197, -0.017456055, -0.009101868, -0.040740967, 0.017303467, -0.019699097, 0.07458496, -0.02583313, 0.009033203, 0.060546875, -0.025268555, -0.029586792, 0.013366699, 0.0031833649, 0....
http://arxiv.org/abs/2309.07462v1
Are Large Language Model-based Evaluators the Solution to Scaling Up Multilingual Evaluation?
Rishav Hada, Varun Gumma, Adrian de Wynter, Harshita Diddee, Mohamed Ahmed, Monojit Choudhury, Kalika Bali, Sunayana Sitaram
cs.CL
Large Language Models (LLMs) have demonstrated impressive performance on Natural Language Processing (NLP) tasks, such as Question Answering, Summarization, and Classification. The use of LLMs as evaluators, that can rank or score the output of other models (usually LLMs) has become increasingly popular, due to the lim...
http://arxiv.org/pdf/2309.07462v1
2023-09-14T06:41:58
2023-09-14T06:41:58
{ "title": [ -0.00680542, -0.00831604, -0.018798828, -0.04208374, -0.02659607, -0.0013418198, -0.037109375, 0.0052871704, -0.013793945, 0.03161621, 0.010673523, -0.023101807, -0.012191772, -0.042877197, -0.020065308, -0.027633667, -0.019927979, 0...
http://arxiv.org/abs/2309.07445v1
SIB-200: A Simple, Inclusive, and Big Evaluation Dataset for Topic Classification in 200+ Languages and Dialects
David Ifeoluwa Adelani, Hannah Liu, Xiaoyu Shen, Nikita Vassilyev, Jesujoba O. Alabi, Yanke Mao, Haonan Gao, Annie En-Shiun Lee
cs.CL
Despite the progress we have recorded in the last few years in multilingual natural language processing, evaluation is typically limited to a small set of languages with available datasets which excludes a large number of low-resource languages. In this paper, we created SIB-200 -- a large-scale open-sourced benchmark ...
http://arxiv.org/pdf/2309.07445v1
2023-09-14T05:56:49
2023-09-14T05:56:49
{ "title": [ -0.006362915, -0.003396988, -0.050567627, -0.05545044, 0.03375244, -0.017227173, -0.042266846, -0.011184692, 0.01890564, 0.014312744, -0.016693115, 0.028198242, -0.021362305, -0.02230835, -0.02557373, 0.003030777, -0.016326904, 0.038...
http://arxiv.org/abs/2309.07430v3
Clinical Text Summarization: Adapting Large Language Models Can Outperform Human Experts
Dave Van Veen, Cara Van Uden, Louis Blankemeier, Jean-Benoit Delbrouck, Asad Aali, Christian Bluethgen, Anuj Pareek, Malgorzata Polacin, Eduardo Pontes Reis, Anna Seehofnerova, Nidhi Rohatgi, Poonam Hosamani, William Collins, Neera Ahuja, Curtis P. Langlotz, Jason Hom, Sergios Gatidis, John Pauly, Akshay S. Chaudhari
cs.CL
Sifting through vast textual data and summarizing key information from electronic health records (EHR) imposes a substantial burden on how clinicians allocate their time. Although large language models (LLMs) have shown immense promise in natural language processing (NLP) tasks, their efficacy on a diverse range of cli...
http://arxiv.org/pdf/2309.07430v3
2023-10-24T02:04:59
2023-09-14T05:15:01
{ "title": [ -0.023773193, -0.0078125, -0.024841309, -0.07727051, -0.022338867, 0.017242432, -0.07745361, -0.016998291, -0.0051345825, 0.04864502, 0.00667572, 0.004131317, 0.0021266937, -0.0015106201, -0.065979004, 0.013801575, 0.0011854172, 0.03...
http://arxiv.org/abs/2309.07429v1
Semantic Parsing in Limited Resource Conditions
Zhuang Li
cs.CL, cs.AI
This thesis explores challenges in semantic parsing, specifically focusing on scenarios with limited data and computational resources. It offers solutions using techniques like automatic data curation, knowledge transfer, active learning, and continual learning. For tasks with no parallel training data, the thesis pr...
http://arxiv.org/pdf/2309.07429v1
2023-09-14T05:03:09
2023-09-14T05:03:09
{ "title": [ -0.026641846, -0.006248474, -0.038391113, -0.08843994, 0.0041007996, 0.011810303, -0.066711426, 0.008758545, -0.0030212402, 0.025222778, -0.03805542, -0.0030708313, -0.027389526, -0.0035686493, -0.052886963, -0.010139465, -0.016921997, ...
http://arxiv.org/abs/2309.07423v1
ChatGPT MT: Competitive for High- (but not Low-) Resource Languages
Nathaniel R. Robinson, Perez Ogayo, David R. Mortensen, Graham Neubig
cs.CL
Large language models (LLMs) implicitly learn to perform a range of language tasks, including machine translation (MT). Previous studies explore aspects of LLMs' MT capabilities. However, there exist a wide variety of languages for which recent LLM MT performance has never before been evaluated. Without published exper...
http://arxiv.org/pdf/2309.07423v1
2023-09-14T04:36:00
2023-09-14T04:36:00
{ "title": [ 0.0023708344, -0.029769897, -0.00434494, -0.033416748, -0.022979736, 0.0018796921, -0.023666382, -0.008148193, -0.031799316, 0.024932861, -0.006427765, 0.01574707, -0.00642395, -0.034729004, 0.000726223, 0.009300232, -0.0234375, 0.04...
http://arxiv.org/abs/2309.07414v2
PromptASR for contextualized ASR with controllable style
Xiaoyu Yang, Wei Kang, Zengwei Yao, Yifan Yang, Liyong Guo, Fangjun Kuang, Long Lin, Daniel Povey
eess.AS, cs.CL, cs.SD
Prompts are crucial to large language models as they provide context information such as topic or logical relationships. Inspired by this, we propose PromptASR, a framework that integrates prompts in end-to-end automatic speech recognition (E2E ASR) systems to achieve contextualized ASR with controllable style of trans...
http://arxiv.org/pdf/2309.07414v2
2023-09-20T09:13:11
2023-09-14T03:43:07
{ "title": [ 0.011878967, 0.00957489, -0.068237305, -0.061920166, -0.016204834, -0.015060425, -0.07092285, 0.0064353943, -0.043884277, 0.047546387, -0.020095825, -0.0053253174, 0.051849365, -0.043823242, 0.023712158, 0.0071983337, 0.023345947, 0....
http://arxiv.org/abs/2309.07413v2
CPPF: A contextual and post-processing-free model for automatic speech recognition
Lei Zhang, Zhengkun Tian, Xiang Chen, Jiaming Sun, Hongyu Xiang, Ke Ding, Guanglu Wan
cs.CL, cs.SD, eess.AS
ASR systems have become increasingly widespread in recent years. However, their textual outputs often require post-processing tasks before they can be practically utilized. To address this issue, we draw inspiration from the multifaceted capabilities of LLMs and Whisper, and focus on integrating multiple ASR text proce...
http://arxiv.org/pdf/2309.07413v2
2023-09-21T03:02:27
2023-09-14T03:40:14
{ "title": [ -0.011169434, 0.0078125, -0.056640625, -0.07952881, 0.018920898, 0.0033607483, -0.059692383, 0.011070251, 0.009284973, 0.043029785, -0.024291992, -0.00081062317, -0.0038833618, -0.009277344, -0.06536865, 0.0009198189, -0.024597168, 0...
http://arxiv.org/abs/2309.07412v1
Advancing Regular Language Reasoning in Linear Recurrent Neural Networks
Ting-Han Fan, Ta-Chung Chi, Alexander I. Rudnicky
cs.CL, cs.LG
In recent studies, linear recurrent neural networks (LRNNs) have achieved Transformer-level performance in natural language modeling and long-range modeling while offering rapid parallel training and constant inference costs. With the resurged interest in LRNNs, we study whether they can learn the hidden rules in train...
http://arxiv.org/pdf/2309.07412v1
2023-09-14T03:36:01
2023-09-14T03:36:01
{ "title": [ -0.02243042, -0.01864624, -0.03225708, -0.09387207, 0.004310608, 0.014343262, -0.066101074, 0.036315918, -0.0158844, 0.01424408, 0.0005559921, 0.008621216, -0.028198242, -0.022750854, -0.022094727, 0.008026123, -0.006374359, 0.020156...
http://arxiv.org/abs/2309.07396v1
DebCSE: Rethinking Unsupervised Contrastive Sentence Embedding Learning in the Debiasing Perspective
Pu Miao, Zeyao Du, Junlin Zhang
cs.CL, cs.AI
Several prior studies have suggested that word frequency biases can cause the Bert model to learn indistinguishable sentence embeddings. Contrastive learning schemes such as SimCSE and ConSERT have already been adopted successfully in unsupervised sentence embedding to improve the quality of embeddings by reducing this...
http://arxiv.org/pdf/2309.07396v1
2023-09-14T02:43:34
2023-09-14T02:43:34
{ "title": [ 0.008544922, 0.016906738, -0.01436615, -0.032592773, 0.0076560974, -0.013557434, -0.019714355, -0.044799805, 0.02734375, 0.012184143, -0.03781128, 0.009399414, -0.0029067993, -0.010948181, -0.038391113, -0.005252838, -0.020309448, 0....
http://arxiv.org/abs/2309.07387v1
VDialogUE: A Unified Evaluation Benchmark for Visually-grounded Dialogue
Yunshui Li, Binyuan Hui, Zhaochao Yin, Wanwei He, Run Luo, Yuxing Long, Min Yang, Fei Huang, Yongbin Li
cs.CL, cs.CV
Visually-grounded dialog systems, which integrate multiple modes of communication such as text and visual inputs, have become an increasingly popular area of investigation. However, the absence of a standardized evaluation framework poses a challenge in assessing the development of this field. To this end, we propose \...
http://arxiv.org/pdf/2309.07387v1
2023-09-14T02:09:20
2023-09-14T02:09:20
{ "title": [ -0.013572693, 0.00022792816, -0.04611206, -0.06768799, 0.01777649, -0.0038108826, -0.044128418, 0.001499176, 0.028182983, 0.0385437, -0.012542725, 0.0058135986, -0.0053596497, -0.011726379, -0.08251953, -0.0021591187, -0.017456055, 0...
http://arxiv.org/abs/2309.07384v1
An Interactive Framework for Profiling News Media Sources
Nikhil Mehta, Dan Goldwasser
cs.CL
The recent rise of social media has led to the spread of large amounts of fake and biased news, content published with the intent to sway beliefs. While detecting and profiling the sources that spread this news is important to maintain a healthy society, it is challenging for automated systems. In this paper, we prop...
http://arxiv.org/pdf/2309.07384v1
2023-09-14T02:03:45
2023-09-14T02:03:45
{ "title": [ 0.0042419434, 0.009597778, -0.035217285, -0.04550171, 0.0060539246, -0.019012451, 0.0038604736, -0.019729614, -0.03527832, 0.042877197, -0.025253296, 0.020355225, -0.03665161, -0.009376526, -0.044036865, -0.016052246, 0.0021514893, 0...
http://arxiv.org/abs/2309.07382v1
Less is More for Long Document Summary Evaluation by LLMs
Yunshu Wu, Hayate Iso, Pouya Pezeshkpour, Nikita Bhutani, Estevam Hruschka
cs.CL
Large Language Models (LLMs) have shown promising performance in summary evaluation tasks, yet they face challenges such as high computational costs and the Lost-in-the-Middle problem where important information in the middle of long documents is often overlooked. To address these issues, this paper introduces a novel ...
http://arxiv.org/pdf/2309.07382v1
2023-09-14T01:59:15
2023-09-14T01:59:15
{ "title": [ -0.012535095, -0.012886047, -0.053497314, -0.018661499, -0.042755127, -0.03781128, -0.02456665, 0.0021133423, -0.050842285, 0.016204834, -0.054534912, 0.01574707, 0.0026683807, -0.036895752, 0.034454346, -0.044555664, -0.005329132, 0...
http://arxiv.org/abs/2309.07369v1
Hybrid Attention-based Encoder-decoder Model for Efficient Language Model Adaptation
Shaoshi Ling, Guoli Ye, Rui Zhao, Yifan Gong
eess.AS, cs.CL, cs.SD
Attention-based encoder-decoder (AED) speech recognition model has been widely successful in recent years. However, the joint optimization of acoustic model and language model in end-to-end manner has created challenges for text adaptation. In particular, effectively, quickly and inexpensively adapting text has become ...
http://arxiv.org/pdf/2309.07369v1
2023-09-14T01:07:36
2023-09-14T01:07:36
{ "title": [ -0.024658203, 0.0010223389, -0.038330078, -0.048553467, -0.013824463, 0.004650116, -0.059143066, 0.00010108948, 0.021484375, 0.047302246, -0.009086609, -0.010681152, 0.03857422, -0.009773254, -0.04196167, 0.02458191, -0.004688263, 0....
http://arxiv.org/abs/2309.07334v1
Learning from Auxiliary Sources in Argumentative Revision Classification
Tazin Afrin, Diane Litman
cs.CL, cs.AI
We develop models to classify desirable reasoning revisions in argumentative writing. We explore two approaches -- multi-task learning and transfer learning -- to take advantage of auxiliary sources of revision data for similar tasks. Results of intrinsic and extrinsic evaluations show that both approaches can indeed i...
http://arxiv.org/pdf/2309.07334v1
2023-09-13T22:08:12
2023-09-13T22:08:12
{ "title": [ -0.017303467, 0.005859375, -0.04171753, -0.07147217, -0.0044555664, -0.00004506111, -0.06561279, -0.0066223145, -0.03326416, 0.003967285, -0.040130615, 0.012466431, -0.030075073, -0.009124756, -0.019866943, -0.04650879, -0.016998291, ...
http://arxiv.org/abs/2309.07315v2
Traveling Words: A Geometric Interpretation of Transformers
Raul Molina
cs.CL, cs.AI, cs.LG
Transformers have significantly advanced the field of natural language processing, but comprehending their internal mechanisms remains a challenge. In this paper, we introduce a novel geometric perspective that elucidates the inner mechanisms of transformer operations. Our primary contribution is illustrating how layer...
http://arxiv.org/pdf/2309.07315v2
2023-09-19T00:34:56
2023-09-13T21:01:03
{ "title": [ -0.026641846, -0.0057907104, -0.0050315857, -0.039001465, 0.00021517277, 0.0018177032, -0.004371643, -0.022506714, -0.024795532, 0.035949707, -0.04144287, 0.01084137, -0.009132385, -0.012161255, -0.029953003, -0.013931274, 0.015991211, ...
http://arxiv.org/abs/2309.07311v3
Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMs
Angelica Chen, Ravid Shwartz-Ziv, Kyunghyun Cho, Matthew L. Leavitt, Naomi Saphra
cs.CL
Most interpretability research in NLP focuses on understanding the behavior and features of a fully trained model. However, certain insights into model behavior may only be accessible by observing the trajectory of the training process. We present a case study of syntax acquisition in masked language models (MLMs) that...
http://arxiv.org/pdf/2309.07311v3
2023-09-28T21:30:20
2023-09-13T20:57:11
{ "title": [ -0.02835083, -0.0012836456, -0.027740479, -0.008552551, -0.02507019, -0.010063171, -0.029525757, -0.008972168, -0.04046631, 0.022735596, -0.024017334, 0.026550293, -0.008605957, -0.025009155, 0.0045547485, -0.025268555, -0.005508423, ...
http://arxiv.org/abs/2309.08632v1
Pretraining on the Test Set Is All You Need
Rylan Schaeffer
cs.CL, cs.AI
Inspired by recent work demonstrating the promise of smaller Transformer-based language models pretrained on carefully curated data, we supercharge such approaches by investing heavily in curating a novel, high quality, non-synthetic data mixture based solely on evaluation benchmarks. Using our novel dataset mixture co...
http://arxiv.org/pdf/2309.08632v1
2023-09-13T19:47:33
2023-09-13T19:47:33
{ "title": [ 0.045684814, -0.004837036, -0.016082764, -0.08288574, -0.01687622, 0.0030555725, -0.055358887, 0.033416748, -0.017547607, 0.04562378, -0.0013341904, 0.0017852783, -0.020965576, -0.027328491, -0.008415222, -0.022247314, 0.038909912, -...
http://arxiv.org/abs/2309.07251v1
In-Contextual Bias Suppression for Large Language Models
Daisuke Oba, Masahiro Kaneko, Danushka Bollegala
cs.CL
Despite their impressive performance in a wide range of NLP tasks, Large Language Models (LLMs) have been reported to encode worrying-levels of gender bias. Prior work has proposed debiasing methods that require human labelled examples, data augmentation and fine-tuning of the LLMs, which are computationally costly. Mo...
http://arxiv.org/pdf/2309.07251v1
2023-09-13T18:39:08
2023-09-13T18:39:08
{ "title": [ -0.008735657, -0.000045120716, -0.05493164, -0.058380127, 0.012969971, -0.013748169, -0.045928955, -0.0016622543, 0.03062439, 0.041778564, -0.02508545, -0.010803223, 0.040283203, -0.016494751, -0.03842163, 0.02166748, -0.022003174, 0...
http://arxiv.org/abs/2309.07124v2
RAIN: Your Language Models Can Align Themselves without Finetuning
Yuhui Li, Fangyun Wei, Jinjing Zhao, Chao Zhang, Hongyang Zhang
cs.CL
Large language models (LLMs) often demonstrate inconsistencies with human preferences. Previous research typically gathered human preference data and then aligned the pre-trained models using reinforcement learning or instruction tuning, a.k.a. the finetuning step. In contrast, aligning frozen LLMs without requiring al...
http://arxiv.org/pdf/2309.07124v2
2023-10-09T03:34:01
2023-09-13T17:59:09
{ "title": [ 0.003063202, -0.034301758, -0.05267334, -0.08251953, -0.035980225, -0.0026683807, -0.052520752, 0.039489746, -0.01701355, 0.036010742, -0.021911621, -0.0016613007, 0.009407043, -0.029220581, -0.012214661, -0.012199402, 0.004497528, 0...
http://arxiv.org/abs/2309.07120v1
Sight Beyond Text: Multi-Modal Training Enhances LLMs in Truthfulness and Ethics
Haoqin Tu, Bingchen Zhao, Chen Wei, Cihang Xie
cs.CL, cs.AI, cs.CV, cs.CY, cs.LG
Multi-modal large language models (MLLMs) are trained based on large language models (LLM), with an enhanced capability to comprehend multi-modal inputs and generate textual responses. While they excel in multi-modal tasks, the pure NLP abilities of MLLMs are often underestimated and left untested. In this study, we ge...
http://arxiv.org/pdf/2309.07120v1
2023-09-13T17:57:21
2023-09-13T17:57:21
{ "title": [ -0.015640259, -0.0051345825, -0.057647705, -0.005279541, -0.039245605, -0.03753662, -0.043029785, -0.016464233, -0.01322937, 0.010360718, -0.015365601, 0.015419006, 0.026992798, -0.05230713, 0.031433105, -0.010871887, -0.03314209, 0....
http://arxiv.org/abs/2309.07098v1
Mitigating Hallucinations and Off-target Machine Translation with Source-Contrastive and Language-Contrastive Decoding
Rico Sennrich, Jannis Vamvas, Alireza Mohammadshahi
cs.CL
Hallucinations and off-target translation remain unsolved problems in machine translation, especially for low-resource languages and massively multilingual models. In this paper, we introduce methods to mitigate both failure cases with a modified decoding objective, without requiring retraining or external models. In s...
http://arxiv.org/pdf/2309.07098v1
2023-09-13T17:15:27
2023-09-13T17:15:27
{ "title": [ -0.006298065, -0.008476257, 0.008300781, -0.026229858, -0.028839111, 0.004310608, -0.016708374, -0.027938843, 0.0024585724, 0.05899048, 0.0062942505, 0.012321472, 0.036895752, -0.02217102, -0.051239014, 0.033325195, -0.0029506683, 0....
http://arxiv.org/abs/2309.07081v1
Can Whisper perform speech-based in-context learning
Siyin Wang, Chao-Han Huck Yang, Ji Wu, Chao Zhang
eess.AS, cs.CL, cs.SD
This paper investigates the in-context learning abilities of the Whisper automatic speech recognition (ASR) models released by OpenAI. A novel speech-based in-context learning (SICL) approach is proposed for test-time adaptation, which can reduce the word error rates (WERs) with only a small number of labelled speech s...
http://arxiv.org/pdf/2309.07081v1
2023-09-13T16:46:27
2023-09-13T16:46:27
{ "title": [ 0.010612488, -0.013221741, -0.026046753, -0.03213501, -0.050445557, -0.01977539, -0.02708435, -0.007820129, -0.033050537, 0.03567505, -0.050994873, -0.003917694, 0.00017178059, -0.030883789, 0.00037002563, -0.03338623, -0.00066661835, ...
http://arxiv.org/abs/2309.07045v1
SafetyBench: Evaluating the Safety of Large Language Models with Multiple Choice Questions
Zhexin Zhang, Leqi Lei, Lindong Wu, Rui Sun, Yongkang Huang, Chong Long, Xiao Liu, Xuanyu Lei, Jie Tang, Minlie Huang
cs.CL
With the rapid development of Large Language Models (LLMs), increasing attention has been paid to their safety concerns. Consequently, evaluating the safety of LLMs has become an essential task for facilitating the broad applications of LLMs. Nevertheless, the absence of comprehensive safety evaluation benchmarks poses...
http://arxiv.org/pdf/2309.07045v1
2023-09-13T15:56:50
2023-09-13T15:56:50
{ "title": [ -0.020339966, -0.01159668, -0.03475952, -0.06451416, 0.013381958, 0.0013017654, -0.039611816, -0.0022583008, 0.00096940994, 0.012512207, -0.013938904, 0.020187378, -0.011756897, -0.026351929, -0.02508545, 0.0037059784, -0.019943237, ...
http://arxiv.org/abs/2309.07034v1
How (Not) to Use Sociodemographic Information for Subjective NLP Tasks
Tilman Beck, Hendrik Schuff, Anne Lauscher, Iryna Gurevych
cs.CL, cs.AI
Annotators' sociodemographic backgrounds (i.e., the individual compositions of their gender, age, educational background, etc.) have a strong impact on their decisions when working on subjective NLP tasks, such as hate speech detection. Often, heterogeneous backgrounds result in high disagreements. To model this variat...
http://arxiv.org/pdf/2309.07034v1
2023-09-13T15:42:06
2023-09-13T15:42:06
{ "title": [ -0.0063552856, -0.008972168, -0.015075684, -0.029022217, -0.01763916, -0.009124756, -0.0013370514, -0.020492554, 0.006046295, 0.0005378723, -0.04119873, 0.003829956, -0.02293396, -0.008087158, -0.035705566, -0.0061187744, -0.013748169, ...
http://arxiv.org/abs/2309.07020v1
Beyond original Research Articles Categorization via NLP
Rosanna Turrisi
cs.CL
This work proposes a novel approach to text categorization -- for unknown categories -- in the context of scientific literature, using Natural Language Processing techniques. The study leverages the power of pre-trained language models, specifically SciBERT, to extract meaningful representations of abstracts from the A...
http://arxiv.org/pdf/2309.07020v1
2023-09-13T15:23:30
2023-09-13T15:23:30
{ "title": [ -0.01927185, 0.0050621033, -0.032165527, -0.022232056, -0.0035972595, -0.017501831, -0.015991211, -0.033355713, -0.02381897, 0.018356323, -0.025497437, 0.021133423, -0.04663086, -0.011230469, -0.035095215, -0.0134887695, -0.015655518, ...
http://arxiv.org/abs/2309.07015v1
Résumé Parsing as Hierarchical Sequence Labeling: An Empirical Study
Federico Retyk, Hermenegildo Fabregat, Juan Aizpuru, Mariana Taglio, Rabih Zbib
cs.CL, cs.AI, cs.IR, I.2.7
Extracting information from r\'esum\'es is typically formulated as a two-stage problem, where the document is first segmented into sections and then each section is processed individually to extract the target entities. Instead, we cast the whole problem as sequence labeling in two levels -- lines and tokens -- and stu...
http://arxiv.org/pdf/2309.07015v1
2023-09-13T15:17:29
2023-09-13T15:17:29
{ "title": [ 0.0013952255, -0.0016469955, -0.02293396, -0.09246826, 0.0022392273, 0.0064430237, -0.04147339, 0.0021972656, -0.023223877, 0.031341553, -0.038330078, 0.026947021, 0.00071525574, 0.014419556, -0.05130005, -0.024383545, -0.038848877, ...
http://arxiv.org/abs/2309.07009v1
OYXOY: A Modern NLP Test Suite for Modern Greek
Konstantinos Kogkalidis, Stergios Chatzikyriakidis, Eirini Chrysovalantou Giannikouri, Vassiliki Katsouli, Christina Klironomou, Christina Koula, Dimitris Papadakis, Thelka Pasparaki, Erofili Psaltaki, Efthymia Sakellariou, Hara Soupiona
cs.CL
This paper serves as a foundational step towards the development of a linguistically motivated and technically relevant evaluation suite for Greek NLP. We initiate this endeavor by introducing four expert-verified evaluation tasks, specifically targeted at natural language inference, word sense disambiguation (through ...
http://arxiv.org/pdf/2309.07009v1
2023-09-13T15:00:56
2023-09-13T15:00:56
{ "title": [ -0.021896362, -0.024673462, -0.030090332, -0.04901123, -0.0017375946, -0.0039787292, -0.014785767, 0.013938904, -0.004016876, 0.02633667, -0.043701172, 0.024398804, 0.026397705, -0.017837524, -0.014282227, 0.0013990402, 0.0014162064, ...
http://arxiv.org/abs/2309.06991v1
Unsupervised Contrast-Consistent Ranking with Language Models
Niklas Stoehr, Pengxiang Cheng, Jing Wang, Daniel Preotiuc-Pietro, Rajarshi Bhowmik
cs.LG, cs.CL, stat.ML
Language models contain ranking-based knowledge and are powerful solvers of in-context ranking tasks. For instance, they may have parametric knowledge about the ordering of countries by size or may be able to rank reviews by sentiment. Recent work focuses on pairwise, pointwise, and listwise prompting techniques to eli...
http://arxiv.org/pdf/2309.06991v1
2023-09-13T14:36:26
2023-09-13T14:36:26
{ "title": [ 0.00819397, 0.021713257, -0.026489258, -0.039764404, 0.026611328, -0.0014238358, -0.039764404, -0.041778564, 0.01902771, 0.030380249, -0.029785156, -0.008018494, -0.00031924248, -0.005004883, -0.05343628, -0.005882263, -0.023986816, ...
http://arxiv.org/abs/2309.06989v1
Remote Inference of Cognitive Scores in ALS Patients Using a Picture Description
Carla Agurto, Guillermo Cecchi, Bo Wen, Ernest Fraenkel, James Berry, Indu Navar, Raquel Norel
cs.CL
Amyotrophic lateral sclerosis is a fatal disease that not only affects movement, speech, and breath but also cognition. Recent studies have focused on the use of language analysis techniques to detect ALS and infer scales for monitoring functional progression. In this paper, we focused on another important aspect, cogn...
http://arxiv.org/pdf/2309.06989v1
2023-09-13T14:30:30
2023-09-13T14:30:30
{ "title": [ 0.029144287, -0.0036888123, -0.008224487, -0.040039062, 0.01133728, -0.0076904297, 0.0040245056, -0.039886475, -0.017669678, 0.0027446747, -0.05758667, -0.016098022, -0.016815186, -0.0345459, -0.0103302, -0.055877686, 0.01058197, 0.0...
http://arxiv.org/abs/2309.06979v1
Auto-Regressive Next-Token Predictors are Universal Learners
Eran Malach
cs.LG, cs.CL
Large language models display remarkable capabilities in logical and mathematical reasoning, allowing them to solve complex tasks. Interestingly, these abilities emerge in networks trained on the simple task of next-token prediction. In this work, we present a theoretical framework for studying auto-regressive next-tok...
http://arxiv.org/pdf/2309.06979v1
2023-09-13T14:15:03
2023-09-13T14:15:03
{ "title": [ 0.012374878, 0.004459381, -0.018035889, -0.0692749, -0.0076446533, 0.008270264, -0.047424316, -0.03466797, -0.010375977, 0.021621704, -0.032440186, 0.02243042, -0.02684021, -0.017242432, -0.026901245, -0.0070343018, -0.008125305, 0.0...
http://arxiv.org/abs/2309.06928v1
Dynamic Causal Disentanglement Model for Dialogue Emotion Detection
Yuting Su, Yichen Wei, Weizhi Nie, Sicheng Zhao, Anan Liu
cs.CL, cs.CV
Emotion detection is a critical technology extensively employed in diverse fields. While the incorporation of commonsense knowledge has proven beneficial for existing emotion detection methods, dialogue-based emotion detection encounters numerous difficulties and challenges due to human agency and the variability of di...
http://arxiv.org/pdf/2309.06928v1
2023-09-13T12:58:09
2023-09-13T12:58:09
{ "title": [ -0.0013494492, -0.0050315857, -0.049804688, -0.05609131, 0.010017395, -0.018920898, -0.02079773, 0.006729126, 0.028701782, 0.014678955, -0.006252289, -0.02168274, 0.02420044, -0.045532227, -0.03253174, 0.00223732, -0.01777649, 0.0397...
http://arxiv.org/abs/2309.06923v1
Native Language Identification with Big Bird Embeddings
Sergey Kramp, Giovanni Cassani, Chris Emmery
cs.CL
Native Language Identification (NLI) intends to classify an author's native language based on their writing in another language. Historically, the task has heavily relied on time-consuming linguistic feature engineering, and transformer-based NLI models have thus far failed to offer effective, practical alternatives. T...
http://arxiv.org/pdf/2309.06923v1
2023-09-13T12:47:40
2023-09-13T12:47:40
{ "title": [ -0.0007405281, -0.007835388, -0.008094788, -0.048614502, -0.0095825195, 0.0031280518, -0.04067993, -0.014015198, 0.006866455, 0.016937256, -0.008651733, 0.004928589, -0.018814087, -0.029800415, -0.025756836, -0.00026392937, 0.009727478, ...
http://arxiv.org/abs/2309.06917v1
Continual Learning with Dirichlet Generative-based Rehearsal
Min Zeng, Wei Xue, Qifeng Liu, Yike Guo
cs.CL, cs.AI, cs.LG
Recent advancements in data-driven task-oriented dialogue systems (ToDs) struggle with incremental learning due to computational constraints and time-consuming issues. Continual Learning (CL) attempts to solve this by avoiding intensive pre-training, but it faces the problem of catastrophic forgetting (CF). While gener...
http://arxiv.org/pdf/2309.06917v1
2023-09-13T12:30:03
2023-09-13T12:30:03
{ "title": [ 0.013114929, 0.007381439, -0.034423828, -0.10870361, 0.013435364, -0.0063972473, -0.049072266, 0.0057411194, -0.004043579, 0.025756836, -0.034362793, 0.031234741, -0.037322998, -0.00566864, -0.034332275, -0.02166748, 0.0039863586, 0....
http://arxiv.org/abs/2309.06908v1
Towards the TopMost: A Topic Modeling System Toolkit
Xiaobao Wu, Fengjun Pan, Anh Tuan Luu
cs.CL, cs.AI, cs.IR, cs.LG
Topic models have been proposed for decades with various applications and recently refreshed by the neural variational inference. However, these topic models adopt totally distinct dataset, implementation, and evaluation settings, which hinders their quick utilization and fair comparisons. This greatly hinders the rese...
http://arxiv.org/pdf/2309.06908v1
2023-09-13T12:10:54
2023-09-13T12:10:54
{ "title": [ -0.002565384, -0.0057411194, -0.026046753, -0.058624268, 0.015205383, 0.012588501, -0.026397705, -0.03744507, 0.032073975, 0.023864746, 0.011878967, 0.019622803, -0.017578125, -0.024856567, -0.047302246, 0.03186035, -0.012641907, 0.0...
http://arxiv.org/abs/2309.06844v1
Gpachov at CheckThat! 2023: A Diverse Multi-Approach Ensemble for Subjectivity Detection in News Articles
Georgi Pachov, Dimitar Dimitrov, Ivan Koychev, Preslav Nakov
cs.CL, cs.AI, cs.MM
The wide-spread use of social networks has given rise to subjective, misleading, and even false information on the Internet. Thus, subjectivity detection can play an important role in ensuring the objectiveness and the quality of a piece of information. This paper presents the solution built by the Gpachov team for the...
http://arxiv.org/pdf/2309.06844v1
2023-09-13T09:49:20
2023-09-13T09:49:20
{ "title": [ 0.019744873, -0.005970001, -0.019836426, -0.0231781, 0.0053901672, -0.027420044, 0.006843567, -0.03845215, 0.014915466, 0.04220581, -0.028915405, 0.02973938, -0.00522995, 0.002450943, -0.058532715, 0.014976501, -0.027801514, 0.037963...
http://arxiv.org/abs/2309.06814v1
Comparative Analysis of Contextual Relation Extraction based on Deep Learning Models
R. Priyadharshini, G. Jeyakodi, P. Shanthi Bala
cs.CL, cs.AI, cs.LG
Contextual Relation Extraction (CRE) is mainly used for constructing a knowledge graph with a help of ontology. It performs various tasks such as semantic search, query answering, and textual entailment. Relation extraction identifies the entities from raw texts and the relations among them. An efficient and accurate C...
http://arxiv.org/pdf/2309.06814v1
2023-09-13T09:05:09
2023-09-13T09:05:09
{ "title": [ 0.010787964, 0.015594482, -0.06311035, -0.0791626, 0.03668213, -0.01335907, -0.040130615, -0.0027427673, 0.020523071, 0.03768921, -0.012809753, 0.0046653748, 0.010597229, -0.010749817, -0.052490234, 0.013946533, -0.016479492, 0.04138...
http://arxiv.org/abs/2309.06794v1
Cognitive Mirage: A Review of Hallucinations in Large Language Models
Hongbin Ye, Tong Liu, Aijia Zhang, Wei Hua, Weiqiang Jia
cs.CL, cs.AI, cs.LG
As large language models continue to develop in the field of AI, text generation systems are susceptible to a worrisome phenomenon known as hallucination. In this study, we summarize recent compelling insights into hallucinations in LLMs. We present a novel taxonomy of hallucinations from various text generation tasks,...
http://arxiv.org/pdf/2309.06794v1
2023-09-13T08:33:09
2023-09-13T08:33:09
{ "title": [ -0.012107849, 0.009262085, -0.024871826, -0.031188965, -0.011749268, -0.021987915, -0.030639648, -0.034820557, 0.00774765, 0.05657959, -0.0020618439, -0.000113487244, 0.03149414, -0.023483276, -0.066345215, 0.021362305, -0.015357971, ...
http://arxiv.org/abs/2309.06759v1
Scaled Prompt-Tuning for Few-Shot Natural Language Generation
Ting Hu, Christoph Meinel, Haojin Yang
cs.CL
The increasingly Large Language Models (LLMs) demonstrate stronger language understanding and generation capabilities, while the memory demand and computation cost of fine-tuning LLMs on downstream tasks are non-negligible. Besides, fine-tuning generally requires a certain amount of data from individual tasks whilst da...
http://arxiv.org/pdf/2309.06759v1
2023-09-13T07:12:31
2023-09-13T07:12:31
{ "title": [ -0.028808594, -0.04534912, -0.028427124, -0.05545044, -0.03994751, 0.013900757, -0.056732178, 0.030227661, -0.0037498474, 0.020385742, -0.032806396, 0.011657715, -0.0023765564, -0.011779785, -0.033050537, -0.009529114, -0.006767273, ...
http://arxiv.org/abs/2309.06748v1
CONVERSER: Few-Shot Conversational Dense Retrieval with Synthetic Data Generation
Chao-Wei Huang, Chen-Yu Hsu, Tsu-Yuan Hsu, Chen-An Li, Yun-Nung Chen
cs.CL, cs.IR
Conversational search provides a natural interface for information retrieval (IR). Recent approaches have demonstrated promising results in applying dense retrieval to conversational IR. However, training dense retrievers requires large amounts of in-domain paired data. This hinders the development of conversational de...
http://arxiv.org/pdf/2309.06748v1
2023-09-13T06:40:24
2023-09-13T06:40:24
{ "title": [ -0.0025596619, -0.018844604, -0.020477295, -0.051727295, -0.021972656, 0.017547607, -0.04437256, 0.0042419434, 0.04547119, 0.023483276, -0.03668213, 0.011360168, -0.013168335, 0.0027275085, -0.05899048, -0.0014591217, -0.007347107, -...
http://arxiv.org/abs/2309.06726v1
Enhancing Keyphrase Generation by BART Finetuning with Splitting and Shuffling
Bin Chen, Mizuho Iwaihara
cs.CL, cs.AI, cs.DL, cs.NE
Keyphrase generation is a task of identifying a set of phrases that best repre-sent the main topics or themes of a given text. Keyphrases are dividend int pre-sent and absent keyphrases. Recent approaches utilizing sequence-to-sequence models show effectiveness on absent keyphrase generation. However, the per-formance ...
http://arxiv.org/pdf/2309.06726v1
2023-09-13T05:02:11
2023-09-13T05:02:11
{ "title": [ -0.030197144, -0.0082092285, -0.02204895, -0.06262207, -0.014350891, 0.014846802, -0.05783081, 0.001282692, -0.008255005, 0.06591797, -0.010864258, 0.03933716, 0.017715454, -0.0031661987, -0.049072266, 0.028640747, -0.020614624, 0.04...
http://arxiv.org/abs/2309.06706v1
Simultaneous Machine Translation with Large Language Models
Minghan Wang, Jinming Zhao, Thuy-Trang Vu, Fatemeh Shiri, Ehsan Shareghi, Gholamreza Haffari
cs.CL
Large language models (LLM) have demonstrated their abilities to solve various natural language processing tasks through dialogue-based interactions. For instance, research indicates that LLMs can achieve competitive performance in offline machine translation tasks for high-resource languages. However, applying LLMs to...
http://arxiv.org/pdf/2309.06706v1
2023-09-13T04:06:47
2023-09-13T04:06:47
{ "title": [ -0.039794922, -0.011833191, -0.019363403, -0.072021484, 0.0015602112, 0.02255249, -0.05822754, 0.0064315796, -0.0025310516, 0.049102783, -0.0019187927, -0.0013608932, 0.016921997, -0.020446777, -0.0423584, 0.02079773, -0.0024261475, ...
http://arxiv.org/abs/2309.06703v1
VLSlice: Interactive Vision-and-Language Slice Discovery
Eric Slyman, Minsuk Kahng, Stefan Lee
cs.CV, cs.CL, cs.HC, cs.LG, I.4.10; I.2.7; J.4
Recent work in vision-and-language demonstrates that large-scale pretraining can learn generalizable models that are efficiently transferable to downstream tasks. While this may improve dataset-scale aggregate metrics, analyzing performance around hand-crafted subgroups targeting specific bias dimensions reveals system...
http://arxiv.org/pdf/2309.06703v1
2023-09-13T04:02:38
2023-09-13T04:02:38
{ "title": [ -0.01184082, -0.015945435, -0.042175293, -0.06384277, 0.0014047623, -0.00919342, -0.044708252, -0.008644104, -0.0064849854, 0.022644043, -0.019989014, 0.00047588348, 0.00034356117, -0.023956299, -0.03390503, 0.0046844482, 0.008522034, ...
http://arxiv.org/abs/2309.06698v1
Benchmarking Procedural Language Understanding for Low-Resource Languages: A Case Study on Turkish
Arda Uzunoğlu, Gözde Gül Şahin
cs.CL
Understanding procedural natural language (e.g., step-by-step instructions) is a crucial step to execution and planning. However, while there are ample corpora and downstream tasks available in English, the field lacks such resources for most languages. To address this gap, we conduct a case study on Turkish procedural...
http://arxiv.org/pdf/2309.06698v1
2023-09-13T03:42:28
2023-09-13T03:42:28
{ "title": [ -0.012504578, -0.018859863, -0.008041382, -0.08306885, -0.026763916, 0.018615723, -0.08569336, 0.014854431, -0.009811401, 0.018218994, -0.0048065186, 0.0073432922, -0.01902771, -0.020889282, -0.04748535, -0.008850098, -0.008934021, 0...
http://arxiv.org/abs/2309.08631v1
Large Language Models Can Infer Psychological Dispositions of Social Media Users
Heinrich Peters, Sandra Matz
cs.CL, cs.AI, cs.CY, cs.HC, cs.LG, cs.SI
As Large Language Models (LLMs) demonstrate increasingly human-like abilities in various natural language processing (NLP) tasks that are bound to become integral to personalized technologies, understanding their capabilities and inherent biases is crucial. Our study investigates the potential of LLMs like ChatGPT to i...
http://arxiv.org/pdf/2309.08631v1
2023-09-13T01:27:48
2023-09-13T01:27:48
{ "title": [ 0.040740967, 0.01826477, -0.016601562, -0.029754639, -0.024658203, -0.034301758, -0.0005249977, -0.014930725, -0.023742676, 0.026519775, -0.03540039, -0.0017709732, 0.021942139, -0.044281006, 0.018554688, -0.013137817, 0.015991211, 0...
http://arxiv.org/abs/2309.06553v3
Query-Dependent Prompt Evaluation and Optimization with Offline Inverse RL
Hao Sun, Alihan Hüyük, Mihaela van der Schaar
cs.CL, cs.AI, cs.LG
In this study, we aim to enhance the arithmetic reasoning ability of Large Language Models (LLMs) through zero-shot prompt optimization. We identify a previously overlooked objective of query dependency in such optimization and elucidate two ensuing challenges that impede the successful and economical design of prompt ...
http://arxiv.org/pdf/2309.06553v3
2023-10-21T16:41:23
2023-09-13T01:12:52
{ "title": [ -0.025863647, -0.0104904175, -0.008430481, -0.0501709, -0.019104004, 0.016586304, -0.03668213, -0.015045166, -0.017288208, -0.0012521744, -0.0024986267, -0.030197144, -0.02041626, -0.047302246, 0.001004219, -0.029266357, -0.009147644, ...
http://arxiv.org/abs/2309.06657v1
Statistical Rejection Sampling Improves Preference Optimization
Tianqi Liu, Yao Zhao, Rishabh Joshi, Misha Khalman, Mohammad Saleh, Peter J. Liu, Jialu Liu
cs.CL
Improving the alignment of language models with human preferences remains an active research challenge. Previous approaches have primarily utilized Reinforcement Learning from Human Feedback (RLHF) via online RL methods such as Proximal Policy Optimization (PPO). Recently, offline methods such as Sequence Likelihood Ca...
http://arxiv.org/pdf/2309.06657v1
2023-09-13T01:07:25
2023-09-13T01:07:25
{ "title": [ 0.018081665, 0.003566742, 0.0001885891, -0.016738892, -0.0022640228, -0.027786255, 0.018295288, -0.04095459, -0.004573822, 0.013206482, -0.020309448, 0.0014820099, -0.021911621, -0.032440186, -0.024963379, 0.000749588, 0.0040626526, ...
http://arxiv.org/abs/2309.06619v1
RT-LM: Uncertainty-Aware Resource Management for Real-Time Inference of Language Models
Yufei Li, Zexin Li, Wei Yang, Cong Liu
cs.LG, cs.CL, cs.SY, eess.SY
Recent advancements in language models (LMs) have gained substantial attentions on their capability to generate human-like responses. Though exhibiting a promising future for various applications such as conversation AI, these LMs face deployment challenges on various devices due to their extreme computational cost and...
http://arxiv.org/pdf/2309.06619v1
2023-09-12T22:22:10
2023-09-12T22:22:10
{ "title": [ -0.051818848, -0.025421143, -0.044036865, -0.08557129, -0.02456665, 0.015174866, -0.10040283, 0.013771057, 0.00642395, 0.02130127, -0.023345947, 0.00333786, 0.0098724365, -0.022277832, -0.030090332, 0.0067749023, -0.0012578964, 0.031...
http://arxiv.org/abs/2309.06589v1
Do Generative Large Language Models need billions of parameters?
Sia Gholami, Marwan Omar
cs.CL, cs.AI
This paper presents novel systems and methodologies for the development of efficient large language models (LLMs). It explores the trade-offs between model size, performance, and computational resources, with the aim of maximizing the efficiency of these AI systems. The research explores novel methods that allow differ...
http://arxiv.org/pdf/2309.06589v1
2023-09-12T20:25:22
2023-09-12T20:25:22
{ "title": [ -0.015777588, -0.0121154785, -0.034332275, -0.07366943, -0.022583008, -0.0028209686, -0.057159424, 0.014564514, -0.014450073, 0.020706177, -0.03555298, -0.0022449493, -0.0040664673, -0.016311646, -0.015716553, -0.0135650635, 0.0049972534...
http://arxiv.org/abs/2309.06541v1
Text Encoders Lack Knowledge: Leveraging Generative LLMs for Domain-Specific Semantic Textual Similarity
Joseph Gatto, Omar Sharif, Parker Seegmiller, Philip Bohlman, Sarah Masud Preum
cs.CL
Amidst the sharp rise in the evaluation of large language models (LLMs) on various tasks, we find that semantic textual similarity (STS) has been under-explored. In this study, we show that STS can be cast as a text generation problem while maintaining strong performance on multiple STS benchmarks. Additionally, we sho...
http://arxiv.org/pdf/2309.06541v1
2023-09-12T19:32:45
2023-09-12T19:32:45
{ "title": [ -0.049438477, -0.002286911, -0.039123535, -0.03475952, -0.018051147, -0.0005540848, -0.054870605, -0.010040283, 0.0181427, 0.045288086, -0.02255249, 0.053588867, 0.014839172, -0.0027751923, -0.03451538, 0.026046753, -0.00970459, 0.03...
http://arxiv.org/abs/2309.15827v1
How We Define Harm Impacts Data Annotations: Explaining How Annotators Distinguish Hateful, Offensive, and Toxic Comments
Angela Schöpke-Gonzalez, Siqi Wu, Sagar Kumar, Paul J. Resnick, Libby Hemphill
cs.CL, cs.CY, K.4.1
Computational social science research has made advances in machine learning and natural language processing that support content moderators in detecting harmful content. These advances often rely on training datasets annotated by crowdworkers for harmful content. In designing instructions for annotation tasks to genera...
http://arxiv.org/pdf/2309.15827v1
2023-09-12T19:23:40
2023-09-12T19:23:40
{ "title": [ 0.026123047, 0.025680542, -0.03543091, -0.04333496, 0.016189575, -0.024429321, -0.037384033, -0.020187378, -0.01739502, 0.030227661, -0.024902344, 0.00033855438, 0.00025486946, -0.022598267, -0.0113220215, -0.03012085, -0.00015461445, ...
http://arxiv.org/abs/2309.06520v2
Minimum Bayes' Risk Decoding for System Combination of Grammatical Error Correction Systems
Vyas Raina, Mark Gales
cs.CL, cs.AI
For sequence-to-sequence tasks it is challenging to combine individual system outputs. Further, there is also often a mismatch between the decoding criterion and the one used for assessment. Minimum Bayes' Risk (MBR) decoding can be used to combine system outputs in a manner that encourages better alignment with the fi...
http://arxiv.org/pdf/2309.06520v2
2023-10-27T14:42:29
2023-09-12T18:51:10
{ "title": [ -0.006580353, -0.007205963, -0.04284668, -0.03640747, -0.023956299, -0.020751953, -0.034423828, 0.0012207031, -0.0018663406, 0.060180664, -0.049194336, 0.012763977, 0.024765015, -0.0063934326, -0.03930664, -0.005130768, -0.01675415, ...
http://arxiv.org/abs/2309.06517v1
Overview of Memotion 3: Sentiment and Emotion Analysis of Codemixed Hinglish Memes
Shreyash Mishra, S Suryavardan, Megha Chakraborty, Parth Patwa, Anku Rani, Aman Chadha, Aishwarya Reganti, Amitava Das, Amit Sheth, Manoj Chinnakotla, Asif Ekbal, Srijan Kumar
cs.CL
Analyzing memes on the internet has emerged as a crucial endeavor due to the impact this multi-modal form of content wields in shaping online discourse. Memes have become a powerful tool for expressing emotions and sentiments, possibly even spreading hate and misinformation, through humor and sarcasm. In this paper, we...
http://arxiv.org/pdf/2309.06517v1
2023-09-12T18:47:29
2023-09-12T18:47:29
{ "title": [ 0.02998352, -0.0063972473, 0.006095886, -0.015235901, -0.03253174, -0.0051841736, -0.020690918, -0.0154418945, 0.03387451, 0.0793457, -0.026931763, 0.0058670044, 0.066711426, -0.012290955, -0.033599854, 0.017425537, 0.0062675476, 0.0...
http://arxiv.org/abs/2309.06503v1
Leveraging Large Language Models and Weak Supervision for Social Media data annotation: an evaluation using COVID-19 self-reported vaccination tweets
Ramya Tekumalla, Juan M. Banda
cs.CL, cs.LG, cs.SI
The COVID-19 pandemic has presented significant challenges to the healthcare industry and society as a whole. With the rapid development of COVID-19 vaccines, social media platforms have become a popular medium for discussions on vaccine-related topics. Identifying vaccine-related tweets and analyzing them can provide ...
http://arxiv.org/pdf/2309.06503v1
2023-09-12T18:18:23
2023-09-12T18:18:23
{ "title": [ 0.034698486, 0.0045814514, -0.022720337, -0.082336426, 0.0022296906, -0.0075569153, -0.042907715, 0.01058197, -0.020431519, 0.028747559, -0.026031494, 0.03378296, -0.0103302, -0.023040771, 0.004421234, -0.003955841, -0.0048065186, 0....
http://arxiv.org/abs/2309.06490v1
Leveraging Large Language Models for Automated Dialogue Analysis
Sarah E. Finch, Ellie S. Paek, Jinho D. Choi
cs.CL
Developing high-performing dialogue systems benefits from the automatic identification of undesirable behaviors in system responses. However, detecting such behaviors remains challenging, as it draws on a breadth of general knowledge and understanding of conversational practices. Although recent research has focused on...
http://arxiv.org/pdf/2309.06490v1
2023-09-12T18:03:55
2023-09-12T18:03:55
{ "title": [ -0.021392822, -0.00818634, -0.041503906, -0.08428955, 0.00605011, 0.00919342, -0.0770874, 0.01411438, 0.010025024, 0.024993896, -0.0031089783, -0.012306213, -0.0017852783, -0.015220642, -0.058410645, 0.006477356, -0.010246277, 0.0404...
http://arxiv.org/abs/2309.06460v1
Widely Interpretable Semantic Representation: Frameless Meaning Representation for Broader Applicability
Lydia Feng, Gregor Williamson, Han He, Jinho D. Choi
cs.CL
This paper presents a novel semantic representation, WISeR, that overcomes challenges for Abstract Meaning Representation (AMR). Despite its strengths, AMR is not easily applied to languages or domains without predefined semantic frames, and its use of numbered arguments results in semantic role labels, which are not d...
http://arxiv.org/pdf/2309.06460v1
2023-09-12T17:44:40
2023-09-12T17:44:40
{ "title": [ -0.009422302, -0.014762878, -0.043273926, -0.066101074, 0.018569946, 0.0033931732, -0.034362793, -0.004951477, 0.019424438, 0.038726807, -0.00957489, -0.0015850067, 0.0064468384, -0.010757446, -0.05050659, 0.017822266, 0.0035324097, ...
http://arxiv.org/abs/2309.07172v1
Exploring Large Language Models for Ontology Alignment
Yuan He, Jiaoyan Chen, Hang Dong, Ian Horrocks
cs.AI, cs.CL, cs.LG
This work investigates the applicability of recent generative Large Language Models (LLMs), such as the GPT series and Flan-T5, to ontology alignment for identifying concept equivalence mappings across ontologies. To test the zero-shot performance of Flan-T5-XXL and GPT-3.5-turbo, we leverage challenging subsets from t...
http://arxiv.org/pdf/2309.07172v1
2023-09-12T17:01:02
2023-09-12T17:01:02
{ "title": [ -0.03652954, -0.008399963, -0.06567383, -0.076538086, 0.012184143, -0.0003786087, -0.06311035, 0.013298035, 0.01838684, 0.033294678, -0.0055618286, -0.022857666, 0.01852417, -0.017318726, -0.04006958, 0.010398865, -0.014442444, 0.035...
http://arxiv.org/abs/2309.08628v2
Recovering from Privacy-Preserving Masking with Large Language Models
Arpita Vats, Zhe Liu, Peng Su, Debjyoti Paul, Yingyi Ma, Yutong Pang, Zeeshan Ahmed, Ozlem Kalinli
cs.CL, cs.CR, cs.LG
Model adaptation is crucial to handle the discrepancy between proxy training data and actual users data received. To effectively perform adaptation, textual data of users is typically stored on servers or their local devices, where downstream natural language processing (NLP) models can be directly trained using such i...
http://arxiv.org/pdf/2309.08628v2
2023-09-23T00:20:05
2023-09-12T16:39:41
{ "title": [ -0.043792725, -0.004310608, -0.06896973, -0.06542969, 0.012550354, -0.002439499, -0.06173706, 0.022094727, -0.005176544, 0.023910522, -0.012542725, 0.006072998, 0.010971069, -0.023529053, -0.024597168, 0.014259338, -0.007369995, 0.03...
http://arxiv.org/abs/2309.06365v1
Cited Text Spans for Citation Text Generation
Xiangci Li, Yi-Hui Lee, Jessica Ouyang
cs.CL
Automatic related work generation must ground their outputs to the content of the cited papers to avoid non-factual hallucinations, but due to the length of scientific documents, existing abstractive approaches have conditioned only on the cited paper \textit{abstracts}. We demonstrate that the abstract is not always t...
http://arxiv.org/pdf/2309.06365v1
2023-09-12T16:28:36
2023-09-12T16:28:36
{ "title": [ -0.017028809, -0.0049552917, -0.033111572, -0.017791748, -0.03302002, -0.01777649, -0.032287598, -0.012199402, -0.03213501, 0.052856445, -0.039367676, 0.0023441315, 0.039978027, -0.020843506, -0.012664795, -0.010902405, -0.0070724487, ...
http://arxiv.org/abs/2309.06363v1
Learning to Predict Concept Ordering for Common Sense Generation
Tianhui Zhang, Danushka Bollegala, Bei Peng
cs.CL
Prior work has shown that the ordering in which concepts are shown to a commonsense generator plays an important role, affecting the quality of the generated sentence. However, it remains a challenge to determine the optimal ordering of a given set of concepts such that a natural sentence covering all the concepts coul...
http://arxiv.org/pdf/2309.06363v1
2023-09-12T16:27:18
2023-09-12T16:27:18
{ "title": [ -0.004348755, -0.013587952, -0.04647827, -0.06402588, -0.0024471283, 0.0021820068, -0.048828125, -0.010498047, 0.009544373, 0.03363037, -0.010848999, 0.012878418, -0.01789856, -0.00749588, -0.047821045, -0.00415802, -0.0056877136, 0....
http://arxiv.org/abs/2309.06275v1
Re-Reading Improves Reasoning in Language Models
Xiaohan Xu, Chongyang Tao, Tao Shen, Can Xu, Hongbo Xu, Guodong Long, Jian-guang Lou
cs.CL
Reasoning presents a significant and challenging issue for Large Language Models (LLMs). The predominant focus of research has revolved around developing diverse prompting strategies to guide and structure the reasoning processes of LLMs. However, these approaches based on decoder-only causal language models often oper...
http://arxiv.org/pdf/2309.06275v1
2023-09-12T14:36:23
2023-09-12T14:36:23
{ "title": [ -0.015670776, -0.0206604, -0.01348114, -0.05456543, -0.07232666, -0.0049209595, -0.10461426, -0.0061416626, 0.012016296, 0.045013428, -0.027297974, 0.003320694, 0.039001465, -0.016540527, -0.034576416, 0.025024414, 0.002363205, 0.020...
http://arxiv.org/abs/2309.06236v1
The first step is the hardest: Pitfalls of Representing and Tokenizing Temporal Data for Large Language Models
Dimitris Spathis, Fahim Kawsar
cs.LG, cs.CL
Large Language Models (LLMs) have demonstrated remarkable generalization across diverse tasks, leading individuals to increasingly use them as personal assistants and universal computing engines. Nevertheless, a notable obstacle emerges when feeding numerical/temporal data into these models, such as data sourced from w...
http://arxiv.org/pdf/2309.06236v1
2023-09-12T13:51:29
2023-09-12T13:51:29
{ "title": [ -0.011543274, -0.019729614, -0.047576904, -0.06896973, -0.018478394, -0.020431519, -0.044769287, 0.019805908, 0.004776001, 0.04324341, -0.054992676, 0.009643555, 0.009025574, -0.00082588196, -0.05404663, -0.011177063, -0.0041389465, ...
http://arxiv.org/abs/2309.06219v2
Human Action Co-occurrence in Lifestyle Vlogs using Graph Link Prediction
Oana Ignat, Santiago Castro, Weiji Li, Rada Mihalcea
cs.CV, cs.CL, cs.CY, cs.IR
We introduce the task of automatic human action co-occurrence identification, i.e., determine whether two human actions can co-occur in the same interval of time. We create and make publicly available the ACE (Action Co-occurrencE) dataset, consisting of a large graph of ~12k co-occurring pairs of visual actions and th...
http://arxiv.org/pdf/2309.06219v2
2023-09-22T18:27:57
2023-09-12T13:38:44
{ "title": [ 0.007144928, 0.011260986, -0.062561035, -0.06854248, 0.009239197, -0.034576416, -0.033325195, 0.018325806, -0.04849243, 0.046142578, 0.009086609, 0.024780273, 0.007434845, -0.014549255, -0.018295288, -0.011703491, -0.009925842, 0.054...
http://arxiv.org/abs/2309.08627v1
Evaluating Dynamic Topic Models
Charu James, Mayank Nagda, Nooshin Haji Ghassemi, Marius Kloft, Sophie Fellenz
cs.CL, cs.IR, cs.LG
There is a lack of quantitative measures to evaluate the progression of topics through time in dynamic topic models (DTMs). Filling this gap, we propose a novel evaluation measure for DTMs that analyzes the changes in the quality of each topic over time. Additionally, we propose an extension combining topic quality wit...
http://arxiv.org/pdf/2309.08627v1
2023-09-12T13:30:25
2023-09-12T13:30:25
{ "title": [ -0.0063095093, 0.009414673, -0.029373169, -0.03677368, 0.008407593, -0.004600525, -0.009689331, -0.038116455, 0.0060272217, 0.028717041, -0.012893677, 0.022491455, -0.036743164, -0.012138367, -0.038848877, 0.015792847, -0.0036582947, ...
http://arxiv.org/abs/2309.06192v2
Improving and Evaluating the Detection of Fragmentation in News Recommendations with the Clustering of News Story Chains
Alessandra Polimeno, Myrthe Reuver, Sanne Vrijenhoek, Antske Fokkens
cs.CL, cs.CY, cs.IR
News recommender systems play an increasingly influential role in shaping information access within democratic societies. However, tailoring recommendations to users' specific interests can result in the divergence of information streams. Fragmented access to information poses challenges to the integrity of the public ...
http://arxiv.org/pdf/2309.06192v2
2023-09-18T13:05:08
2023-09-12T13:01:20
{ "title": [ -0.013412476, -0.006565094, -0.05307007, -0.04522705, 0.0051651, -0.007083893, -0.012908936, 0.00029444695, -0.030151367, 0.041534424, -0.01940918, 0.04333496, 0.006526947, -0.0008687973, -0.024795532, -0.004219055, -0.003850937, 0.0...
http://arxiv.org/abs/2309.06179v1
Glancing Future for Simultaneous Machine Translation
Shoutao Guo, Shaolei Zhang, Yang Feng
cs.CL, cs.AI
Simultaneous machine translation (SiMT) outputs translation while reading the source sentence. Unlike conventional sequence-to-sequence (seq2seq) training, existing SiMT methods adopt the prefix-to-prefix (prefix2prefix) training, where the model predicts target tokens based on partial source tokens. However, the prefi...
http://arxiv.org/pdf/2309.06179v1
2023-09-12T12:46:20
2023-09-12T12:46:20
{ "title": [ -0.023757935, -0.023208618, 0.00027537346, -0.07989502, -0.027252197, 0.013893127, -0.038269043, 0.025680542, -0.007686615, 0.07525635, -0.005897522, -0.0038223267, 0.0362854, -0.014640808, -0.06048584, 0.03137207, 0.004798889, 0.024...
http://arxiv.org/abs/2309.06175v2
AKEM: Aligning Knowledge Base to Queries with Ensemble Model for Entity Recognition and Linking
Di Lu, Zhongping Liang, Caixia Yuan, Xiaojie Wang
cs.CL, cs.IR
This paper presents a novel approach to address the Entity Recognition and Linking Challenge at NLPCC 2015. The task involves extracting named entity mentions from short search queries and linking them to entities within a reference Chinese knowledge base. To tackle this problem, we first expand the existing knowledge ...
http://arxiv.org/pdf/2309.06175v2
2023-09-13T03:53:43
2023-09-12T12:37:37
{ "title": [ -0.016738892, 0.0044136047, -0.077819824, -0.07098389, 0.032806396, -0.0047187805, -0.03475952, 0.0057640076, 0.021896362, 0.02645874, -0.031021118, 0.004219055, 0.020568848, -0.00774765, -0.040863037, 0.0039978027, -0.0110321045, 0....
http://arxiv.org/abs/2309.06163v1
Overview of GUA-SPA at IberLEF 2023: Guarani-Spanish Code Switching Analysis
Luis Chiruzzo, Marvin Agüero-Torales, Gustavo Giménez-Lugo, Aldo Alvarez, Yliana Rodríguez, Santiago Góngora, Thamar Solorio
cs.CL
We present the first shared task for detecting and analyzing code-switching in Guarani and Spanish, GUA-SPA at IberLEF 2023. The challenge consisted of three tasks: identifying the language of a token, NER, and a novel task of classifying the way a Spanish span is used in the code-switched context. We annotated a corpu...
http://arxiv.org/pdf/2309.06163v1
2023-09-12T12:18:18
2023-09-12T12:18:18
{ "title": [ -0.043060303, 0.025924683, -0.050354004, 0.045074463, -0.012535095, -0.02456665, -0.032226562, -0.018096924, -0.018112183, 0.05014038, 0.013191223, -0.026290894, 0.06585693, -0.0463562, 0.016586304, -0.001964569, 0.0066986084, 0.0217...
http://arxiv.org/abs/2309.06135v1
Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts
Zhi-Yi Chin, Chieh-Ming Jiang, Ching-Chun Huang, Pin-Yu Chen, Wei-Chen Chiu
cs.CL, cs.CV
Text-to-image diffusion models, e.g. Stable Diffusion (SD), lately have shown remarkable ability in high-quality content generation, and become one of the representatives for the recent wave of transformative AI. Nevertheless, such advance comes with an intensifying concern about the misuse of this generative technolog...
http://arxiv.org/pdf/2309.06135v1
2023-09-12T11:19:36
2023-09-12T11:19:36
{ "title": [ 0.018234253, 0.008140564, 0.004699707, -0.006427765, -0.009239197, 0.0009589195, -0.01322937, -0.057922363, -0.0010967255, 0.010543823, -0.028930664, 0.021224976, -0.023910522, -0.031677246, -0.017242432, -0.010894775, -0.007232666, ...
http://arxiv.org/abs/2309.06132v2
Measuring vagueness and subjectivity in texts: from symbolic to neural VAGO
Benjamin Icard, Vincent Claveau, Ghislain Atemezing, Paul Égré
cs.CL, cs.AI, 68T07, 68T50
We present a hybrid approach to the automated measurement of vagueness and subjectivity in texts. We first introduce the expert system VAGO, we illustrate it on a small benchmark of fact vs. opinion sentences, and then test it on the larger French press corpus FreSaDa to confirm the higher prevalence of subjective mark...
http://arxiv.org/pdf/2309.06132v2
2023-10-23T18:16:19
2023-09-12T11:18:29
{ "title": [ -0.028945923, 0.0056419373, -0.018096924, -0.00623703, -0.009376526, -0.009773254, 0.0037498474, -0.039794922, 0.010925293, 0.034240723, -0.014137268, 0.0063056946, -0.013511658, -0.01361084, -0.07165527, -0.0020942688, -0.025177002, ...
http://arxiv.org/abs/2309.06131v1
Annotating Data for Fine-Tuning a Neural Ranker? Current Active Learning Strategies are not Better than Random Selection
Sophia Althammer, Guido Zuccon, Sebastian Hofstätter, Suzan Verberne, Allan Hanbury
cs.IR, cs.CL
Search methods based on Pretrained Language Models (PLM) have demonstrated great effectiveness gains compared to statistical and early neural ranking models. However, fine-tuning PLM-based rankers requires a great amount of annotated training data. Annotating data involves a large manual effort and thus is expensive, e...
http://arxiv.org/pdf/2309.06131v1
2023-09-12T11:17:42
2023-09-12T11:17:42
{ "title": [ 0.021896362, 0.010612488, 0.014549255, 0.018798828, -0.0018320084, -0.034576416, 0.053619385, -0.035308838, -0.0059127808, 0.02809143, 0.0067214966, 0.009208679, -0.055023193, -0.044921875, -0.011352539, -0.016494751, 0.02279663, 0.0...
http://arxiv.org/abs/2310.05960v1
Fingerprint Attack: Client De-Anonymization in Federated Learning
Qiongkai Xu, Trevor Cohn, Olga Ohrimenko
cs.CR, cs.AI, cs.CL, cs.LG
Federated Learning allows collaborative training without data sharing in settings where participants do not trust the central server and one another. Privacy can be further improved by ensuring that communication between the participants and the server is anonymized through a shuffle; decoupling the participant identit...
http://arxiv.org/pdf/2310.05960v1
2023-09-12T11:10:30
2023-09-12T11:10:30
{ "title": [ -0.00017023087, 0.042236328, -0.06616211, -0.051574707, 0.043182373, -0.027069092, 0.008651733, -0.025146484, -0.037963867, 0.030792236, 0.003944397, 0.002714157, -0.001821518, -0.032409668, -0.02267456, -0.012466431, -0.008171082, -...
http://arxiv.org/abs/2309.06126v1
AstroLLaMA: Towards Specialized Foundation Models in Astronomy
Tuan Dung Nguyen, Yuan-Sen Ting, Ioana Ciucă, Charlie O'Neill, Ze-Chang Sun, Maja Jabłońska, Sandor Kruk, Ernest Perkowski, Jack Miller, Jason Li, Josh Peek, Kartheik Iyer, Tomasz Różański, Pranav Khetarpal, Sharaf Zaman, David Brodrick, Sergio J. Rodríguez Méndez, Thang Bui, Alyssa Goodman, Alberto Accomazzi, Jill Nai...
astro-ph.IM, astro-ph.CO, astro-ph.GA, astro-ph.HE, cs.CL, cs.LG
Large language models excel in many human-language tasks but often falter in highly specialized domains like scholarly astronomy. To bridge this gap, we introduce AstroLLaMA, a 7-billion-parameter model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv. Optimized for traditional causal language ...
http://arxiv.org/pdf/2309.06126v1
2023-09-12T11:02:27
2023-09-12T11:02:27
{ "title": [ -0.03677368, -0.0035095215, -0.025161743, -0.022140503, -0.011024475, -0.0012264252, -0.021499634, -0.034423828, 0.003112793, 0.031799316, -0.0025615692, 0.0070266724, -0.004398346, -0.025482178, -0.040130615, 0.013160706, 0.022994995, ...
http://arxiv.org/abs/2309.06112v1
Characterizing Latent Perspectives of Media Houses Towards Public Figures
Sharath Srivatsa, Srinath Srinivasa
cs.CL, cs.IR
Media houses reporting on public figures, often come with their own biases stemming from their respective worldviews. A characterization of these underlying patterns helps us in better understanding and interpreting news stories. For this, we need diverse or subjective summarizations, which may not be amenable for clas...
http://arxiv.org/pdf/2309.06112v1
2023-09-12T10:27:39
2023-09-12T10:27:39
{ "title": [ 0.016082764, -0.0020179749, -0.0287323, -0.02758789, -0.009132385, -0.036621094, 0.025115967, -0.008934021, -0.024429321, 0.03262329, -0.048553467, 0.017425537, -0.013061523, -0.004459381, -0.030685425, -0.026153564, -0.014091492, 0....
http://arxiv.org/abs/2309.06105v1
Towards Visual Taxonomy Expansion
Tinghui Zhu, Jingping Liu, Jiaqing Liang, Haiyun Jiang, Yanghua Xiao, Zongyu Wang, Rui Xie, Yunsen Xian
cs.CV, cs.CL
Taxonomy expansion task is essential in organizing the ever-increasing volume of new concepts into existing taxonomies. Most existing methods focus exclusively on using textual semantics, leading to an inability to generalize to unseen terms and the "Prototypical Hypernym Problem." In this paper, we propose Visual Taxo...
http://arxiv.org/pdf/2309.06105v1
2023-09-12T10:17:28
2023-09-12T10:17:28
{ "title": [ 0.028579712, 0.007396698, -0.032226562, -0.05105591, 0.023986816, -0.021987915, -0.0019226074, -0.028656006, 0.014450073, 0.005962372, -0.048034668, 0.006011963, -0.044158936, -0.016723633, -0.03491211, -0.0113220215, 0.02243042, 0.0...
http://arxiv.org/abs/2309.06089v1
Measuring Catastrophic Forgetting in Cross-Lingual Transfer Paradigms: Exploring Tuning Strategies
Boshko Koloski, Blaž Škrlj, Marko Robnik-Šikonja, Senja Pollak
cs.CL, cs.LG
The cross-lingual transfer is a promising technique to solve tasks in less-resourced languages. In this empirical study, we compare two fine-tuning approaches combined with zero-shot and full-shot learning approaches for large language models in a cross-lingual setting. As fine-tuning strategies, we compare parameter-e...
http://arxiv.org/pdf/2309.06089v1
2023-09-12T09:37:08
2023-09-12T09:37:08
{ "title": [ -0.020904541, 0.0076141357, -0.036010742, -0.047729492, -0.002231598, -0.004283905, -0.028762817, -0.00089645386, -0.006980896, 0.048706055, -0.008758545, 0.035064697, 0.044891357, -0.023025513, -0.02041626, 0.021621704, -0.0017280579, ...
http://arxiv.org/abs/2309.06085v2
BHASA: A Holistic Southeast Asian Linguistic and Cultural Evaluation Suite for Large Language Models
Wei Qi Leong, Jian Gang Ngui, Yosephine Susanto, Hamsawardhini Rengarajan, Kengatharaiyer Sarveswaran, William Chandra Tjhi
cs.CL
The rapid development of Large Language Models (LLMs) and the emergence of novel abilities with scale have necessitated the construction of holistic, diverse and challenging benchmarks such as HELM and BIG-bench. However, at the moment, most of these benchmarks focus only on performance in English and evaluations that ...
http://arxiv.org/pdf/2309.06085v2
2023-09-19T03:44:17
2023-09-12T09:31:25
{ "title": [ -0.022033691, -0.008712769, -0.028625488, -0.0619812, 0.008628845, -0.0059814453, -0.04675293, 0.005596161, -0.010375977, 0.013053894, -0.017562866, 0.009963989, -0.016647339, -0.021957397, -0.035125732, -0.013145447, -0.02696228, 0....
http://arxiv.org/abs/2309.06057v1
RAP-Gen: Retrieval-Augmented Patch Generation with CodeT5 for Automatic Program Repair
Weishi Wang, Yue Wang, Shafiq Joty, Steven C. H. Hoi
cs.SE, cs.CL
Automatic program repair (APR) is crucial to reduce manual debugging efforts for developers and improve software reliability. While conventional search-based techniques typically rely on heuristic rules or a redundancy assumption to mine fix patterns, recent years have witnessed the surge of deep learning (DL) based ap...
http://arxiv.org/pdf/2309.06057v1
2023-09-12T08:52:56
2023-09-12T08:52:56
{ "title": [ -0.016784668, -0.019241333, -0.023880005, -0.097961426, -0.0031337738, 0.013267517, -0.072509766, 0.028335571, -0.016525269, 0.031311035, -0.022399902, 0.028381348, -0.008979797, -0.016860962, -0.015106201, 0.008171082, -0.014892578, ...
http://arxiv.org/abs/2309.06054v1
How does representation impact in-context learning: A exploration on a synthetic task
Jingwen Fu, Tao Yang, Yuwang Wang, Yan Lu, Nanning Zheng
cs.LG, cs.CL, cs.CV
In-context learning, i.e., learning from in-context samples, is an impressive ability of Transformer. However, the mechanism driving the in-context learning is not yet fully understood. In this study, we aim to investigate from an underexplored perspective of representation learning. The representation is more complex ...
http://arxiv.org/pdf/2309.06054v1
2023-09-12T08:45:25
2023-09-12T08:45:25
{ "title": [ 0.023544312, 0.008087158, -0.033996582, -0.04171753, 0.016357422, -0.035583496, 0.022979736, -0.0134887695, -0.0068359375, 0.026153564, -0.025634766, 0.0071640015, -0.017303467, -0.018005371, -0.031921387, -0.021606445, 0.000019133091, ...
http://arxiv.org/abs/2309.06453v1
Narrowing the Gap between Supervised and Unsupervised Sentence Representation Learning with Large Language Model
Mingxin Li, Richong Zhang, Zhijie Nie, Yongyi Mao
cs.CL, cs.LG
Sentence Representation Learning (SRL) is a fundamental task in Natural Language Processing (NLP), with Contrastive learning of Sentence Embeddings (CSE) as the mainstream technique due to its superior performance. An intriguing phenomenon in CSE is the significant performance gap between supervised and unsupervised me...
http://arxiv.org/pdf/2309.06453v1
2023-09-12T08:16:58
2023-09-12T08:16:58
{ "title": [ -0.015380859, 0.0043411255, -0.03451538, -0.05380249, 0.006679535, -0.0038375854, -0.048675537, -0.0051879883, 0.003435135, 0.021392822, -0.025482178, 0.0064888, -0.000344038, -0.0132751465, -0.02268982, 0.00085163116, -0.009735107, ...
http://arxiv.org/abs/2309.06009v1
Content Reduction, Surprisal and Information Density Estimation for Long Documents
Shaoxiong Ji, Wei Sun, Pekka Marttinen
cs.CL
Many computational linguistic methods have been proposed to study the information content of languages. We consider two interesting research questions: 1) how is information distributed over long documents, and 2) how does content reduction, such as token selection and text summarization, affect the information density...
http://arxiv.org/pdf/2309.06009v1
2023-09-12T07:08:22
2023-09-12T07:08:22
{ "title": [ -0.012741089, 0.022766113, -0.03829956, -0.016540527, 0.009773254, -0.03274536, -0.009254456, -0.030273438, 0.02029419, 0.035583496, -0.03451538, -0.01651001, 0.026947021, -0.018753052, -0.031173706, -0.00084638596, -0.0054893494, 0....
http://arxiv.org/abs/2309.07927v2
Kid-Whisper: Towards Bridging the Performance Gap in Automatic Speech Recognition for Children VS. Adults
Ahmed Adel Attia, Jing Liu, Wei Ai, Dorottya Demszky, Carol Espy-Wilson
eess.AS, cs.CL, cs.SD
Recent advancements in Automatic Speech Recognition (ASR) systems, exemplified by Whisper, have demonstrated the potential of these systems to approach human-level performance given sufficient data. However, this progress doesn't readily extend to ASR for children due to the limited availability of suitable child-speci...
http://arxiv.org/pdf/2309.07927v2
2023-09-18T09:56:20
2023-09-12T06:58:18
{ "title": [ -0.02458191, -0.009361267, -0.033172607, -0.046417236, -0.0045661926, -0.0020942688, -0.029251099, -0.004627228, -0.022033691, 0.05319214, -0.044647217, 0.01309967, 0.0135269165, 0.0034866333, -0.054901123, -0.018722534, -0.014434814, ...
http://arxiv.org/abs/2309.08626v1
Improving Robustness of Neural Inverse Text Normalization via Data-Augmentation, Semi-Supervised Learning, and Post-Aligning Method
Juntae Kim, Minkyu Lim, Seokjin Hong
cs.CL
Inverse text normalization (ITN) is crucial for converting spoken-form into written-form, especially in the context of automatic speech recognition (ASR). While most downstream tasks of ASR rely on written-form, ASR systems often output spoken-form, highlighting the necessity for robust ITN in product-level ASR-based a...
http://arxiv.org/pdf/2309.08626v1
2023-09-12T06:05:57
2023-09-12T06:05:57
{ "title": [ -0.018554688, 0.017623901, -0.030899048, -0.028884888, 0.008262634, 0.0015678406, -0.03604126, -0.02734375, -0.01751709, 0.051208496, -0.013465881, 0.038757324, 0.010795593, 0.0038452148, -0.034362793, 0.011009216, -0.0031929016, 0.0...
http://arxiv.org/abs/2309.08625v1
Performance of ChatGPT-3.5 and GPT-4 on the United States Medical Licensing Examination With and Without Distractions
Myriam Safrai, Amos Azaria
cs.CL
As Large Language Models (LLMs) are predictive models building their response based on the words in the prompts, there is a risk that small talk and irrelevant information may alter the response and the suggestion given. Therefore, this study aims to investigate the impact of medical data mixed with small talk on the a...
http://arxiv.org/pdf/2309.08625v1
2023-09-12T05:54:45
2023-09-12T05:54:45
{ "title": [ 0.047698975, 0.014030457, 0.020599365, -0.010551453, 0.003189087, -0.02267456, 0.010551453, -0.05041504, -0.009284973, 0.033599854, -0.013389587, 0.014472961, 0.0317688, -0.034240723, 0.0043029785, 0.01134491, -0.025344849, 0.0274047...
http://arxiv.org/abs/2309.05973v1
Circuit Breaking: Removing Model Behaviors with Targeted Ablation
Maximilian Li, Xander Davies, Max Nadeau
cs.CL, cs.LG
Language models often exhibit behaviors that improve performance on a pre-training objective but harm performance on downstream tasks. We propose a novel approach to removing undesirable behaviors by ablating a small number of causal pathways between model components, with the intention of disabling the computational c...
http://arxiv.org/pdf/2309.05973v1
2023-09-12T05:51:56
2023-09-12T05:51:56
{ "title": [ 0.050476074, 0.030548096, 0.0028762817, -0.03955078, 0.018676758, -0.02204895, -0.010810852, -0.057922363, -0.029876709, 0.033843994, 0.028884888, -0.008033752, 0.032043457, -0.060333252, 0.013786316, 0.027420044, 0.0068244934, 0.019...
http://arxiv.org/abs/2309.05961v2
Evaluating the Ebb and Flow: An In-depth Analysis of Question-Answering Trends across Diverse Platforms
Rima Hazra, Agnik Saha, Somnath Banerjee, Animesh Mukherjee
cs.SI, cs.CL, cs.IR, cs.LG
Community Question Answering (CQA) platforms steadily gain popularity as they provide users with fast responses to their queries. The swiftness of these responses is contingent on a mixture of query-specific and user-related elements. This paper scrutinizes these contributing factors within the context of six highly po...
http://arxiv.org/pdf/2309.05961v2
2023-10-31T06:16:40
2023-09-12T05:03:28
{ "title": [ 0.030166626, -0.011856079, -0.039794922, -0.046691895, 0.006214142, -0.040161133, -0.009246826, 0.007926941, 0.01096344, 0.035339355, -0.030776978, 0.013748169, 0.027389526, 0.0028133392, -0.049743652, 0.010734558, 0.0063934326, -0.0...
http://arxiv.org/abs/2309.05958v1
The Moral Machine Experiment on Large Language Models
Kazuhiro Takemoto
cs.CL, cs.CY, cs.HC
As large language models (LLMs) become more deeply integrated into various sectors, understanding how they make moral judgments has become crucial, particularly in the realm of autonomous driving. This study utilized the Moral Machine framework to investigate the ethical decision-making tendencies of prominent LLMs, in...
http://arxiv.org/pdf/2309.05958v1
2023-09-12T04:49:39
2023-09-12T04:49:39
{ "title": [ -0.02458191, -0.0069885254, -0.039733887, -0.032226562, 0.0031833649, -0.017044067, -0.017852783, -0.0047454834, -0.011749268, 0.03652954, -0.026809692, 0.002292633, 0.015914917, -0.013519287, -0.03265381, 0.011123657, -0.02381897, 0...