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/2302.08081v1
Exploring the Limits of ChatGPT for Query or Aspect-based Text Summarization
Xianjun Yang, Yan Li, Xinlu Zhang, Haifeng Chen, Wei Cheng
cs.CL, cs.AI
Text summarization has been a crucial problem in natural language processing (NLP) for several decades. It aims to condense lengthy documents into shorter versions while retaining the most critical information. Various methods have been proposed for text summarization, including extractive and abstractive summarization...
http://arxiv.org/pdf/2302.08081v1
2023-02-16T04:41:30
2023-02-16T04:41:30
{ "title": [ 0.007911682, -0.018112183, 0.0038471222, -0.058776855, -0.023010254, 0.027145386, -0.037597656, 0.0064849854, -0.0076141357, 0.07293701, -0.0004415512, 0.032470703, 0.012069702, 0.0054893494, -0.060638428, 0.025360107, -0.0072746277, ...
http://arxiv.org/abs/2302.08079v1
Document Flattening: Beyond Concatenating Context for Document-Level Neural Machine Translation
Minghao Wu, George Foster, Lizhen Qu, Gholamreza Haffari
cs.CL
Existing work in document-level neural machine translation commonly concatenates several consecutive sentences as a pseudo-document, and then learns inter-sentential dependencies. This strategy limits the model's ability to leverage information from distant context. We overcome this limitation with a novel Document Fla...
http://arxiv.org/pdf/2302.08079v1
2023-02-16T04:38:34
2023-02-16T04:38:34
{ "title": [ -0.0138168335, -0.010314941, -0.004184723, -0.06964111, -0.00049352646, 0.020462036, -0.048614502, -0.004550934, 0.0057754517, 0.052520752, -0.01626587, 0.0038776398, 0.025650024, -0.0025501251, -0.042022705, 0.011375427, 0.00036525726, ...
http://arxiv.org/abs/2302.08068v2
LabelPrompt: Effective Prompt-based Learning for Relation Classification
Wenjie Zhang, Xiaoning Song, Zhenhua Feng, Tianyang Xu, Xiaojun Wu
cs.CL, cs.AI, cs.IR, cs.LG
Recently, prompt-based learning has gained popularity across many natural language processing (NLP) tasks by reformulating them into a cloze-style format to better align pre-trained language models (PLMs) with downstream tasks. However, applying this approach to relation classification poses unique challenges. Specific...
http://arxiv.org/pdf/2302.08068v2
2023-08-13T03:04:08
2023-02-16T04:06:25
{ "title": [ 0.0057258606, -0.0067100525, -0.0552063, -0.0690918, 0.008171082, -0.019592285, -0.038330078, -0.0024318695, 0.02960205, -0.018234253, -0.041992188, -0.002828598, -0.01638794, -0.030380249, -0.01864624, -0.02154541, -0.011711121, 0.0...
http://arxiv.org/abs/2302.10315v1
Generalization algorithm of multimodal pre-training model based on graph-text self-supervised training
Zhangxiaobing, Tangzhenhao, Longzi, Fuxianghua
cs.CL, cs.AI
Recently, a large number of studies have shown that the introduction of visual information can effectively improve the effect of neural machine translation (NMT). Its effectiveness largely depends on the availability of a large number of bilingual parallel sentence pairs and manual image annotation. The lack of images ...
http://arxiv.org/pdf/2302.10315v1
2023-02-16T03:34:08
2023-02-16T03:34:08
{ "title": [ -0.021697998, 0.023529053, -0.03591919, -0.05340576, -0.011779785, 0.006832123, -0.06072998, -0.010444641, -0.028579712, 0.04824829, 0.021957397, 0.036468506, -0.015533447, -0.0345459, -0.015296936, 0.007472992, -0.011703491, 0.05700...
http://arxiv.org/abs/2302.08043v3
GraphPrompt: Unifying Pre-Training and Downstream Tasks for Graph Neural Networks
Zemin Liu, Xingtong Yu, Yuan Fang, Xinming Zhang
cs.LG, cs.CL
Graphs can model complex relationships between objects, enabling a myriad of Web applications such as online page/article classification and social recommendation. While graph neural networks(GNNs) have emerged as a powerful tool for graph representation learning, in an end-to-end supervised setting, their performance ...
http://arxiv.org/pdf/2302.08043v3
2023-02-25T03:00:10
2023-02-16T02:51:38
{ "title": [ -0.014152527, -0.0043945312, -0.043518066, -0.059295654, 0.0075416565, -0.001832962, -0.043121338, -0.0059051514, 0.009651184, 0.02267456, -0.010551453, 0.020217896, -0.017059326, -0.021957397, -0.014656067, 0.01838684, 0.003376007, ...
http://arxiv.org/abs/2302.09300v1
Search-Engine-augmented Dialogue Response Generation with Cheaply Supervised Query Production
Ante Wang, Linfeng Song, Qi Liu, Haitao Mi, Longyue Wang, Zhaopeng Tu, Jinsong Su, Dong Yu
cs.CL, cs.AI
Knowledge-aided dialogue response generation aims at augmenting chatbots with relevant external knowledge in the hope of generating more informative responses. The majority of previous work assumes that the relevant knowledge is given as input or retrieved from a static pool of knowledge. However, this assumption viola...
http://arxiv.org/pdf/2302.09300v1
2023-02-16T01:58:10
2023-02-16T01:58:10
{ "title": [ 0.010322571, -0.018600464, -0.020202637, -0.05682373, -0.0021686554, -0.0018043518, -0.013504028, 0.01411438, -0.012306213, 0.016281128, -0.0154953, 0.015670776, -0.03314209, -0.020126343, -0.046569824, -0.019119263, -0.015449524, 0....
http://arxiv.org/abs/2302.07994v1
À-la-carte Prompt Tuning (APT): Combining Distinct Data Via Composable Prompting
Benjamin Bowman, Alessandro Achille, Luca Zancato, Matthew Trager, Pramuditha Perera, Giovanni Paolini, Stefano Soatto
cs.LG, cs.AI, cs.CL, cs.CV
We introduce \`A-la-carte Prompt Tuning (APT), a transformer-based scheme to tune prompts on distinct data so that they can be arbitrarily composed at inference time. The individual prompts can be trained in isolation, possibly on different devices, at different times, and on different distributions or domains. Further...
http://arxiv.org/pdf/2302.07994v1
2023-02-15T23:51:09
2023-02-15T23:51:09
{ "title": [ -0.017944336, -0.018188477, -0.06463623, -0.111328125, -0.010696411, -0.004135132, -0.08428955, 0.025314331, 0.0035953522, 0.018585205, -0.02810669, 0.016433716, -0.018157959, -0.025878906, -0.008491516, -0.0064048767, 0.008903503, 0...
http://arxiv.org/abs/2302.07974v1
Tree-Based Representation and Generation of Natural and Mathematical Language
Alexander Scarlatos, Andrew Lan
cs.CL
Mathematical language in scientific communications and educational scenarios is important yet relatively understudied compared to natural languages. Recent works on mathematical language focus either on representing stand-alone mathematical expressions, especially in their natural tree format, or mathematical reasoning...
http://arxiv.org/pdf/2302.07974v1
2023-02-15T22:38:34
2023-02-15T22:38:34
{ "title": [ -0.03515625, -0.01159668, -0.04559326, -0.0635376, 0.0053596497, 0.0090408325, -0.042236328, 0.013313293, 0.01486969, 0.0340271, -0.013473511, -0.009857178, -0.00869751, -0.012863159, -0.05331421, 0.009399414, 0.0017118454, 0.0238952...
http://arxiv.org/abs/2302.07926v1
Commonsense Reasoning for Conversational AI: A Survey of the State of the Art
Christopher Richardson, Larry Heck
cs.CL, cs.LG
Large, transformer-based pretrained language models like BERT, GPT, and T5 have demonstrated a deep understanding of contextual semantics and language syntax. Their success has enabled significant advances in conversational AI, including the development of open-dialogue systems capable of coherent, salient conversation...
http://arxiv.org/pdf/2302.07926v1
2023-02-15T19:55:57
2023-02-15T19:55:57
{ "title": [ -0.008399963, -0.02798462, -0.022537231, -0.05859375, -0.0020217896, 0.0052757263, -0.038970947, 0.0020599365, 0.02835083, 0.015975952, -0.015945435, -0.0008239746, -0.0093307495, -0.032165527, -0.03805542, 0.0030765533, -0.019760132, ...
http://arxiv.org/abs/2302.07912v1
Meeting the Needs of Low-Resource Languages: The Value of Automatic Alignments via Pretrained Models
Abteen Ebrahimi, Arya D. McCarthy, Arturo Oncevay, Luis Chiruzzo, John E. Ortega, Gustavo A. Giménez-Lugo, Rolando Coto-Solano, Katharina Kann
cs.CL
Large multilingual models have inspired a new class of word alignment methods, which work well for the model's pretraining languages. However, the languages most in need of automatic alignment are low-resource and, thus, not typically included in the pretraining data. In this work, we ask: How do modern aligners perfor...
http://arxiv.org/pdf/2302.07912v1
2023-02-15T19:06:17
2023-02-15T19:06:17
{ "title": [ -0.016799927, -0.014213562, -0.03338623, -0.076416016, -0.022827148, 0.0035476685, -0.09979248, -0.00049877167, -0.00198555, 0.048614502, 0.004714966, -0.0044403076, 0.0155181885, -0.03353882, -0.044799805, 0.01197052, -0.0013628006, ...
http://arxiv.org/abs/2302.07866v1
Do Deep Neural Networks Capture Compositionality in Arithmetic Reasoning?
Keito Kudo, Yoichi Aoki, Tatsuki Kuribayashi, Ana Brassard, Masashi Yoshikawa, Keisuke Sakaguchi, Kentaro Inui
cs.CL, cs.AI
Compositionality is a pivotal property of symbolic reasoning. However, how well recent neural models capture compositionality remains underexplored in the symbolic reasoning tasks. This study empirically addresses this question by systematically examining recently published pre-trained seq2seq models with a carefully c...
http://arxiv.org/pdf/2302.07866v1
2023-02-15T18:59:04
2023-02-15T18:59:04
{ "title": [ 0.00022351742, 0.00894928, -0.05041504, -0.08154297, 0.01260376, -0.018920898, -0.038146973, 0.019699097, 0.009407043, 0.034088135, -0.0070610046, -0.012390137, 0.017166138, -0.026611328, -0.038726807, 0.0053367615, -0.016433716, 0.0...
http://arxiv.org/abs/2302.07863v4
Speculative Decoding with Big Little Decoder
Sehoon Kim, Karttikeya Mangalam, Suhong Moon, Jitendra Malik, Michael W. Mahoney, Amir Gholami, Kurt Keutzer
cs.CL
The recent emergence of Large Language Models based on the Transformer architecture has enabled dramatic advancements in the field of Natural Language Processing. However, these models have long inference latency, which limits their deployment and makes them prohibitively expensive for various real-time applications. T...
http://arxiv.org/pdf/2302.07863v4
2023-10-12T17:23:56
2023-02-15T18:55:29
{ "title": [ -0.029067993, -0.016479492, -0.02848816, -0.024093628, -0.07647705, -0.007843018, -0.048095703, 0.0059547424, -0.019500732, 0.05731201, -0.033935547, -0.036499023, 0.034179688, -0.03326416, -0.01713562, -0.0016860962, 0.032196045, 0....
http://arxiv.org/abs/2302.07856v1
Dictionary-based Phrase-level Prompting of Large Language Models for Machine Translation
Marjan Ghazvininejad, Hila Gonen, Luke Zettlemoyer
cs.CL, cs.LG
Large language models (LLMs) demonstrate remarkable machine translation (MT) abilities via prompting, even though they were not explicitly trained for this task. However, even given the incredible quantities of data they are trained on, LLMs can struggle to translate inputs with rare words, which are common in low reso...
http://arxiv.org/pdf/2302.07856v1
2023-02-15T18:46:42
2023-02-15T18:46:42
{ "title": [ -0.03567505, -0.014289856, -0.01007843, -0.0680542, -0.018203735, 0.02166748, -0.06903076, 0.012481689, 0.0112838745, 0.025939941, -0.03213501, 0.0016822815, 0.02293396, -0.014137268, -0.04534912, 0.01121521, 0.008369446, -0.00493621...
http://arxiv.org/abs/2302.07845v3
NL2CMD: An Updated Workflow for Natural Language to Bash Commands Translation
Quchen Fu, Zhongwei Teng, Marco Georgaklis, Jules White, Douglas C. Schmidt
cs.CL, cs.AI, cs.PF
Translating natural language into Bash Commands is an emerging research field that has gained attention in recent years. Most efforts have focused on producing more accurate translation models. To the best of our knowledge, only two datasets are available, with one based on the other. Both datasets involve scraping thr...
http://arxiv.org/pdf/2302.07845v3
2023-06-18T16:27:16
2023-02-15T18:31:36
{ "title": [ -0.051513672, -0.02671814, -0.05102539, -0.040252686, -0.02017212, 0.0042152405, -0.05709839, 0.013641357, -0.028366089, 0.060668945, 0.013496399, 0.0069084167, 0.051452637, -0.047668457, -0.008460999, 0.026016235, 0.0037822723, -0.0...
http://arxiv.org/abs/2302.07842v1
Augmented Language Models: a Survey
Grégoire Mialon, Roberto Dessì, Maria Lomeli, Christoforos Nalmpantis, Ram Pasunuru, Roberta Raileanu, Baptiste Rozière, Timo Schick, Jane Dwivedi-Yu, Asli Celikyilmaz, Edouard Grave, Yann LeCun, Thomas Scialom
cs.CL
This survey reviews works in which language models (LMs) are augmented with reasoning skills and the ability to use tools. The former is defined as decomposing a potentially complex task into simpler subtasks while the latter consists in calling external modules such as a code interpreter. LMs can leverage these augmen...
http://arxiv.org/pdf/2302.07842v1
2023-02-15T18:25:52
2023-02-15T18:25:52
{ "title": [ -0.031341553, 0.0023288727, -0.024673462, -0.06427002, 0.006439209, 0.0256958, -0.074401855, -0.007904053, -0.010681152, 0.047851562, 0.009590149, -0.00062704086, 0.011505127, -0.023788452, -0.048950195, 0.019088745, -0.029449463, 0....
http://arxiv.org/abs/2302.07778v2
Measuring the Instability of Fine-Tuning
Yupei Du, Dong Nguyen
cs.CL
Fine-tuning pre-trained language models on downstream tasks with varying random seeds has been shown to be unstable, especially on small datasets. Many previous studies have investigated this instability and proposed methods to mitigate it. However, most studies only used the standard deviation of performance scores (S...
http://arxiv.org/pdf/2302.07778v2
2023-10-01T10:38:39
2023-02-15T16:55:15
{ "title": [ 0.020187378, 0.010017395, 0.025253296, -0.011154175, -0.031204224, 0.00894165, -0.027709961, -0.046722412, 0.0014467239, 0.041015625, -0.03555298, -0.0014953613, -0.0011529922, -0.013244629, -0.018341064, -0.0077056885, 0.030532837, ...
http://arxiv.org/abs/2302.07748v4
Whats New? Identifying the Unfolding of New Events in Narratives
Seyed Mahed Mousavi, Shohei Tanaka, Gabriel Roccabruna, Koichiro Yoshino, Satoshi Nakamura, Giuseppe Riccardi
cs.CL
Narratives include a rich source of events unfolding over time and context. Automatic understanding of these events provides a summarised comprehension of the narrative for further computation (such as reasoning). In this paper, we study the Information Status (IS) of the events and propose a novel challenging task: th...
http://arxiv.org/pdf/2302.07748v4
2023-08-08T08:08:12
2023-02-15T15:54:01
{ "title": [ -0.0030288696, 0.0009536743, -0.022537231, -0.07183838, 0.02458191, -0.007949829, -0.043060303, -0.010253906, -0.023986816, 0.01889038, -0.008110046, 0.047180176, -0.01474762, -0.002817154, -0.01739502, 0.0021038055, 0.013458252, 0.0...
http://arxiv.org/abs/2302.07679v1
On graph-based reentrancy-free semantic parsing
Alban Petit, Caio Corro
cs.CL, cs.LG
We propose a novel graph-based approach for semantic parsing that resolves two problems observed in the literature: (1) seq2seq models fail on compositional generalization tasks; (2) previous work using phrase structure parsers cannot cover all the semantic parses observed in treebanks. We prove that both MAP inference...
http://arxiv.org/pdf/2302.07679v1
2023-02-15T14:14:09
2023-02-15T14:14:09
{ "title": [ -0.028884888, -0.0046653748, -0.022766113, -0.088256836, -0.0039138794, 0.025787354, -0.08428955, 0.0070266724, -0.012329102, 0.033233643, -0.0036125183, 0.013771057, -0.026107788, -0.007774353, -0.04486084, 0.011062622, -0.02368164, ...
http://arxiv.org/abs/2302.07636v2
DP-BART for Privatized Text Rewriting under Local Differential Privacy
Timour Igamberdiev, Ivan Habernal
cs.CR, cs.CL
Privatized text rewriting with local differential privacy (LDP) is a recent approach that enables sharing of sensitive textual documents while formally guaranteeing privacy protection to individuals. However, existing systems face several issues, such as formal mathematical flaws, unrealistic privacy guarantees, privat...
http://arxiv.org/pdf/2302.07636v2
2023-06-06T14:17:46
2023-02-15T13:07:34
{ "title": [ -0.036499023, -0.017578125, -0.041778564, -0.10662842, -0.0039253235, 0.022094727, -0.08459473, 0.032287598, -0.033233643, 0.00818634, -0.02407837, 0.04345703, 0.0012168884, -0.018096924, 0.012832642, -0.006095886, -0.021774292, 0.03...
http://arxiv.org/abs/2302.07588v1
Word class representations spontaneously emerge in a deep neural network trained on next word prediction
Kishore Surendra, Achim Schilling, Paul Stoewer, Andreas Maier, Patrick Krauss
cs.CL, cs.AI, q-bio.NC
How do humans learn language, and can the first language be learned at all? These fundamental questions are still hotly debated. In contemporary linguistics, there are two major schools of thought that give completely opposite answers. According to Chomsky's theory of universal grammar, language cannot be learned becau...
http://arxiv.org/pdf/2302.07588v1
2023-02-15T11:02:50
2023-02-15T11:02:50
{ "title": [ -0.002357483, 0.008346558, -0.03643799, -0.03527832, 0.011955261, -0.021652222, -0.013694763, -0.012199402, -0.02558899, 0.04449463, -0.023757935, 0.029907227, -0.018936157, 0.013160706, -0.039123535, -0.013786316, -0.0017080307, 0.0...
http://arxiv.org/abs/2302.07492v1
Envisioning the Next-Gen Document Reader
Catherine Yeh, Nedim Lipka, Franck Dernoncourt
cs.CL, cs.AI, cs.HC, cs.LG
People read digital documents on a daily basis to share, exchange, and understand information in electronic settings. However, current document readers create a static, isolated reading experience, which does not support users' goals of gaining more knowledge and performing additional tasks through document interaction...
http://arxiv.org/pdf/2302.07492v1
2023-02-15T06:43:12
2023-02-15T06:43:12
{ "title": [ 0.00092983246, -0.02281189, 0.005924225, -0.018997192, -0.080444336, 0.0032844543, -0.030700684, 0.008575439, -0.009712219, 0.062805176, -0.0390625, -0.029693604, 0.016952515, -0.0115737915, -0.03579712, -0.024032593, 0.017120361, 0....
http://arxiv.org/abs/2302.09068v1
A Pilot Evaluation of ChatGPT and DALL-E 2 on Decision Making and Spatial Reasoning
Zhisheng Tang, Mayank Kejriwal
cs.AI, cs.CL, cs.CV
We conduct a pilot study selectively evaluating the cognitive abilities (decision making and spatial reasoning) of two recently released generative transformer models, ChatGPT and DALL-E 2. Input prompts were constructed following neutral a priori guidelines, rather than adversarial intent. Post hoc qualitative analysi...
http://arxiv.org/pdf/2302.09068v1
2023-02-15T05:04:49
2023-02-15T05:04:49
{ "title": [ -0.015411377, -0.009536743, -0.01977539, 0.018325806, -0.010307312, -0.01953125, 0.018554688, -0.02835083, -0.009284973, 0.008216858, -0.023727417, 0.0035419464, -0.018417358, -0.023498535, -0.012413025, -0.0067596436, -0.012390137, ...
http://arxiv.org/abs/2302.07459v2
The Capacity for Moral Self-Correction in Large Language Models
Deep Ganguli, Amanda Askell, Nicholas Schiefer, Thomas I. Liao, Kamilė Lukošiūtė, Anna Chen, Anna Goldie, Azalia Mirhoseini, Catherine Olsson, Danny Hernandez, Dawn Drain, Dustin Li, Eli Tran-Johnson, Ethan Perez, Jackson Kernion, Jamie Kerr, Jared Mueller, Joshua Landau, Kamal Ndousse, Karina Nguyen, Liane Lovitt, Mic...
cs.CL
We test the hypothesis that language models trained with reinforcement learning from human feedback (RLHF) have the capability to "morally self-correct" -- to avoid producing harmful outputs -- if instructed to do so. We find strong evidence in support of this hypothesis across three different experiments, each of whic...
http://arxiv.org/pdf/2302.07459v2
2023-02-18T21:30:27
2023-02-15T04:25:40
{ "title": [ -0.007587433, 0.005836487, -0.040802002, -0.040039062, 0.004535675, -0.029769897, -0.015930176, -0.00094652176, -0.0067253113, 0.039855957, -0.017913818, -0.007621765, 0.0012369156, -0.010498047, -0.0625, -0.00223732, -0.014564514, 0...
http://arxiv.org/abs/2302.07452v1
How to Train Your DRAGON: Diverse Augmentation Towards Generalizable Dense Retrieval
Sheng-Chieh Lin, Akari Asai, Minghan Li, Barlas Oguz, Jimmy Lin, Yashar Mehdad, Wen-tau Yih, Xilun Chen
cs.IR, cs.CL
Various techniques have been developed in recent years to improve dense retrieval (DR), such as unsupervised contrastive learning and pseudo-query generation. Existing DRs, however, often suffer from effectiveness tradeoffs between supervised and zero-shot retrieval, which some argue was due to the limited model capaci...
http://arxiv.org/pdf/2302.07452v1
2023-02-15T03:53:26
2023-02-15T03:53:26
{ "title": [ 0.010932922, 0.011436462, -0.048675537, -0.06732178, 0.036743164, -0.004924774, -0.009140015, -0.0036811829, 0.02670288, 0.032409668, -0.024337769, 0.022140503, -0.004638672, 0.0014333725, -0.043182373, 0.022628784, 0.0028190613, 0.0...
http://arxiv.org/abs/2302.07388v1
Adding Instructions during Pretraining: Effective Way of Controlling Toxicity in Language Models
Shrimai Prabhumoye, Mostofa Patwary, Mohammad Shoeybi, Bryan Catanzaro
cs.CL, cs.AI
Pretrained large language models have become indispensable for solving various natural language processing (NLP) tasks. However, safely deploying them in real world applications is challenging because they generate toxic content. To address this challenge, we propose two novel pretraining data augmentation strategies t...
http://arxiv.org/pdf/2302.07388v1
2023-02-14T23:00:42
2023-02-14T23:00:42
{ "title": [ -0.0023021698, -0.0022563934, -0.03149414, -0.07318115, -0.0010538101, -0.0082473755, -0.050994873, -0.012718201, -0.041900635, 0.039031982, -0.0034866333, 0.023757935, 0.024108887, -0.032165527, -0.0025901794, 0.005680084, 0.004989624, ...
http://arxiv.org/abs/2302.07371v2
BiasTestGPT: Using ChatGPT for Social Bias Testing of Language Models
Rafal Kocielnik, Shrimai Prabhumoye, Vivian Zhang, Roy Jiang, R. Michael Alvarez, Anima Anandkumar
cs.CL, cs.CY, 68T50, I.2.7; J.5; K.4.1
Pretrained Language Models (PLMs) harbor inherent social biases that can result in harmful real-world implications. Such social biases are measured through the probability values that PLMs output for different social groups and attributes appearing in a set of test sentences. However, bias testing is currently cumberso...
http://arxiv.org/pdf/2302.07371v2
2023-09-04T04:34:57
2023-02-14T22:07:57
{ "title": [ -0.023239136, -0.01209259, -0.034698486, -0.024795532, -0.006843567, -0.009796143, -0.028076172, -0.01852417, -0.011405945, 0.03387451, -0.033599854, 0.01134491, 0.023483276, -0.015464783, -0.023406982, 0.0031528473, -0.0418396, 0.06...
http://arxiv.org/abs/2302.07346v1
ScatterShot: Interactive In-context Example Curation for Text Transformation
Tongshuang Wu, Hua Shen, Daniel S. Weld, Jeffrey Heer, Marco Tulio Ribeiro
cs.HC, cs.CL
The in-context learning capabilities of LLMs like GPT-3 allow annotators to customize an LLM to their specific tasks with a small number of examples. However, users tend to include only the most obvious patterns when crafting examples, resulting in underspecified in-context functions that fall short on unseen cases. Fu...
http://arxiv.org/pdf/2302.07346v1
2023-02-14T21:13:31
2023-02-14T21:13:31
{ "title": [ 0.008956909, -0.027130127, -0.038391113, -0.03074646, -0.015823364, -0.020843506, -0.0059013367, -0.0066108704, 0.014045715, 0.042144775, -0.029968262, 0.0017652512, 0.054138184, -0.027877808, -0.008361816, -0.010292053, 0.0006365776, ...
http://arxiv.org/abs/2302.07324v2
READIN: A Chinese Multi-Task Benchmark with Realistic and Diverse Input Noises
Chenglei Si, Zhengyan Zhang, Yingfa Chen, Xiaozhi Wang, Zhiyuan Liu, Maosong Sun
cs.CL
For many real-world applications, the user-generated inputs usually contain various noises due to speech recognition errors caused by linguistic variations1 or typographical errors (typos). Thus, it is crucial to test model performance on data with realistic input noises to ensure robustness and fairness. However, litt...
http://arxiv.org/pdf/2302.07324v2
2023-05-25T01:04:08
2023-02-14T20:14:39
{ "title": [ -0.04345703, -0.004962921, -0.017684937, -0.04928589, -0.016296387, 0.005554199, -0.04916382, -0.018356323, -0.00061416626, 0.047790527, -0.0042533875, 0.017318726, -0.012199402, -0.012901306, -0.054779053, 0.01474762, -0.008590698, ...
http://arxiv.org/abs/2302.07322v2
TRESTLE: Toolkit for Reproducible Execution of Speech, Text and Language Experiments
Changye Li, Weizhe Xu, Trevor Cohen, Martin Michalowski, Serguei Pakhomov
cs.CL
The evidence is growing that machine and deep learning methods can learn the subtle differences between the language produced by people with various forms of cognitive impairment such as dementia and cognitively healthy individuals. Valuable public data repositories such as TalkBank have made it possible for researcher...
http://arxiv.org/pdf/2302.07322v2
2023-03-14T19:35:05
2023-02-14T20:07:31
{ "title": [ -0.036499023, -0.022262573, -0.061767578, -0.06286621, 0.0049057007, -0.007091522, -0.066833496, 0.0059013367, -0.007598877, 0.025131226, 0.006427765, 0.009208679, 0.009384155, -0.027389526, -0.039764404, 0.01626587, -0.013931274, 0....
http://arxiv.org/abs/2302.07232v1
A Psycholinguistic Analysis of BERT's Representations of Compounds
Lars Buijtelaar, Sandro Pezzelle
cs.CL, I.2.7; J.5
This work studies the semantic representations learned by BERT for compounds, that is, expressions such as sunlight or bodyguard. We build on recent studies that explore semantic information in Transformers at the word level and test whether BERT aligns with human semantic intuitions when dealing with expressions (e.g....
http://arxiv.org/pdf/2302.07232v1
2023-02-14T18:23:15
2023-02-14T18:23:15
{ "title": [ 0.02784729, -0.03152466, 0.0005888939, -0.056518555, -0.05545044, -0.013366699, -0.030761719, 0.02999878, -0.05041504, 0.04714966, -0.020507812, 0.009941101, 0.03982544, -0.023605347, 0.02986145, -0.00894928, -0.021759033, 0.04040527...
http://arxiv.org/abs/2302.07189v4
Reveal the Unknown: Out-of-Knowledge-Base Mention Discovery with Entity Linking
Hang Dong, Jiaoyan Chen, Yuan He, Yinan Liu, Ian Horrocks
cs.CL, I.2.7
Discovering entity mentions that are out of a Knowledge Base (KB) from texts plays a critical role in KB maintenance, but has not yet been fully explored. The current methods are mostly limited to the simple threshold-based approach and feature-based classification, and the datasets for evaluation are relatively rare. ...
http://arxiv.org/pdf/2302.07189v4
2023-09-01T20:32:05
2023-02-14T17:00:06
{ "title": [ -0.0028152466, 0.0011014938, -0.054595947, -0.048461914, 0.010429382, -0.027359009, 0.0058937073, -0.014801025, -0.00617218, 0.014198303, -0.04196167, 0.039611816, 0.021469116, -0.020736694, -0.0031871796, -0.0129852295, 0.00071525574, ...
http://arxiv.org/abs/2302.07159v1
A Friendly Face: Do Text-to-Image Systems Rely on Stereotypes when the Input is Under-Specified?
Kathleen C. Fraser, Svetlana Kiritchenko, Isar Nejadgholi
cs.CY, cs.CL
As text-to-image systems continue to grow in popularity with the general public, questions have arisen about bias and diversity in the generated images. Here, we investigate properties of images generated in response to prompts which are visually under-specified, but contain salient social attributes (e.g., 'a portrait...
http://arxiv.org/pdf/2302.07159v1
2023-02-14T16:11:06
2023-02-14T16:11:06
{ "title": [ 0.0072517395, 0.026519775, -0.022140503, 0.01828003, -0.0029735565, -0.034942627, 0.024795532, -0.048858643, 0.028503418, 0.02406311, -0.072021484, 0.018692017, 0.00012099743, -0.011962891, -0.02357483, -0.037017822, -0.012016296, 0....
http://arxiv.org/abs/2302.07139v1
Modeling Complex Event Scenarios via Simple Entity-focused Questions
Mahnaz Koupaee, Greg Durrett, Nathanael Chambers, Niranjan Balasubramanian
cs.CL
Event scenarios are often complex and involve multiple event sequences connected through different entity participants. Exploring such complex scenarios requires an ability to branch through different sequences, something that is difficult to achieve with standard event language modeling. To address this, we propose a ...
http://arxiv.org/pdf/2302.07139v1
2023-02-14T15:48:56
2023-02-14T15:48:56
{ "title": [ 0.009590149, -0.0027980804, -0.043823242, -0.10308838, 0.027862549, 0.0060768127, -0.044403076, 0.012634277, 0.020217896, 0.014915466, 0.0035209656, -0.0012750626, -0.0048828125, -0.03274536, -0.03668213, 0.021774292, -0.01335907, -0...
http://arxiv.org/abs/2302.07124v1
Exploiting Summarization Data to Help Text Simplification
Renliang Sun, Zhixian Yang, Xiaojun Wan
cs.CL, cs.AI
One of the major problems with text simplification is the lack of high-quality data. The sources of simplification datasets are limited to Wikipedia and Newsela, restricting further development of this field. In this paper, we analyzed the similarity between text summarization and text simplification and exploited summ...
http://arxiv.org/pdf/2302.07124v1
2023-02-14T15:32:04
2023-02-14T15:32:04
{ "title": [ -0.002986908, 0.0035800934, -0.045074463, -0.073913574, -0.02104187, -0.014709473, -0.049560547, 0.0037727356, 0.0077590942, 0.061431885, -0.01965332, 0.023986816, 0.0107803345, -0.00047326088, -0.04815674, 0.008659363, 0.010864258, ...
http://arxiv.org/abs/2302.07739v1
Meta-Learning Triplet Network with Adaptive Margins for Few-Shot Named Entity Recognition
Chengcheng Han, Renyu Zhu, Jun Kuang, FengJiao Chen, Xiang Li, Ming Gao, Xuezhi Cao, Wei Wu
cs.CL
Meta-learning methods have been widely used in few-shot named entity recognition (NER), especially prototype-based methods. However, the Other(O) class is difficult to be represented by a prototype vector because there are generally a large number of samples in the class that have miscellaneous semantics. To solve the ...
http://arxiv.org/pdf/2302.07739v1
2023-02-14T15:10:26
2023-02-14T15:10:26
{ "title": [ 0.0032901764, -0.0023975372, -0.032165527, -0.04562378, -0.004875183, -0.0025978088, -0.04550171, -0.012763977, 0.029129028, 0.039367676, -0.032348633, 0.02998352, -0.016174316, 0.011856079, -0.06506348, 0.008682251, 0.013160706, 0.0...
http://arxiv.org/abs/2302.07027v3
AdapterSoup: Weight Averaging to Improve Generalization of Pretrained Language Models
Alexandra Chronopoulou, Matthew E. Peters, Alexander Fraser, Jesse Dodge
cs.CL
Pretrained language models (PLMs) are trained on massive corpora, but often need to specialize to specific domains. A parameter-efficient adaptation method suggests training an adapter for each domain on the task of language modeling. This leads to good in-domain scores but can be impractical for domain- or resource-re...
http://arxiv.org/pdf/2302.07027v3
2023-03-28T13:37:54
2023-02-14T13:09:23
{ "title": [ -0.016540527, -0.007423401, -0.041900635, -0.06866455, 0.0015821457, -0.0033168793, -0.056640625, 0.010787964, -0.0024528503, 0.02482605, -0.019226074, 0.0073165894, -0.0011043549, -0.013961792, -0.043670654, 0.0021800995, -0.0023612976,...
http://arxiv.org/abs/2302.07729v3
Generation of Highlights from Research Papers Using Pointer-Generator Networks and SciBERT Embeddings
Tohida Rehman, Debarshi Kumar Sanyal, Samiran Chattopadhyay, Plaban Kumar Bhowmick, Partha Pratim Das
cs.CL, cs.AI, cs.LG
Nowadays many research articles are prefaced with research highlights to summarize the main findings of the paper. Highlights not only help researchers precisely and quickly identify the contributions of a paper, they also enhance the discoverability of the article via search engines. We aim to automatically construct ...
http://arxiv.org/pdf/2302.07729v3
2023-09-17T16:45:44
2023-02-14T12:45:14
{ "title": [ 0.010124207, 0.0048980713, -0.020385742, -0.034088135, -0.012268066, -0.010261536, -0.025268555, -0.020980835, -0.009254456, 0.049224854, 0.0126953125, 0.0065727234, -0.016494751, -0.008880615, -0.06390381, -0.00630188, -0.021453857, ...
http://arxiv.org/abs/2302.06976v1
Investigating Multi-source Active Learning for Natural Language Inference
Ard Snijders, Douwe Kiela, Katerina Margatina
cs.CL
In recent years, active learning has been successfully applied to an array of NLP tasks. However, prior work often assumes that training and test data are drawn from the same distribution. This is problematic, as in real-life settings data may stem from several sources of varying relevance and quality. We show that fou...
http://arxiv.org/pdf/2302.06976v1
2023-02-14T11:10:18
2023-02-14T11:10:18
{ "title": [ -0.02003479, -0.009292603, -0.026046753, -0.067993164, 0.00315094, -0.001458168, -0.040161133, 0.0121154785, -0.0020065308, 0.015144348, -0.01499176, 0.0128479, -0.036102295, -0.020385742, -0.022735596, 0.0024852753, -0.00566864, 0.0...
http://arxiv.org/abs/2302.06951v1
Few-shot learning approaches for classifying low resource domain specific software requirements
Anmol Nayak, Hari Prasad Timmapathini, Vidhya Murali, Atul Anil Gohad
cs.CL
With the advent of strong pre-trained natural language processing models like BERT, DeBERTa, MiniLM, T5, the data requirement for industries to fine-tune these models to their niche use cases has drastically reduced (typically to a few hundred annotated samples for achieving a reasonable performance). However, the avai...
http://arxiv.org/pdf/2302.06951v1
2023-02-14T10:19:23
2023-02-14T10:19:23
{ "title": [ -0.013595581, -0.022094727, -0.021713257, -0.049072266, -0.015213013, 0.004211426, -0.030517578, 0.025787354, -0.012863159, 0.0141067505, -0.009025574, 0.009544373, -0.04034424, -0.035308838, -0.033050537, -0.02848816, -0.0029354095, ...
http://arxiv.org/abs/2302.06871v1
Learning gain differences between ChatGPT and human tutor generated algebra hints
Zachary A. Pardos, Shreya Bhandari
cs.CY, cs.CL, cs.HC
Large Language Models (LLMs), such as ChatGPT, are quickly advancing AI to the frontiers of practical consumer use and leading industries to re-evaluate how they allocate resources for content production. Authoring of open educational resources and hint content within adaptive tutoring systems is labor intensive. Shoul...
http://arxiv.org/pdf/2302.06871v1
2023-02-14T07:20:48
2023-02-14T07:20:48
{ "title": [ -0.0124053955, 0.0017223358, -0.018615723, 0.026763916, -0.049072266, -0.020629883, -0.0019140244, -0.022018433, -0.03173828, 0.04776001, -0.024505615, 0.03451538, 0.013931274, -0.020507812, 0.019454956, 0.01084137, -0.003063202, 0.0...
http://arxiv.org/abs/2302.06868v1
SwitchPrompt: Learning Domain-Specific Gated Soft Prompts for Classification in Low-Resource Domains
Koustava Goswami, Lukas Lange, Jun Araki, Heike Adel
cs.CL, cs.AI
Prompting pre-trained language models leads to promising results across natural language processing tasks but is less effective when applied in low-resource domains, due to the domain gap between the pre-training data and the downstream task. In this work, we bridge this gap with a novel and lightweight prompting metho...
http://arxiv.org/pdf/2302.06868v1
2023-02-14T07:14:08
2023-02-14T07:14:08
{ "title": [ -0.009567261, -0.0030002594, -0.0395813, -0.042663574, 0.008155823, -0.024154663, -0.02848816, -0.003566742, 0.0021781921, -0.000091671944, -0.04837036, 0.011199951, -0.0051078796, -0.028320312, -0.00699234, -0.023803711, -0.01651001, ...
http://arxiv.org/abs/2302.06860v2
BLIAM: Literature-based Data Synthesis for Synergistic Drug Combination Prediction
Cai Yang, Addie Woicik, Hoifung Poon, Sheng Wang
cs.CL
Language models pre-trained on scientific literature corpora have substantially advanced scientific discovery by offering high-quality feature representations for downstream applications. However, these features are often not interpretable, and thus can reveal limited insights to domain experts. Instead of obtaining fe...
http://arxiv.org/pdf/2302.06860v2
2023-02-16T05:26:25
2023-02-14T06:48:52
{ "title": [ -0.016937256, 0.018035889, -0.040771484, -0.07104492, 0.004714966, -0.006290436, -0.07635498, -0.01965332, -0.007835388, 0.022827148, -0.012878418, 0.014526367, -0.037994385, -0.017288208, -0.05596924, -0.015052795, -0.024673462, 0.0...
http://arxiv.org/abs/2302.07268v5
AI Chat Assistants can Improve Conversations about Divisive Topics
Lisa P. Argyle, Ethan Busby, Joshua Gubler, Chris Bail, Thomas Howe, Christopher Rytting, David Wingate
cs.HC, cs.AI, cs.CL
A rapidly increasing amount of human conversation occurs online. But divisiveness and conflict can fester in text-based interactions on social media platforms, in messaging apps, and on other digital forums. Such toxicity increases polarization and, importantly, corrodes the capacity of diverse societies to develop eff...
http://arxiv.org/pdf/2302.07268v5
2023-10-20T23:28:08
2023-02-14T06:42:09
{ "title": [ 0.029571533, -0.02558899, -0.002439499, -0.024276733, -0.044067383, -0.017623901, -0.0032310486, -0.0022239685, 0.022064209, 0.03640747, -0.035491943, -0.01663208, 0.040039062, -0.022125244, 0.0017518997, -0.019882202, -0.006668091, ...
http://arxiv.org/abs/2302.06829v2
The Role of Semantic Parsing in Understanding Procedural Text
Hossein Rajaby Faghihi, Parisa Kordjamshidi, Choh Man Teng, James Allen
cs.CL, cs.SC
In this paper, we investigate whether symbolic semantic representations, extracted from deep semantic parsers, can help reasoning over the states of involved entities in a procedural text. We consider a deep semantic parser~(TRIPS) and semantic role labeling as two sources of semantic parsing knowledge. First, we propo...
http://arxiv.org/pdf/2302.06829v2
2023-05-18T02:41:52
2023-02-14T04:59:33
{ "title": [ -0.043884277, -0.011230469, -0.031433105, -0.07043457, -0.021636963, 0.0118255615, -0.09033203, 0.00027871132, -0.01979065, 0.024108887, -0.020584106, 0.01109314, -0.017105103, -0.009628296, -0.04623413, -0.011741638, -0.01499176, 0....
http://arxiv.org/abs/2302.06784v1
The Stable Entropy Hypothesis and Entropy-Aware Decoding: An Analysis and Algorithm for Robust Natural Language Generation
Kushal Arora, Timothy J. O'Donnell, Doina Precup, Jason Weston, Jackie C. K. Cheung
cs.CL
State-of-the-art language generation models can degenerate when applied to open-ended generation problems such as text completion, story generation, or dialog modeling. This degeneration usually shows up in the form of incoherence, lack of vocabulary diversity, and self-repetition or copying from the context. In this p...
http://arxiv.org/pdf/2302.06784v1
2023-02-14T02:02:33
2023-02-14T02:02:33
{ "title": [ -0.038909912, -0.000828743, -0.045806885, -0.0546875, -0.004295349, 0.0096588135, -0.063964844, -0.006416321, 0.00068998337, 0.05722046, -0.009246826, 0.021728516, 0.033447266, -0.0026741028, -0.06896973, 0.026168823, -0.010002136, 0...
http://arxiv.org/abs/2302.06761v3
Language Model Analysis for Ontology Subsumption Inference
Yuan He, Jiaoyan Chen, Ernesto Jiménez-Ruiz, Hang Dong, Ian Horrocks
cs.CL, cs.AI, cs.LO
Investigating whether pre-trained language models (LMs) can function as knowledge bases (KBs) has raised wide research interests recently. However, existing works focus on simple, triple-based, relational KBs, but omit more sophisticated, logic-based, conceptualised KBs such as OWL ontologies. To investigate an LM's kn...
http://arxiv.org/pdf/2302.06761v3
2023-05-08T17:21:40
2023-02-14T00:21:56
{ "title": [ -0.024139404, -0.01272583, -0.041870117, -0.08746338, -0.0067977905, 0.0019207001, -0.06036377, 0.00856781, 0.027786255, -0.0030288696, -0.0501709, -0.013145447, 0.010261536, -0.01777649, -0.022583008, -0.0003695488, -0.017166138, 0....
http://arxiv.org/abs/2302.06729v1
STREET: A Multi-Task Structured Reasoning and Explanation Benchmark
Danilo Ribeiro, Shen Wang, Xiaofei Ma, Henry Zhu, Rui Dong, Deguang Kong, Juliette Burger, Anjelica Ramos, William Wang, Zhiheng Huang, George Karypis, Bing Xiang, Dan Roth
cs.CL, cs.AI, I.2.7; I.2.6
We introduce STREET, a unified multi-task and multi-domain natural language reasoning and explanation benchmark. Unlike most existing question-answering (QA) datasets, we expect models to not only answer questions, but also produce step-by-step structured explanations describing how premises in the question are used to...
http://arxiv.org/pdf/2302.06729v1
2023-02-13T22:34:02
2023-02-13T22:34:02
{ "title": [ -0.040374756, -0.022750854, -0.0423584, -0.103149414, -0.019332886, 0.02104187, -0.0770874, 0.021774292, 0.002286911, -0.0021419525, -0.012001038, 0.0143585205, -0.032196045, -0.023605347, -0.0021152496, -0.00068855286, 0.004348755, ...
http://arxiv.org/abs/2302.06716v3
Machine Learning Model Attribution Challenge
Elizabeth Merkhofer, Deepesh Chaudhari, Hyrum S. Anderson, Keith Manville, Lily Wong, João Gante
cs.LG, cs.CL, cs.CR
We present the findings of the Machine Learning Model Attribution Challenge. Fine-tuned machine learning models may derive from other trained models without obvious attribution characteristics. In this challenge, participants identify the publicly-available base models that underlie a set of anonymous, fine-tuned large...
http://arxiv.org/pdf/2302.06716v3
2023-02-17T15:48:46
2023-02-13T22:05:27
{ "title": [ 0.00009429455, -0.0087509155, -0.02861023, -0.017211914, -0.0075187683, -0.027770996, -0.020523071, -0.026367188, -0.019485474, 0.017303467, -0.023345947, -0.008277893, -0.0028438568, -0.04244995, 0.00655365, -0.010047913, -0.007385254, ...
http://arxiv.org/abs/2302.06706v1
On the Planning Abilities of Large Language Models (A Critical Investigation with a Proposed Benchmark)
Karthik Valmeekam, Sarath Sreedharan, Matthew Marquez, Alberto Olmo, Subbarao Kambhampati
cs.AI, cs.CL, cs.LG
Intrigued by the claims of emergent reasoning capabilities in LLMs trained on general web corpora, in this paper, we set out to investigate their planning capabilities. We aim to evaluate (1) how good LLMs are by themselves in generating and validating simple plans in commonsense planning tasks (of the type that humans...
http://arxiv.org/pdf/2302.06706v1
2023-02-13T21:37:41
2023-02-13T21:37:41
{ "title": [ -0.031051636, -0.011650085, -0.023010254, -0.059051514, -0.016067505, 0.006893158, -0.06137085, -0.00137043, -0.007156372, 0.01737976, -0.022598267, 0.01197052, -0.0069618225, -0.016540527, -0.025390625, -0.0030899048, -0.008407593, ...
http://arxiv.org/abs/2302.06692v2
Guiding Pretraining in Reinforcement Learning with Large Language Models
Yuqing Du, Olivia Watkins, Zihan Wang, Cédric Colas, Trevor Darrell, Pieter Abbeel, Abhishek Gupta, Jacob Andreas
cs.LG, cs.AI, cs.CL
Reinforcement learning algorithms typically struggle in the absence of a dense, well-shaped reward function. Intrinsically motivated exploration methods address this limitation by rewarding agents for visiting novel states or transitions, but these methods offer limited benefits in large environments where most discove...
http://arxiv.org/pdf/2302.06692v2
2023-09-15T02:42:40
2023-02-13T21:16:03
{ "title": [ -0.013290405, -0.0050468445, -0.026885986, -0.047546387, 0.0054740906, -0.008872986, -0.021987915, -0.0093688965, -0.031921387, 0.016418457, -0.012481689, 0.015670776, -0.036376953, -0.022979736, -0.009178162, -0.016662598, 0.00075531006...
http://arxiv.org/abs/2302.06690v1
Bag of Tricks for In-Distribution Calibration of Pretrained Transformers
Jaeyoung Kim, Dongbin Na, Sungchul Choi, Sungbin Lim
cs.CL, cs.LG
While pre-trained language models (PLMs) have become a de-facto standard promoting the accuracy of text classification tasks, recent studies find that PLMs often predict over-confidently. Although various calibration methods have been proposed, such as ensemble learning and data augmentation, most of the methods have b...
http://arxiv.org/pdf/2302.06690v1
2023-02-13T21:11:52
2023-02-13T21:11:52
{ "title": [ 0.000051558018, 0.01474762, -0.030410767, -0.02079773, -0.008430481, -0.021865845, -0.016067505, -0.040039062, -0.035003662, 0.027755737, -0.035064697, 0.019836426, -0.013191223, -0.029373169, 0.0039596558, -0.013626099, 0.02671814, ...
http://arxiv.org/abs/2302.06675v4
Symbolic Discovery of Optimization Algorithms
Xiangning Chen, Chen Liang, Da Huang, Esteban Real, Kaiyuan Wang, Yao Liu, Hieu Pham, Xuanyi Dong, Thang Luong, Cho-Jui Hsieh, Yifeng Lu, Quoc V. Le
cs.LG, cs.AI, cs.CL, cs.CV, cs.NE
We present a method to formulate algorithm discovery as program search, and apply it to discover optimization algorithms for deep neural network training. We leverage efficient search techniques to explore an infinite and sparse program space. To bridge the large generalization gap between proxy and target tasks, we al...
http://arxiv.org/pdf/2302.06675v4
2023-05-08T21:49:57
2023-02-13T20:27:30
{ "title": [ -0.050628662, -0.0020008087, -0.020477295, -0.0001333952, -0.017166138, -0.0035648346, -0.0178833, -0.025314331, -0.03463745, 0.045562744, -0.0025749207, 0.026870728, -0.04425049, -0.007610321, -0.036102295, 0.0026855469, 0.01689148, ...
http://arxiv.org/abs/2302.06674v3
PK-ICR: Persona-Knowledge Interactive Context Retrieval for Grounded Dialogue
Minsik Oh, Joosung Lee, Jiwei Li, Guoyin Wang
cs.CL, cs.IR
Identifying relevant persona or knowledge for conversational systems is critical to grounded dialogue response generation. However, each grounding has been mostly researched in isolation with more practical multi-context dialogue tasks introduced in recent works. We define Persona and Knowledge Dual Context Identificat...
http://arxiv.org/pdf/2302.06674v3
2023-10-19T15:43:06
2023-02-13T20:27:26
{ "title": [ -0.012123108, 0.0020141602, -0.066101074, -0.07763672, 0.030838013, -0.0053100586, -0.022079468, 0.014770508, 0.007675171, 0.04626465, -0.022659302, 0.00982666, 0.012794495, -0.012809753, -0.06237793, -0.0027370453, -0.024978638, 0.0...
http://arxiv.org/abs/2302.06605v2
UniAdapter: Unified Parameter-Efficient Transfer Learning for Cross-modal Modeling
Haoyu Lu, Yuqi Huo, Guoxing Yang, Zhiwu Lu, Wei Zhan, Masayoshi Tomizuka, Mingyu Ding
cs.CV, cs.CL
Large-scale vision-language pre-trained models have shown promising transferability to various downstream tasks. As the size of these foundation models and the number of downstream tasks grow, the standard full fine-tuning paradigm becomes unsustainable due to heavy computational and storage costs. This paper proposes ...
http://arxiv.org/pdf/2302.06605v2
2023-05-21T17:50:30
2023-02-13T18:59:10
{ "title": [ -0.011421204, 0.0072250366, -0.030700684, -0.050994873, 0.011787415, -0.0025558472, -0.052978516, -0.04147339, 0.011024475, 0.023269653, -0.016235352, 0.03125, 0.015266418, -0.020828247, 0.0043754578, 0.023117065, -0.011390686, 0.045...
http://arxiv.org/abs/2302.06600v2
Task-Specific Skill Localization in Fine-tuned Language Models
Abhishek Panigrahi, Nikunj Saunshi, Haoyu Zhao, Sanjeev Arora
cs.CL, cs.LG
Pre-trained language models can be fine-tuned to solve diverse NLP tasks, including in few-shot settings. Thus fine-tuning allows the model to quickly pick up task-specific ``skills,'' but there has been limited study of where these newly-learnt skills reside inside the massive model. This paper introduces the term ski...
http://arxiv.org/pdf/2302.06600v2
2023-07-02T01:55:00
2023-02-13T18:55:52
{ "title": [ -0.021942139, 0.0030021667, -0.021148682, -0.03387451, -0.0076408386, -0.008277893, -0.037628174, -0.008972168, -0.013664246, 0.04449463, -0.0014667511, 0.000644207, 0.02128601, -0.015945435, -0.048675537, 0.010147095, -0.0146102905, ...
http://arxiv.org/abs/2302.06598v1
Gradient-Based Automated Iterative Recovery for Parameter-Efficient Tuning
Maximilian Mozes, Tolga Bolukbasi, Ann Yuan, Frederick Liu, Nithum Thain, Lucas Dixon
cs.CL
Pretrained large language models (LLMs) are able to solve a wide variety of tasks through transfer learning. Various explainability methods have been developed to investigate their decision making process. TracIn (Pruthi et al., 2020) is one such gradient-based method which explains model inferences based on the influe...
http://arxiv.org/pdf/2302.06598v1
2023-02-13T18:54:58
2023-02-13T18:54:58
{ "title": [ -0.031921387, -0.0060768127, -0.02619934, -0.0814209, 0.007118225, 0.016235352, -0.07800293, -0.0088272095, -0.009521484, 0.039886475, 0.010856628, 0.022415161, 0.026412964, -0.018432617, -0.015098572, 0.022842407, 0.008735657, -0.00...
http://arxiv.org/abs/2302.06589v1
Improving Out-of-Distribution Generalization of Neural Rerankers with Contextualized Late Interaction
Xinyu Zhang, Minghan Li, Jimmy Lin
cs.IR, cs.CL
Recent progress in information retrieval finds that embedding query and document representation into multi-vector yields a robust bi-encoder retriever on out-of-distribution datasets. In this paper, we explore whether late interaction, the simplest form of multi-vector, is also helpful to neural rerankers that only use...
http://arxiv.org/pdf/2302.06589v1
2023-02-13T18:42:17
2023-02-13T18:42:17
{ "title": [ 0.04208374, 0.025619507, 0.01876831, -0.044555664, 0.021606445, 0.0075416565, -0.0063552856, -0.054656982, 0.009590149, 0.03390503, -0.0234375, 0.019500732, -0.025131226, -0.015151978, -0.025009155, -0.004676819, -0.008041382, 0.0158...
http://arxiv.org/abs/2302.06579v1
AbLit: A Resource for Analyzing and Generating Abridged Versions of English Literature
Melissa Roemmele, Kyle Shaffer, Katrina Olsen, Yiyi Wang, Steve DeNeefe
cs.CL
Creating an abridged version of a text involves shortening it while maintaining its linguistic qualities. In this paper, we examine this task from an NLP perspective for the first time. We present a new resource, AbLit, which is derived from abridged versions of English literature books. The dataset captures passage-le...
http://arxiv.org/pdf/2302.06579v1
2023-02-13T18:33:26
2023-02-13T18:33:26
{ "title": [ 0.00806427, 0.005760193, -0.05621338, -0.05609131, 0.0043945312, -0.0055999756, -0.038208008, -0.0024719238, -0.02949524, 0.0473938, 0.008834839, 0.01386261, 0.048034668, -0.05090332, -0.004863739, 0.013450623, -0.0059280396, 0.02117...
http://arxiv.org/abs/2302.07735v1
Targeted Attack on GPT-Neo for the SATML Language Model Data Extraction Challenge
Ali Al-Kaswan, Maliheh Izadi, Arie van Deursen
cs.CL, cs.AI, cs.CR
Previous work has shown that Large Language Models are susceptible to so-called data extraction attacks. This allows an attacker to extract a sample that was contained in the training data, which has massive privacy implications. The construction of data extraction attacks is challenging, current attacks are quite inef...
http://arxiv.org/pdf/2302.07735v1
2023-02-13T18:00:44
2023-02-13T18:00:44
{ "title": [ -0.014373779, -0.015434265, -0.016418457, -0.052581787, -0.02949524, 0.0061569214, -0.05682373, -0.01007843, -0.026184082, 0.0057907104, -0.00004351139, -0.011070251, -0.024612427, -0.039642334, -0.010467529, -0.016296387, -0.013000488, ...
http://arxiv.org/abs/2302.06560v1
Large Scale Multi-Lingual Multi-Modal Summarization Dataset
Yash Verma, Anubhav Jangra, Raghvendra Kumar, Sriparna Saha
cs.CL, cs.MM
Significant developments in techniques such as encoder-decoder models have enabled us to represent information comprising multiple modalities. This information can further enhance many downstream tasks in the field of information retrieval and natural language processing; however, improvements in multi-modal techniques...
http://arxiv.org/pdf/2302.06560v1
2023-02-13T18:00:23
2023-02-13T18:00:23
{ "title": [ -0.008575439, 0.010856628, -0.054382324, -0.08544922, 0.020935059, -0.0047912598, -0.0769043, -0.0025939941, 0.009498596, 0.05657959, 0.00020873547, 0.014968872, 0.02015686, -0.011695862, -0.034301758, 0.02583313, 0.00894928, 0.03527...
http://arxiv.org/abs/2302.07267v6
Diminished Diversity-of-Thought in a Standard Large Language Model
Peter S. Park, Philipp Schoenegger, Chongyang Zhu
cs.HC, cs.AI, cs.CL, 68T50, I.2.7
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that...
http://arxiv.org/pdf/2302.07267v6
2023-09-13T07:44:42
2023-02-13T17:57:50
{ "title": [ -0.018997192, -0.0053596497, -0.058288574, -0.037628174, -0.001786232, -0.027282715, -0.031204224, 0.010749817, 0.0066719055, 0.02897644, -0.037750244, -0.0074653625, 0.022232056, -0.023010254, -0.03781128, 0.0138549805, -0.006755829, ...
http://arxiv.org/abs/2302.06555v1
Implications of the Convergence of Language and Vision Model Geometries
Jiaang Li, Yova Kementchedjhieva, Anders Søgaard
cs.CL
Large-scale pretrained language models (LMs) are said to ``lack the ability to connect [their] utterances to the world'' (Bender and Koller, 2020). If so, we would expect LM representations to be unrelated to representations in computer vision models. To investigate this, we present an empirical evaluation across three...
http://arxiv.org/pdf/2302.06555v1
2023-02-13T17:55:54
2023-02-13T17:55:54
{ "title": [ -0.012901306, -0.0032081604, -0.017654419, -0.029708862, -0.03262329, -0.019561768, -0.05029297, -0.024642944, 0.00007522106, 0.040649414, -0.026123047, -0.027832031, 0.030303955, -0.028259277, -0.030975342, -0.0012321472, 0.022781372, ...
http://arxiv.org/abs/2302.06541v2
Towards Agile Text Classifiers for Everyone
Maximilian Mozes, Jessica Hoffmann, Katrin Tomanek, Muhamed Kouate, Nithum Thain, Ann Yuan, Tolga Bolukbasi, Lucas Dixon
cs.CL
Text-based safety classifiers are widely used for content moderation and increasingly to tune generative language model behavior - a topic of growing concern for the safety of digital assistants and chatbots. However, different policies require different classifiers, and safety policies themselves improve from iteratio...
http://arxiv.org/pdf/2302.06541v2
2023-10-21T11:49:09
2023-02-13T17:34:13
{ "title": [ -0.012184143, -0.0016698837, -0.04510498, -0.04852295, 0.0076942444, -0.012275696, -0.029922485, -0.00073719025, -0.0037288666, 0.032104492, -0.033355713, 0.029022217, -0.02734375, 0.010231018, -0.06542969, -0.012214661, -0.009803772, ...
http://arxiv.org/abs/2302.06488v1
Why Can't Discourse Parsing Generalize? A Thorough Investigation of the Impact of Data Diversity
Yang Janet Liu, Amir Zeldes
cs.CL
Recent advances in discourse parsing performance create the impression that, as in other NLP tasks, performance for high-resource languages such as English is finally becoming reliable. In this paper we demonstrate that this is not the case, and thoroughly investigate the impact of data diversity on RST parsing stabili...
http://arxiv.org/pdf/2302.06488v1
2023-02-13T16:11:58
2023-02-13T16:11:58
{ "title": [ 0.005680084, 0.02229309, -0.056243896, -0.09161377, 0.025039673, -0.018127441, -0.052520752, -0.0008239746, -0.023010254, 0.018493652, -0.02911377, 0.01852417, -0.02999878, 0.0018053055, -0.048034668, -0.0259552, -0.020141602, 0.0635...
http://arxiv.org/abs/2302.06461v1
A Study on ReLU and Softmax in Transformer
Kai Shen, Junliang Guo, Xu Tan, Siliang Tang, Rui Wang, Jiang Bian
cs.CL, cs.AI, cs.LG
The Transformer architecture consists of self-attention and feed-forward networks (FFNs) which can be viewed as key-value memories according to previous works. However, FFN and traditional memory utilize different activation functions (i.e., ReLU and Softmax respectively), which makes them not equivalent. In this paper...
http://arxiv.org/pdf/2302.06461v1
2023-02-13T15:41:20
2023-02-13T15:41:20
{ "title": [ -0.004825592, 0.006187439, -0.022201538, -0.022247314, 0.002948761, -0.022964478, -0.009735107, -0.017944336, -0.037475586, 0.039001465, -0.0335083, 0.0064926147, 0.022399902, -0.03475952, 0.012580872, -0.00089788437, 0.008872986, 0....
http://arxiv.org/abs/2302.06459v2
Encoding Sentence Position in Context-Aware Neural Machine Translation with Concatenation
Lorenzo Lupo, Marco Dinarelli, Laurent Besacier
cs.CL
Context-aware translation can be achieved by processing a concatenation of consecutive sentences with the standard Transformer architecture. This paper investigates the intuitive idea of providing the model with explicit information about the position of the sentences contained in the concatenation window. We compare v...
http://arxiv.org/pdf/2302.06459v2
2023-04-04T19:05:47
2023-02-13T15:39:08
{ "title": [ -0.006866455, -0.004627228, 0.0009627342, -0.043518066, -0.0035152435, 0.016677856, -0.02772522, -0.005870819, 0.0035266876, 0.060058594, -0.0040779114, -0.0107421875, 0.046325684, -0.0178833, -0.04095459, 0.02859497, -0.011795044, 0...
http://arxiv.org/abs/2302.06448v1
Joint Span Segmentation and Rhetorical Role Labeling with Data Augmentation for Legal Documents
T. Y. S. S. Santosh, Philipp Bock, Matthias Grabmair
cs.CL, cs.IR
Segmentation and Rhetorical Role Labeling of legal judgements play a crucial role in retrieval and adjacent tasks, including case summarization, semantic search, argument mining etc. Previous approaches have formulated this task either as independent classification or sequence labeling of sentences. In this work, we re...
http://arxiv.org/pdf/2302.06448v1
2023-02-13T15:28:02
2023-02-13T15:28:02
{ "title": [ -0.021240234, -0.018966675, -0.057373047, -0.048675537, -0.021087646, -0.0049095154, -0.05215454, 0.029052734, -0.016799927, 0.008522034, -0.044189453, 0.025924683, 0.0005450249, -0.009407043, 0.002840042, -0.01763916, -0.026535034, ...
http://arxiv.org/abs/2302.10282v1
Paparazzi: A Deep Dive into the Capabilities of Language and Vision Models for Grounding Viewpoint Descriptions
Henrik Voigt, Jan Hombeck, Monique Meuschke, Kai Lawonn, Sina Zarrieß
cs.CV, cs.AI, cs.CL, cs.LG
Existing language and vision models achieve impressive performance in image-text understanding. Yet, it is an open question to what extent they can be used for language understanding in 3D environments and whether they implicitly acquire 3D object knowledge, e.g. about different views of an object. In this paper, we in...
http://arxiv.org/pdf/2302.10282v1
2023-02-13T15:18:27
2023-02-13T15:18:27
{ "title": [ -0.017288208, -0.016281128, -0.021087646, -0.034118652, -0.03668213, -0.013038635, -0.053955078, -0.016983032, 0.009757996, 0.027679443, -0.054870605, -0.01121521, 0.02027893, -0.021881104, -0.01386261, -0.026443481, 0.017532349, 0.0...
http://arxiv.org/abs/2302.06426v2
Linguistic ambiguity analysis in ChatGPT
Miguel Ortega-Martín, Óscar García-Sierra, Alfonso Ardoiz, Jorge Álvarez, Juan Carlos Armenteros, Adrián Alonso
cs.CL, cs.AI
Linguistic ambiguity is and has always been one of the main challenges in Natural Language Processing (NLP) systems. Modern Transformer architectures like BERT, T5 or more recently InstructGPT have achieved some impressive improvements in many NLP fields, but there is still plenty of work to do. Motivated by the uproar...
http://arxiv.org/pdf/2302.06426v2
2023-02-20T08:55:39
2023-02-13T15:03:07
{ "title": [ -0.043121338, -0.012939453, -0.01940918, -0.011703491, -0.0109939575, 0.019104004, -0.034576416, -0.0158844, -0.01020813, 0.028366089, -0.0034179688, 0.011833191, 0.0069732666, -0.022216797, -0.02758789, 0.018188477, -0.028457642, 0....
http://arxiv.org/abs/2302.06397v2
Type-Aware Decomposed Framework for Few-Shot Named Entity Recognition
Yongqi Li, Yu Yu, Tieyun Qian
cs.CL
Despite the recent success achieved by several two-stage prototypical networks in few-shot named entity recognition (NER) task, the overdetected false spans at the span detection stage and the inaccurate and unstable prototypes at the type classification stage remain to be challenging problems. In this paper, we propos...
http://arxiv.org/pdf/2302.06397v2
2023-10-16T12:31:12
2023-02-13T14:36:21
{ "title": [ -0.0034694672, -0.02154541, -0.027893066, -0.064086914, 0.00044035912, 0.0074539185, -0.03543091, 0.02796936, 0.023406982, 0.012390137, -0.03930664, 0.008110046, -0.00073194504, -0.0022945404, -0.03692627, -0.005619049, 0.0028972626, ...
http://arxiv.org/abs/2302.06341v1
Text2shape Deep Retrieval Model: Generating Initial Cases for Mechanical Part Redesign under the Context of Case-Based Reasoning
Tianshuo Zang, Maolin Yang, Wentao Yong, Pingyu Jiang
cs.AI, cs.CL, cs.IR
Retrieving the similar solutions from the historical case base for new design requirements is the first step in mechanical part redesign under the context of case-based reasoning. However, the manual retrieving method has the problem of low efficiency when the case base is large. Additionally, it is difficult for simpl...
http://arxiv.org/pdf/2302.06341v1
2023-02-13T13:24:11
2023-02-13T13:24:11
{ "title": [ -0.031951904, -0.0041046143, -0.045532227, -0.0090789795, -0.02671814, 0.0028076172, -0.056152344, -0.023956299, -0.016555786, 0.034606934, -0.016357422, 0.014556885, 0.022384644, -0.02217102, 0.00043034554, 0.0076141357, 0.007835388, ...
http://arxiv.org/abs/2302.06337v3
Learning from Noisy Crowd Labels with Logics
Zhijun Chen, Hailong Sun, Haoqian He, Pengpeng Chen
cs.LG, cs.AI, cs.CL, cs.HC
This paper explores the integration of symbolic logic knowledge into deep neural networks for learning from noisy crowd labels. We introduce Logic-guided Learning from Noisy Crowd Labels (Logic-LNCL), an EM-alike iterative logic knowledge distillation framework that learns from both noisy labeled data and logic rules o...
http://arxiv.org/pdf/2302.06337v3
2023-03-19T13:01:47
2023-02-13T13:14:23
{ "title": [ -0.0049095154, 0.024993896, -0.039764404, -0.062683105, 0.025802612, -0.007030487, -0.009124756, 0.004890442, -0.0010490417, 0.016708374, -0.06933594, 0.03253174, -0.02696228, 0.012664795, -0.025100708, -0.019165039, 0.01008606, 0.02...
http://arxiv.org/abs/2302.06321v2
Parameter-efficient Modularised Bias Mitigation via AdapterFusion
Deepak Kumar, Oleg Lesota, George Zerveas, Daniel Cohen, Carsten Eickhoff, Markus Schedl, Navid Rekabsaz
cs.CL, cs.AI
Large pre-trained language models contain societal biases and carry along these biases to downstream tasks. Current in-processing bias mitigation approaches (like adversarial training) impose debiasing by updating a model's parameters, effectively transferring the model to a new, irreversible debiased state. In this wo...
http://arxiv.org/pdf/2302.06321v2
2023-06-18T20:14:15
2023-02-13T12:39:45
{ "title": [ -0.014656067, -0.007019043, -0.053497314, -0.09625244, 0.020751953, 0.009170532, -0.051635742, 0.014877319, 0.0022602081, 0.059051514, 0.0033512115, -0.021835327, 0.03930664, -0.022018433, -0.05609131, 0.0060653687, -0.0059165955, 0....
http://arxiv.org/abs/2302.06198v3
Distinguishability Calibration to In-Context Learning
Hongjing Li, Hanqi Yan, Yanran Li, Li Qian, Yulan He, Lin Gui
cs.CL
Recent years have witnessed increasing interests in prompt-based learning in which models can be trained on only a few annotated instances, making them suitable in low-resource settings. When using prompt-based learning for text classification, the goal is to use a pre-trained language model (PLM) to predict a missing ...
http://arxiv.org/pdf/2302.06198v3
2023-05-10T09:16:53
2023-02-13T09:15:00
{ "title": [ -0.0025577545, 0.030075073, -0.05987549, -0.049194336, 0.027389526, -0.019470215, -0.0077285767, -0.02003479, -0.003627777, 0.0048828125, -0.04660034, 0.009521484, -0.033966064, -0.02659607, -0.012275696, -0.028640747, -0.005683899, ...
http://arxiv.org/abs/2302.06174v1
Evaluation of Word Embeddings for the Social Sciences
Ricardo Schiffers, Dagmar Kern, Daniel Hienert
cs.CL
Word embeddings are an essential instrument in many NLP tasks. Most available resources are trained on general language from Web corpora or Wikipedia dumps. However, word embeddings for domain-specific language are rare, in particular for the social science domain. Therefore, in this work, we describe the creation and ...
http://arxiv.org/pdf/2302.06174v1
2023-02-13T08:23:03
2023-02-13T08:23:03
{ "title": [ 0.0001348257, -0.008682251, -0.022949219, -0.036712646, 0.010040283, -0.0049476624, -0.020095825, -0.01940918, 0.008651733, 0.02508545, -0.022521973, 0.023468018, 0.0008459091, -0.007896423, -0.037261963, 0.01838684, -0.012886047, 0....
http://arxiv.org/abs/2302.06155v1
Identifying Semantically Difficult Samples to Improve Text Classification
Shashank Mujumdar, Stuti Mehta, Hima Patel, Suman Mitra
cs.CL, cs.AI
In this paper, we investigate the effect of addressing difficult samples from a given text dataset on the downstream text classification task. We define difficult samples as being non-obvious cases for text classification by analysing them in the semantic embedding space; specifically - (i) semantically similar samples...
http://arxiv.org/pdf/2302.06155v1
2023-02-13T07:33:46
2023-02-13T07:33:46
{ "title": [ -0.005935669, 0.0038700104, -0.027557373, -0.028717041, 0.0129776, -0.010536194, -0.014526367, -0.03164673, 0.019424438, 0.020065308, -0.046783447, 0.036102295, -0.02545166, 0.015930176, -0.05834961, -0.006340027, -0.024276733, 0.063...
http://arxiv.org/abs/2302.06132v3
NNKGC: Improving Knowledge Graph Completion with Node Neighborhoods
Irene Li, Boming Yang
cs.CL
Knowledge graph completion (KGC) aims to discover missing relations of query entities. Current text-based models utilize the entity name and description to infer the tail entity given the head entity and a certain relation. Existing approaches also consider the neighborhood of the head entity. However, these methods te...
http://arxiv.org/pdf/2302.06132v3
2023-10-19T07:47:44
2023-02-13T06:38:25
{ "title": [ -0.038116455, -0.023773193, -0.06982422, -0.08929443, 0.0015659332, 0.017974854, -0.06903076, 0.018218994, 0.005393982, 0.01209259, -0.037841797, 0.03286743, -0.008224487, 0.0010032654, -0.017623901, 0.004333496, 0.0018053055, 0.0315...
http://arxiv.org/abs/2302.06100v2
Can GPT-3 Perform Statutory Reasoning?
Andrew Blair-Stanek, Nils Holzenberger, Benjamin Van Durme
cs.CL, cs.AI
Statutory reasoning is the task of reasoning with facts and statutes, which are rules written in natural language by a legislature. It is a basic legal skill. In this paper we explore the capabilities of the most capable GPT-3 model, text-davinci-003, on an established statutory-reasoning dataset called SARA. We consid...
http://arxiv.org/pdf/2302.06100v2
2023-05-10T19:17:23
2023-02-13T04:56:11
{ "title": [ 0.02835083, 0.019241333, -0.015594482, -0.034423828, -0.0137786865, -0.014335632, -0.04269409, -0.021377563, -0.0035743713, 0.0040245056, -0.050048828, -0.013381958, -0.026519775, -0.017654419, 0.009567261, -0.031799316, -0.004611969, ...
http://arxiv.org/abs/2302.06078v2
NYCU-TWO at Memotion 3: Good Foundation, Good Teacher, then you have Good Meme Analysis
Yu-Chien Tang, Kuang-Da Wang, Ting-Yun Ou, Wen-Chih Peng
cs.CL, cs.CV
This paper presents a robust solution to the Memotion 3.0 Shared Task. The goal of this task is to classify the emotion and the corresponding intensity expressed by memes, which are usually in the form of images with short captions on social media. Understanding the multi-modal features of the given memes will be the k...
http://arxiv.org/pdf/2302.06078v2
2023-02-14T03:53:02
2023-02-13T03:25:37
{ "title": [ 0.05130005, -0.0049819946, 0.012039185, -0.022888184, -0.007709503, -0.025238037, 0.0031604767, -0.02709961, 0.009727478, 0.042388916, 0.00091362, 0.030166626, -0.00081825256, -0.028411865, 0.0010643005, -0.0012378693, 0.0025997162, ...
http://arxiv.org/abs/2302.07727v2
Knowledge Enhanced Semantic Communication Receiver
Bingyan Wang, Rongpeng Li, Jianhang Zhu, Zhifeng Zhao, Honggang Zhang
cs.CL, cs.LG
In recent years, with the rapid development of deep learning and natural language processing technologies, semantic communication has become a topic of great interest in the field of communication. Although existing deep learning-based semantic communication approaches have shown many advantages, they still do not make...
http://arxiv.org/pdf/2302.07727v2
2023-04-15T10:25:24
2023-02-13T01:49:51
{ "title": [ -0.05947876, -0.0059661865, -0.07873535, -0.04437256, -0.01184082, -0.015319824, -0.048034668, 0.003583908, 0.01979065, 0.05947876, 0.0009365082, -0.017440796, 0.049926758, -0.04144287, -0.032684326, 0.035217285, 0.025222778, 0.01757...
http://arxiv.org/abs/2302.06029v1
Emotion Detection in Unfix-length-Context Conversation
Xiaochen Zhang, Daniel Tang
cs.CL, cs.LG
We leverage different context windows when predicting the emotion of different utterances. New modules are included to realize variable-length context: 1) two speaker-aware units, which explicitly model inner- and inter-speaker dependencies to form distilled conversational context, and 2) a top-k normalization layer, w...
http://arxiv.org/pdf/2302.06029v1
2023-02-13T00:06:47
2023-02-13T00:06:47
{ "title": [ -0.013839722, 0.0044288635, -0.045043945, -0.036315918, 0.0061073303, -0.013320923, -0.035247803, -0.014389038, 0.027053833, 0.023406982, -0.022079468, -0.009101868, 0.011230469, -0.02116394, -0.041809082, -0.0020065308, -0.009254456, ...
http://arxiv.org/abs/2304.11079v1
Academic Writing with GPT-3.5: Reflections on Practices, Efficacy and Transparency
Oğuz 'Oz' Buruk
cs.CL, cs.AI, cs.HC, cs.LG
The debate around the use of GPT 3.5 has been a popular topic among academics since the release of ChatGPT. Whilst some have argued for the advantages of GPT 3.5 in enhancing academic writing, others have raised concerns such as plagiarism, the spread of false information, and ecological issues. The need for finding wa...
http://arxiv.org/pdf/2304.11079v1
2023-02-12T22:05:08
2023-02-12T22:05:08
{ "title": [ -0.005367279, 0.037200928, -0.018737793, -0.021133423, 0.011566162, -0.012611389, -0.047027588, -0.062072754, -0.022201538, 0.024261475, -0.011039734, -0.0042686462, 0.018630981, -0.044403076, -0.0074882507, -0.007041931, -0.02180481, ...
http://arxiv.org/abs/2302.06008v1
ASR Bundestag: A Large-Scale political debate dataset in German
Johannes Wirth, René Peinl
cs.CL, cs.AI, cs.LG, cs.SD, eess.AS
We present ASR Bundestag, a dataset for automatic speech recognition in German, consisting of 610 hours of aligned audio-transcript pairs for supervised training as well as 1,038 hours of unlabeled audio snippets for self-supervised learning, based on raw audio data and transcriptions from plenary sessions and committe...
http://arxiv.org/pdf/2302.06008v1
2023-02-12T21:45:18
2023-02-12T21:45:18
{ "title": [ 0.0031013489, -0.0050354004, -0.049041748, -0.04727173, -0.00077676773, -0.025268555, -0.029159546, -0.0051002502, -0.012687683, 0.040527344, -0.003868103, 0.0052337646, 0.024734497, -0.03010559, -0.019378662, 0.0005221367, 0.017501831, ...
http://arxiv.org/abs/2302.10281v1
LiT Tuned Models for Efficient Species Detection
Andre Nakkab, Benjamin Feuer, Chinmay Hegde
cs.CV, cs.AI, cs.CL
Recent advances in training vision-language models have demonstrated unprecedented robustness and transfer learning effectiveness; however, standard computer vision datasets are image-only, and therefore not well adapted to such training methods. Our paper introduces a simple methodology for adapting any fine-grained i...
http://arxiv.org/pdf/2302.10281v1
2023-02-12T20:36:55
2023-02-12T20:36:55
{ "title": [ -0.02520752, -0.0022964478, -0.02507019, -0.02796936, -0.0037269592, -0.013313293, -0.0073394775, -0.021575928, 0.0060272217, 0.014785767, -0.03555298, -0.004135132, -0.017547607, -0.023742676, -0.03552246, -0.0072746277, 0.011810303, ...
http://arxiv.org/abs/2302.05981v3
MarioGPT: Open-Ended Text2Level Generation through Large Language Models
Shyam Sudhakaran, Miguel González-Duque, Claire Glanois, Matthias Freiberger, Elias Najarro, Sebastian Risi
cs.AI, cs.CL, cs.LG
Procedural Content Generation (PCG) is a technique to generate complex and diverse environments in an automated way. However, while generating content with PCG methods is often straightforward, generating meaningful content that reflects specific intentions and constraints remains challenging. Furthermore, many PCG alg...
http://arxiv.org/pdf/2302.05981v3
2023-11-08T09:43:51
2023-02-12T19:12:24
{ "title": [ -0.021820068, -0.013114929, -0.024124146, -0.065979004, -0.004585266, 0.025131226, -0.068115234, -0.01789856, 0.013442993, 0.04107666, 0.02671814, 0.014801025, 0.027053833, -0.025360107, -0.03186035, 0.04586792, -0.028442383, 0.05798...
http://arxiv.org/abs/2302.07740v1
Team Triple-Check at Factify 2: Parameter-Efficient Large Foundation Models with Feature Representations for Multi-Modal Fact Verification
Wei-Wei Du, Hong-Wei Wu, Wei-Yao Wang, Wen-Chih Peng
cs.CL, cs.AI, cs.CV, cs.LG
Multi-modal fact verification has become an important but challenging issue on social media due to the mismatch between the text and images in the misinformation of news content, which has been addressed by considering cross-modalities to identify the veracity of the news in recent years. In this paper, we propose the ...
http://arxiv.org/pdf/2302.07740v1
2023-02-12T18:08:54
2023-02-12T18:08:54
{ "title": [ -0.016830444, 0.008666992, -0.057159424, -0.115112305, 0.023406982, 0.011138916, -0.08319092, -0.0018138885, 0.010437012, 0.034973145, -0.018585205, 0.02027893, 0.012275696, -0.011436462, -0.0345459, 0.0019302368, -0.026031494, 0.059...
http://arxiv.org/abs/2302.05965v3
RESDSQL: Decoupling Schema Linking and Skeleton Parsing for Text-to-SQL
Haoyang Li, Jing Zhang, Cuiping Li, Hong Chen
cs.CL
One of the recent best attempts at Text-to-SQL is the pre-trained language model. Due to the structural property of the SQL queries, the seq2seq model takes the responsibility of parsing both the schema items (i.e., tables and columns) and the skeleton (i.e., SQL keywords). Such coupled targets increase the difficulty ...
http://arxiv.org/pdf/2302.05965v3
2023-04-10T14:58:49
2023-02-12T17:41:16
{ "title": [ -0.024124146, -0.02835083, -0.04876709, -0.07098389, 0.01033783, 0.008544922, -0.02432251, 0.02192688, 0.016647339, 0.03503418, -0.015113831, 0.0010986328, 0.009887695, -0.023757935, -0.0413208, -0.0023460388, -0.0032310486, 0.026702...
http://arxiv.org/abs/2302.05963v1
Analyzing the Effectiveness of the Underlying Reasoning Tasks in Multi-hop Question Answering
Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, Akiko Aizawa
cs.CL
To explain the predicted answers and evaluate the reasoning abilities of models, several studies have utilized underlying reasoning (UR) tasks in multi-hop question answering (QA) datasets. However, it remains an open question as to how effective UR tasks are for the QA task when training models on both tasks in an end...
http://arxiv.org/pdf/2302.05963v1
2023-02-12T17:32:55
2023-02-12T17:32:55
{ "title": [ 0.0034179688, -0.010650635, -0.04135132, -0.042297363, 0.0019311905, -0.021591187, -0.009849548, 0.016311646, 0.01737976, 0.02809143, -0.029144287, 0.0079422, 0.019180298, -0.009254456, -0.040496826, 0.0032405853, -0.0064430237, 0.02...
http://arxiv.org/abs/2302.05932v1
Stabilized In-Context Learning with Pre-trained Language Models for Few Shot Dialogue State Tracking
Derek Chen, Kun Qian, Zhou Yu
cs.CL
Prompt-based methods with large pre-trained language models (PLMs) have shown impressive unaided performance across many NLP tasks. These models improve even further with the addition of a few labeled in-context exemplars to guide output generation. However, for more complex tasks such as dialogue state tracking (DST),...
http://arxiv.org/pdf/2302.05932v1
2023-02-12T15:05:10
2023-02-12T15:05:10
{ "title": [ 0.021072388, -0.01675415, -0.020095825, -0.054779053, -0.012527466, -0.019302368, -0.022964478, 0.0017004013, -0.008125305, 0.012191772, -0.023727417, 0.013809204, -0.006767273, -0.027801514, -0.0026435852, -0.021575928, -0.00010150671, ...
http://arxiv.org/abs/2302.05913v1
An Extended Sequence Tagging Vocabulary for Grammatical Error Correction
Stuart Mesham, Christopher Bryant, Marek Rei, Zheng Yuan
cs.CL
We extend a current sequence-tagging approach to Grammatical Error Correction (GEC) by introducing specialised tags for spelling correction and morphological inflection using the SymSpell and LemmInflect algorithms. Our approach improves generalisation: the proposed new tagset allows a smaller number of tags to correct...
http://arxiv.org/pdf/2302.05913v1
2023-02-12T13:31:53
2023-02-12T13:31:53
{ "title": [ 0.002401352, -0.001080513, -0.025054932, -0.07501221, 0.014732361, 0.004711151, -0.035827637, 0.002986908, 0.0022239685, 0.0385437, -0.040649414, 0.021591187, -0.000582695, 0.00025248528, -0.048797607, -0.008911133, -0.03253174, 0.04...
http://arxiv.org/abs/2302.05900v1
Investigating the Effect of Relative Positional Embeddings on AMR-to-Text Generation with Structural Adapters
Sebastien Montella, Alexis Nasr, Johannes Heinecke, Frederic Bechet, Lina M. Rojas-Barahona
cs.CL
Text generation from Abstract Meaning Representation (AMR) has substantially benefited from the popularized Pretrained Language Models (PLMs). Myriad approaches have linearized the input graph as a sequence of tokens to fit the PLM tokenization requirements. Nevertheless, this transformation jeopardizes the structural ...
http://arxiv.org/pdf/2302.05900v1
2023-02-12T12:43:36
2023-02-12T12:43:36
{ "title": [ -0.008964539, -0.016342163, -0.012619019, -0.028900146, -0.037628174, 0.0020866394, -0.06390381, -0.017562866, -0.009918213, 0.051208496, 0.0038433075, 0.0084991455, 0.026901245, -0.02130127, -0.027832031, 0.012390137, -0.010444641, ...
http://arxiv.org/abs/2302.05898v1
Entanglement as a Method to Reduce Uncertainty
Diederik Aerts, Jonito Aerts Argëlles, Lester Beltran, Suzette Geriente, Sandro Sozzo
q-bio.NC, cs.CL, quant-ph
In physics, entanglement 'reduces' the entropy of an entity, because the (von Neumann) entropy of, e.g., a composite bipartite entity in a pure entangled state is systematically lower than the entropy of the component sub-entities. We show here that this 'genuinely non-classical reduction of entropy as a result of comp...
http://arxiv.org/pdf/2302.05898v1
2023-02-12T12:18:01
2023-02-12T12:18:01
{ "title": [ -0.0178833, 0.014213562, -0.06317139, -0.02053833, -0.008354187, -0.041015625, 0.010520935, -0.014953613, -0.018310547, 0.032562256, 0.025299072, -0.0034160614, 0.03253174, -0.053527832, 0.022094727, 0.04067993, 0.027282715, 0.017654...
http://arxiv.org/abs/2302.05895v2
Discourse Structure Extraction from Pre-Trained and Fine-Tuned Language Models in Dialogues
Chuyuan Li, Patrick Huber, Wen Xiao, Maxime Amblard, Chloé Braud, Giuseppe Carenini
cs.CL
Discourse processing suffers from data sparsity, especially for dialogues. As a result, we explore approaches to build discourse structures for dialogues, based on attention matrices from Pre-trained Language Models (PLMs). We investigate multiple tasks for fine-tuning and show that the dialogue-tailored Sentence Order...
http://arxiv.org/pdf/2302.05895v2
2023-06-25T10:19:00
2023-02-12T11:26:10
{ "title": [ -0.019744873, -0.013130188, -0.04006958, -0.08929443, 0.002155304, 0.0051651, -0.07342529, 0.018447876, 0.0020256042, 0.02217102, 0.018051147, -0.0043411255, -0.013198853, -0.02420044, -0.04977417, 0.01134491, -0.008216858, 0.0163116...
http://arxiv.org/abs/2302.05892v1
TextDefense: Adversarial Text Detection based on Word Importance Entropy
Lujia Shen, Xuhong Zhang, Shouling Ji, Yuwen Pu, Chunpeng Ge, Xing Yang, Yanghe Feng
cs.CL, cs.AI, cs.CR
Currently, natural language processing (NLP) models are wildly used in various scenarios. However, NLP models, like all deep models, are vulnerable to adversarially generated text. Numerous works have been working on mitigating the vulnerability from adversarial attacks. Nevertheless, there is no comprehensive defense ...
http://arxiv.org/pdf/2302.05892v1
2023-02-12T11:12:44
2023-02-12T11:12:44
{ "title": [ -0.027893066, 0.0050201416, -0.0423584, -0.018615723, 0.011543274, -0.008773804, -0.008544922, -0.018417358, -0.014732361, 0.056274414, -0.014427185, 0.03277588, -0.003894806, 0.008415222, -0.06951904, -0.0039482117, -0.016281128, 0....
http://arxiv.org/abs/2302.05888v1
Position Matters! Empirical Study of Order Effect in Knowledge-grounded Dialogue
Hsuan Su, Shachi H Kumar, Sahisnu Mazumder, Wenda Chen, Ramesh Manuvinakurike, Eda Okur, Saurav Sahay, Lama Nachman, Shang-Tse Chen, Hung-yi Lee
cs.CL, cs.AI, cs.LG
With the power of large pretrained language models, various research works have integrated knowledge into dialogue systems. The traditional techniques treat knowledge as part of the input sequence for the dialogue system, prepending a set of knowledge statements in front of dialogue history. However, such a mechanism f...
http://arxiv.org/pdf/2302.05888v1
2023-02-12T10:13:00
2023-02-12T10:13:00
{ "title": [ 0.0021858215, -0.000054836273, -0.07543945, -0.066101074, 0.020812988, -0.039733887, -0.02255249, 0.0008087158, -0.014915466, 0.033355713, 0.0062828064, -0.008216858, 0.0057792664, -0.02935791, -0.060424805, -0.013420105, -0.015312195, ...
http://arxiv.org/abs/2302.07728v1
AIDA: Legal Judgment Predictions for Non-Professional Fact Descriptions via Partial-and-Imbalanced Domain Adaptation
Guangyi Xiao, Xinlong Liu, Hao Chen, Jingzhi Guo, Zhiguo Gong
cs.CL, cs.AI
In this paper, we study the problem of legal domain adaptation problem from an imbalanced source domain to a partial target domain. The task aims to improve legal judgment predictions for non-professional fact descriptions. We formulate this task as a partial-and-imbalanced domain adaptation problem. Though deep domain...
http://arxiv.org/pdf/2302.07728v1
2023-02-12T07:04:55
2023-02-12T07:04:55
{ "title": [ -0.016464233, -0.020751953, -0.0335083, -0.048828125, -0.021942139, -0.010276794, -0.03048706, 0.0049209595, 0.017410278, 0.032104492, -0.009437561, 0.0012664795, 0.02658081, -0.0074653625, -0.044128418, 0.012107849, -0.030670166, 0....
http://arxiv.org/abs/2302.05852v1
"Why is this misleading?": Detecting News Headline Hallucinations with Explanations
Jiaming Shen, Jialu Liu, Dan Finnie, Negar Rahmati, Michael Bendersky, Marc Najork
cs.CL, cs.AI, cs.IR
Automatic headline generation enables users to comprehend ongoing news events promptly and has recently become an important task in web mining and natural language processing. With the growing need for news headline generation, we argue that the hallucination issue, namely the generated headlines being not supported by...
http://arxiv.org/pdf/2302.05852v1
2023-02-12T04:21:49
2023-02-12T04:21:49
{ "title": [ -0.006679535, 0.0070877075, -0.0072135925, -0.030975342, 0.007003784, -0.0035076141, -0.00049448013, -0.03463745, -0.019592285, 0.012924194, -0.016464233, 0.02355957, -0.01626587, 0.0002245903, -0.05078125, -0.005783081, -0.022842407, ...
http://arxiv.org/abs/2302.07730v3
Transformer models: an introduction and catalog
Xavier Amatriain, Ananth Sankar, Jie Bing, Praveen Kumar Bodigutla, Timothy J. Hazen, Michaeel Kazi
cs.CL
In the past few years we have seen the meteoric appearance of dozens of foundation models of the Transformer family, all of which have memorable and sometimes funny, but not self-explanatory, names. The goal of this paper is to offer a somewhat comprehensive but simple catalog and classification of the most popular Tra...
http://arxiv.org/pdf/2302.07730v3
2023-05-25T05:42:23
2023-02-12T01:26:49
{ "title": [ -0.00042772293, 0.02243042, -0.024734497, -0.049194336, 0.020553589, -0.0058784485, -0.024810791, -0.041931152, -0.0181427, 0.044525146, -0.0067863464, -0.0055007935, 0.026779175, -0.041046143, -0.016174316, 0.02180481, 0.012962341, ...
http://arxiv.org/abs/2302.05817v2
Level Generation Through Large Language Models
Graham Todd, Sam Earle, Muhammad Umair Nasir, Michael Cerny Green, Julian Togelius
cs.AI, cs.CL, cs.NE
Large Language Models (LLMs) are powerful tools, capable of leveraging their training on natural language to write stories, generate code, and answer questions. But can they generate functional video game levels? Game levels, with their complex functional constraints and spatial relationships in more than one dimension...
http://arxiv.org/pdf/2302.05817v2
2023-06-01T05:43:02
2023-02-11T23:34:42
{ "title": [ -0.03353882, -0.016448975, -0.030776978, -0.07519531, -0.008155823, 0.03173828, -0.07019043, -0.001953125, 0.015304565, 0.01979065, 0.014343262, -0.023254395, -0.012069702, -0.041137695, -0.034729004, 0.019515991, -0.00995636, 0.0322...
http://arxiv.org/abs/2302.05759v1
Improving Sign Recognition with Phonology
Lee Kezar, Jesse Thomason, Zed Sevcikova Sehyr
cs.CL, cs.CV, I.2.7; I.2.10; I.5.4
We use insights from research on American Sign Language (ASL) phonology to train models for isolated sign language recognition (ISLR), a step towards automatic sign language understanding. Our key insight is to explicitly recognize the role of phonology in sign production to achieve more accurate ISLR than existing wor...
http://arxiv.org/pdf/2302.05759v1
2023-02-11T18:51:23
2023-02-11T18:51:23
{ "title": [ -0.0022087097, 0.010383606, -0.014518738, -0.01979065, -0.041381836, -0.008277893, -0.028213501, -0.006713867, -0.005466461, 0.033996582, -0.05960083, -0.014663696, -0.01689148, -0.022598267, -0.019439697, -0.024932861, 0.0011196136, ...