id
stringlengths
33
34
title
stringlengths
6
218
authors
stringlengths
5
6.98k
categories
stringlengths
5
188
summary
stringlengths
34
3.26k
link_pdf
stringlengths
33
34
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/2310.15941v1
This is not a Dataset: A Large Negation Benchmark to Challenge Large Language Models
Iker García-Ferrero, Begoña Altuna, Javier Álvez, Itziar Gonzalez-Dios, German Rigau
cs.CL
Although large language models (LLMs) have apparently acquired a certain level of grammatical knowledge and the ability to make generalizations, they fail to interpret negation, a crucial step in Natural Language Processing. We try to clarify the reasons for the sub-optimal performance of LLMs understanding negation. W...
http://arxiv.org/pdf/2310.15941v1
2023-10-24T15:38:21
2023-10-24T15:38:21
{ "title": [ -0.01651001, -0.01461792, -0.017868042, -0.06109619, -0.0022525787, -0.003419876, -0.04663086, -0.010795593, -0.00096940994, 0.02355957, -0.029083252, 0.016067505, -0.0020656586, -0.00957489, -0.0385437, -0.0013160706, -0.0093307495, ...
http://arxiv.org/abs/2310.15929v1
E-Sparse: Boosting the Large Language Model Inference through Entropy-based N:M Sparsity
Yun Li, Lin Niu, Xipeng Zhang, Kai Liu, Jianchen Zhu, Zhanhui Kang
cs.LG, cs.AI, cs.CL
Traditional pruning methods are known to be challenging to work in Large Language Models (LLMs) for Generative AI because of their unaffordable training process and large computational demands. For the first time, we introduce the information entropy of hidden state features into a pruning metric design, namely E-Spars...
http://arxiv.org/pdf/2310.15929v1
2023-10-24T15:27:15
2023-10-24T15:27:15
{ "title": [ -0.05670166, -0.020339966, -0.07348633, -0.074645996, -0.0028076172, 0.0059432983, -0.084228516, 0.029800415, 0.020217896, 0.027496338, -0.01499176, 0.0011348724, 0.030029297, -0.022918701, -0.03656006, 0.016220093, 0.0018835068, 0.0...
http://arxiv.org/abs/2310.15921v1
Contrastive Learning-based Sentence Encoders Implicitly Weight Informative Words
Hiroto Kurita, Goro Kobayashi, Sho Yokoi, Kentaro Inui
cs.CL
The performance of sentence encoders can be significantly improved through the simple practice of fine-tuning using contrastive loss. A natural question arises: what characteristics do models acquire during contrastive learning? This paper theoretically and experimentally shows that contrastive-based sentence encoders ...
http://arxiv.org/pdf/2310.15921v1
2023-10-24T15:22:04
2023-10-24T15:22:04
{ "title": [ -0.003276825, 0.008636475, -0.025527954, -0.04083252, -0.01802063, -0.020828247, -0.0209198, -0.016220093, 0.020736694, 0.039733887, -0.056915283, 0.011352539, 0.009124756, -0.0038909912, -0.029541016, -0.004989624, -0.0041542053, 0....
http://arxiv.org/abs/2310.15916v1
In-Context Learning Creates Task Vectors
Roee Hendel, Mor Geva, Amir Globerson
cs.CL
In-context learning (ICL) in Large Language Models (LLMs) has emerged as a powerful new learning paradigm. However, its underlying mechanism is still not well understood. In particular, it is challenging to map it to the "standard" machine learning framework, where one uses a training set $S$ to find a best-fitting fun...
http://arxiv.org/pdf/2310.15916v1
2023-10-24T15:17:14
2023-10-24T15:17:14
{ "title": [ 0.06390381, -0.0045433044, -0.00932312, -0.027008057, -0.035308838, -0.039276123, -0.007545471, -0.014511108, -0.06506348, 0.039367676, -0.021499634, 0.014503479, -0.008049011, -0.020141602, 0.038269043, -0.027923584, 0.03265381, 0.0...
http://arxiv.org/abs/2310.15910v1
Characterizing Mechanisms for Factual Recall in Language Models
Qinan Yu, Jack Merullo, Ellie Pavlick
cs.CL, cs.AI
Language Models (LMs) often must integrate facts they memorized in pretraining with new information that appears in a given context. These two sources can disagree, causing competition within the model, and it is unclear how an LM will resolve the conflict. On a dataset that queries for knowledge of world capitals, we ...
http://arxiv.org/pdf/2310.15910v1
2023-10-24T15:15:18
2023-10-24T15:15:18
{ "title": [ -0.008377075, -0.008003235, -0.031188965, -0.064575195, -0.009506226, -0.013221741, -0.037994385, 0.0077171326, 0.0059776306, 0.048858643, -0.0031909943, 0.00932312, 0.013000488, -0.004306793, -0.06555176, 0.031799316, -0.014968872, ...
http://arxiv.org/abs/2310.15905v1
Is Probing All You Need? Indicator Tasks as an Alternative to Probing Embedding Spaces
Tal Levy, Omer Goldman, Reut Tsarfaty
cs.CL, cs.AI, cs.LG
The ability to identify and control different kinds of linguistic information encoded in vector representations of words has many use cases, especially for explainability and bias removal. This is usually done via a set of simple classification tasks, termed probes, to evaluate the information encoded in the embedding ...
http://arxiv.org/pdf/2310.15905v1
2023-10-24T15:08:12
2023-10-24T15:08:12
{ "title": [ 0.010429382, -0.019073486, 0.011474609, 0.0038986206, -0.047851562, -0.017150879, 0.013755798, -0.02558899, -0.011734009, 0.0073776245, -0.02243042, -0.04019165, -0.0006432533, -0.060302734, 0.007659912, -0.060394287, 0.032958984, -0...
http://arxiv.org/abs/2310.15904v1
Do Stochastic Parrots have Feelings Too? Improving Neural Detection of Synthetic Text via Emotion Recognition
Alan Cowap, Yvette Graham, Jennifer Foster
cs.CL, cs.LG
Recent developments in generative AI have shone a spotlight on high-performance synthetic text generation technologies. The now wide availability and ease of use of such models highlights the urgent need to provide equally powerful technologies capable of identifying synthetic text. With this in mind, we draw inspirati...
http://arxiv.org/pdf/2310.15904v1
2023-10-24T15:07:35
2023-10-24T15:07:35
{ "title": [ -0.025482178, -0.00059223175, -0.020050049, -0.02229309, -0.0075798035, -0.007217407, -0.021774292, -0.009307861, 0.014831543, 0.019638062, -0.04574585, 0.0006804466, 0.00356102, -0.016555786, -0.0473938, -0.013717651, -0.0056648254, ...
http://arxiv.org/abs/2310.15896v1
BianQue: Balancing the Questioning and Suggestion Ability of Health LLMs with Multi-turn Health Conversations Polished by ChatGPT
Yirong Chen, Zhenyu Wang, Xiaofen Xing, huimin zheng, Zhipei Xu, Kai Fang, Junhong Wang, Sihang Li, Jieling Wu, Qi Liu, Xiangmin Xu
cs.CL, cs.HC
Large language models (LLMs) have performed well in providing general and extensive health suggestions in single-turn conversations, exemplified by systems such as ChatGPT, ChatGLM, ChatDoctor, DoctorGLM, and etc. However, the limited information provided by users during single turn results in inadequate personalizatio...
http://arxiv.org/pdf/2310.15896v1
2023-10-24T14:57:34
2023-10-24T14:57:34
{ "title": [ -0.038482666, -0.021881104, -0.030731201, 0.0043907166, -0.037506104, 0.004940033, -0.029769897, -0.005203247, -0.023971558, 0.014259338, -0.033325195, 0.031433105, 0.008598328, -0.039611816, 0.040161133, -0.039794922, -0.02859497, 0...
http://arxiv.org/abs/2310.18363v1
A Contextualized Real-Time Multimodal Emotion Recognition for Conversational Agents using Graph Convolutional Networks in Reinforcement Learning
Fathima Abdul Rahman, Guang Lu
cs.CL, cs.HC, cs.LG, I.2.10
Owing to the recent developments in Generative Artificial Intelligence (GenAI) and Large Language Models (LLM), conversational agents are becoming increasingly popular and accepted. They provide a human touch by interacting in ways familiar to us and by providing support as virtual companions. Therefore, it is importan...
http://arxiv.org/pdf/2310.18363v1
2023-10-24T14:31:17
2023-10-24T14:31:17
{ "title": [ -0.00082731247, -0.0042266846, -0.034210205, -0.03527832, -0.0124435425, -0.022644043, -0.030288696, 0.000014126301, 0.013008118, 0.024719238, -0.02204895, -0.006134033, 0.016082764, -0.03363037, -0.019851685, -0.007904053, -0.012367249,...
http://arxiv.org/abs/2310.18362v1
SoK: Memorization in General-Purpose Large Language Models
Valentin Hartmann, Anshuman Suri, Vincent Bindschaedler, David Evans, Shruti Tople, Robert West
cs.CL, cs.CR, cs.LG
Large Language Models (LLMs) are advancing at a remarkable pace, with myriad applications under development. Unlike most earlier machine learning models, they are no longer built for one specific application but are designed to excel in a wide range of tasks. A major part of this success is due to their huge training d...
http://arxiv.org/pdf/2310.18362v1
2023-10-24T14:25:53
2023-10-24T14:25:53
{ "title": [ -0.037109375, -0.011642456, -0.03564453, -0.07476807, -0.0031547546, 0.014526367, -0.05822754, 0.013496399, 0.007987976, 0.044158936, -0.008834839, 0.005607605, 0.019607544, -0.016677856, -0.045410156, 0.013961792, -0.015716553, 0.02...
http://arxiv.org/abs/2310.15852v1
Using Artificial French Data to Understand the Emergence of Gender Bias in Transformer Language Models
Lina Conti, Guillaume Wisniewski
cs.CL, cs.AI
Numerous studies have demonstrated the ability of neural language models to learn various linguistic properties without direct supervision. This work takes an initial step towards exploring the less researched topic of how neural models discover linguistic properties of words, such as gender, as well as the rules gover...
http://arxiv.org/pdf/2310.15852v1
2023-10-24T14:08:37
2023-10-24T14:08:37
{ "title": [ -0.04058838, -0.0005722046, -0.056884766, -0.010505676, -0.009979248, -0.029769897, -0.026535034, -0.0007472038, -0.009864807, 0.037322998, -0.03781128, -0.012962341, 0.046203613, -0.014549255, -0.02218628, -0.014320374, -0.030288696, ...
http://arxiv.org/abs/2310.15851v1
Self-Guard: Empower the LLM to Safeguard Itself
Zezhong Wang, Fangkai Yang, Lu Wang, Pu Zhao, Hongru Wang, Liang Chen, Qingwei Lin, Kam-Fai Wong
cs.CL
The jailbreak attack can bypass the safety measures of a Large Language Model (LLM), generating harmful content. This misuse of LLM has led to negative societal consequences. Currently, there are two main approaches to address jailbreak attacks: safety training and safeguards. Safety training focuses on further trainin...
http://arxiv.org/pdf/2310.15851v1
2023-10-24T14:08:26
2023-10-24T14:08:26
{ "title": [ 0.019042969, 0.00043797493, -0.053375244, -0.033813477, -0.026473999, -0.032958984, -0.005870819, 0.004447937, -0.06781006, 0.0023880005, 0.02243042, 0.022262573, -0.021881104, -0.07373047, 0.043426514, -0.034088135, 0.021530151, 0.0...
http://arxiv.org/abs/2310.15836v1
A Diffusion Weighted Graph Framework for New Intent Discovery
Wenkai Shi, Wenbin An, Feng Tian, Qinghua Zheng, QianYing Wang, Ping Chen
cs.CL, cs.AI, cs.LG
New Intent Discovery (NID) aims to recognize both new and known intents from unlabeled data with the aid of limited labeled data containing only known intents. Without considering structure relationships between samples, previous methods generate noisy supervisory signals which cannot strike a balance between quantity ...
http://arxiv.org/pdf/2310.15836v1
2023-10-24T13:43:01
2023-10-24T13:43:01
{ "title": [ -0.01146698, -0.0031108856, -0.063964844, -0.050323486, 0.015975952, -0.022033691, -0.011734009, 0.0063591003, 0.000060141087, 0.009635925, -0.016998291, 0.023361206, -0.039031982, -0.020889282, -0.022781372, -0.013397217, -0.02078247, ...
http://arxiv.org/abs/2310.15829v1
Unnatural language processing: How do language models handle machine-generated prompts?
Corentin Kervadec, Francesca Franzon, Marco Baroni
cs.CL
Language model prompt optimization research has shown that semantically and grammatically well-formed manually crafted prompts are routinely outperformed by automatically generated token sequences with no apparent meaning or syntactic structure, including sequences of vectors from a model's embedding space. We use mach...
http://arxiv.org/pdf/2310.15829v1
2023-10-24T13:32:20
2023-10-24T13:32:20
{ "title": [ -0.028701782, 0.0029850006, -0.019897461, -0.014808655, -0.010948181, -0.0011301041, -0.029022217, -0.021591187, -0.006454468, 0.018539429, -0.037628174, -0.008918762, 0.012710571, -0.008575439, -0.040771484, -0.0046539307, -0.012207031,...
http://arxiv.org/abs/2310.15823v2
Rosetta Stone at the Arabic Reverse Dictionary Shared Task: A Hop From Language Modeling To Word--Definition Alignment
Ahmed ElBakry, Mohamed Gabr, Muhammad ElNokrashy, Badr AlKhamissi
cs.CL, cs.AI
A Reverse Dictionary is a tool enabling users to discover a word based on its provided definition, meaning, or description. Such a technique proves valuable in various scenarios, aiding language learners who possess a description of a word without its identity, and benefiting writers seeking precise terminology. These ...
http://arxiv.org/pdf/2310.15823v2
2023-11-10T13:02:40
2023-10-24T13:23:57
{ "title": [ -0.051116943, -0.008369446, -0.049743652, 0.022949219, -0.031341553, -0.028533936, -0.0051498413, 0.0054359436, -0.016448975, 0.054107666, -0.010726929, -0.006313324, 0.03781128, -0.021209717, -0.008636475, -0.013343811, 0.02468872, ...
http://arxiv.org/abs/2310.15819v1
Generative Language Models Exhibit Social Identity Biases
Tiancheng Hu, Yara Kyrychenko, Steve Rathje, Nigel Collier, Sander van der Linden, Jon Roozenbeek
cs.CL, cs.CY
The surge in popularity of large language models has given rise to concerns about biases that these models could learn from humans. In this study, we investigate whether ingroup solidarity and outgroup hostility, fundamental social biases known from social science, are present in 51 large language models. We find that ...
http://arxiv.org/pdf/2310.15819v1
2023-10-24T13:17:40
2023-10-24T13:17:40
{ "title": [ 0.020935059, 0.00472641, -0.031234741, -0.02017212, -0.049438477, -0.034729004, -0.019454956, 0.0055770874, -0.019699097, 0.05014038, -0.025222778, -0.010795593, 0.033599854, -0.032318115, 0.010887146, -0.004360199, -0.014022827, 0.0...
http://arxiv.org/abs/2310.15799v1
DALE: Generative Data Augmentation for Low-Resource Legal NLP
Sreyan Ghosh, Chandra Kiran Evuru, Sonal Kumar, S Ramaneswaran, S Sakshi, Utkarsh Tyagi, Dinesh Manocha
cs.CL, cs.AI
We present DALE, a novel and effective generative Data Augmentation framework for low-resource LEgal NLP. DALE addresses the challenges existing frameworks pose in generating effective data augmentations of legal documents - legal language, with its specialized vocabulary and complex semantics, morphology, and syntax, ...
http://arxiv.org/pdf/2310.15799v1
2023-10-24T12:50:28
2023-10-24T12:50:28
{ "title": [ -0.020019531, -0.035949707, -0.061767578, -0.07849121, -0.0048065186, 0.0034332275, -0.058532715, 0.032470703, 0.018844604, 0.01625061, -0.060516357, 0.031280518, 0.0024414062, 0.007091522, -0.03552246, 0.00554657, -0.031921387, 0.07...
http://arxiv.org/abs/2310.15797v1
Random Entity Quantization for Parameter-Efficient Compositional Knowledge Graph Representation
Jiaang Li, Quan Wang, Yi Liu, Licheng Zhang, Zhendong Mao
cs.AI, cs.CL, cs.LG
Representation Learning on Knowledge Graphs (KGs) is essential for downstream tasks. The dominant approach, KG Embedding (KGE), represents entities with independent vectors and faces the scalability challenge. Recent studies propose an alternative way for parameter efficiency, which represents entities by composing ent...
http://arxiv.org/pdf/2310.15797v1
2023-10-24T12:48:52
2023-10-24T12:48:52
{ "title": [ -0.034606934, -0.005508423, -0.07562256, -0.087524414, 0.009170532, 0.003376007, -0.07513428, 0.01651001, 0.00440979, 0.043518066, 0.020858765, 0.003917694, 0.010696411, -0.030731201, -0.029907227, 0.018539429, -0.00030493736, 0.0512...
http://arxiv.org/abs/2310.15793v1
Improving generalization in large language models by learning prefix subspaces
Louis Falissard, Vincent Guigue, Laure Soulier
cs.LG, cs.AI, cs.CL
This article focuses on large language models (LLMs) fine-tuning in the scarce data regime (also known as the "few-shot" learning setting). We propose a method to increase the generalization capabilities of LLMs based on neural network subspaces. This optimization method, recently introduced in computer vision, aims to...
http://arxiv.org/pdf/2310.15793v1
2023-10-24T12:44:09
2023-10-24T12:44:09
{ "title": [ -0.038360596, -0.010894775, -0.036346436, -0.04916382, -0.010032654, 0.0096206665, -0.055755615, 0.0118637085, 0.00048446655, 0.031463623, -0.028747559, 0.016952515, -0.0057411194, -0.0062828064, -0.039276123, 0.00015282631, -0.025238037...
http://arxiv.org/abs/2310.18360v1
Guiding LLM to Fool Itself: Automatically Manipulating Machine Reading Comprehension Shortcut Triggers
Mosh Levy, Shauli Ravfogel, Yoav Goldberg
cs.CL, cs.AI
Recent applications of LLMs in Machine Reading Comprehension (MRC) systems have shown impressive results, but the use of shortcuts, mechanisms triggered by features spuriously correlated to the true label, has emerged as a potential threat to their reliability. We analyze the problem from two angles: LLMs as editors, g...
http://arxiv.org/pdf/2310.18360v1
2023-10-24T12:37:06
2023-10-24T12:37:06
{ "title": [ -0.051208496, -0.018081665, -0.010559082, 0.017929077, -0.072387695, -0.009056091, -0.0395813, -0.0435791, -0.008850098, 0.04724121, -0.032836914, 0.0076065063, 0.02192688, -0.028213501, -0.023406982, -0.0129470825, -0.0005121231, 0....
http://arxiv.org/abs/2310.15777v2
MindLLM: Pre-training Lightweight Large Language Model from Scratch, Evaluations and Domain Applications
Yizhe Yang, Huashan Sun, Jiawei Li, Runheng Liu, Yinghao Li, Yuhang Liu, Heyan Huang, Yang Gao
cs.CL, cs.AI
Large Language Models (LLMs) have demonstrated remarkable performance across various natural language tasks, marking significant strides towards general artificial intelligence. While general artificial intelligence is leveraged by developing increasingly large-scale models, there could be another branch to develop lig...
http://arxiv.org/pdf/2310.15777v2
2023-10-29T01:17:53
2023-10-24T12:22:34
{ "title": [ -0.028442383, -0.024002075, -0.045959473, -0.078063965, -0.0014572144, 0.0044898987, -0.06518555, 0.0070266724, -0.010932922, 0.021347046, 0.007507324, 0.0023937225, 0.009185791, -0.033721924, -0.016662598, 0.013504028, -0.013153076, ...
http://arxiv.org/abs/2310.15773v1
BLESS: Benchmarking Large Language Models on Sentence Simplification
Tannon Kew, Alison Chi, Laura Vásquez-Rodríguez, Sweta Agrawal, Dennis Aumiller, Fernando Alva-Manchego, Matthew Shardlow
cs.CL
We present BLESS, a comprehensive performance benchmark of the most recent state-of-the-art large language models (LLMs) on the task of text simplification (TS). We examine how well off-the-shelf LLMs can solve this challenging task, assessing a total of 44 models, differing in size, architecture, pre-training methods,...
http://arxiv.org/pdf/2310.15773v1
2023-10-24T12:18:17
2023-10-24T12:18:17
{ "title": [ -0.034179688, -0.011871338, -0.037597656, -0.07342529, -0.0014734268, 0.01940918, -0.08441162, -0.0018558502, 0.0073280334, 0.042541504, 0.0005569458, 0.025436401, 0.00737381, -0.012290955, -0.052825928, 0.021377563, -0.0033931732, 0...
http://arxiv.org/abs/2310.15758v1
Learning From Free-Text Human Feedback -- Collect New Datasets Or Extend Existing Ones?
Dominic Petrak, Nafise Sadat Moosavi, Ye Tian, Nikolai Rozanov, Iryna Gurevych
cs.CL
Learning from free-text human feedback is essential for dialog systems, but annotated data is scarce and usually covers only a small fraction of error types known in conversational AI. Instead of collecting and annotating new datasets from scratch, recent advances in synthetic dialog generation could be used to augment...
http://arxiv.org/pdf/2310.15758v1
2023-10-24T12:01:11
2023-10-24T12:01:11
{ "title": [ 0.00010740757, -0.0015325546, -0.011817932, -0.030853271, -0.017456055, -0.01486969, -0.030014038, -0.03640747, -0.015670776, 0.021774292, -0.055358887, 0.02017212, -0.05053711, 0.0029754639, -0.05419922, -0.04849243, 0.0234375, 0.02...
http://arxiv.org/abs/2310.15757v1
Do Differences in Values Influence Disagreements in Online Discussions?
Michiel van der Meer, Piek Vossen, Catholijn M. Jonker, Pradeep K. Murukannaiah
cs.CL
Disagreements are common in online discussions. Disagreement may foster collaboration and improve the quality of a discussion under some conditions. Although there exist methods for recognizing disagreement, a deeper understanding of factors that influence disagreement is lacking in the literature. We investigate a hyp...
http://arxiv.org/pdf/2310.15757v1
2023-10-24T12:00:59
2023-10-24T12:00:59
{ "title": [ 0.022613525, 0.015487671, -0.057495117, -0.011772156, 0.0096206665, -0.045562744, 0.025268555, -0.0014791489, -0.027114868, 0.005050659, -0.0044937134, -0.012046814, -0.009559631, -0.049468994, 0.02015686, -0.042419434, 0.0059661865, ...
http://arxiv.org/abs/2310.15752v1
Integrating Language Models into Direct Speech Translation: An Inference-Time Solution to Control Gender Inflection
Dennis Fucci, Marco Gaido, Sara Papi, Mauro Cettolo, Matteo Negri, Luisa Bentivogli
cs.CL, cs.AI
When translating words referring to the speaker, speech translation (ST) systems should not resort to default masculine generics nor rely on potentially misleading vocal traits. Rather, they should assign gender according to the speakers' preference. The existing solutions to do so, though effective, are hardly feasibl...
http://arxiv.org/pdf/2310.15752v1
2023-10-24T11:55:16
2023-10-24T11:55:16
{ "title": [ -0.030014038, -0.010124207, -0.039520264, -0.078308105, -0.008026123, -0.00066041946, -0.07196045, 0.020324707, 0.0032043457, 0.04812622, -0.013298035, -0.0023651123, 0.040771484, -0.024978638, -0.049987793, 0.0064353943, -0.018295288, ...
http://arxiv.org/abs/2310.15746v1
Failures Pave the Way: Enhancing Large Language Models through Tuning-free Rule Accumulation
Zeyuan Yang, Peng Li, Yang Liu
cs.CL
Large Language Models (LLMs) have showcased impressive performance. However, due to their inability to capture relationships among samples, these frozen LLMs inevitably keep repeating similar mistakes. In this work, we propose our Tuning-free Rule Accumulation (TRAN) framework, which guides LLMs in improving their perf...
http://arxiv.org/pdf/2310.15746v1
2023-10-24T11:40:34
2023-10-24T11:40:34
{ "title": [ 0.010444641, -0.03186035, -0.029571533, -0.13024902, -0.0039405823, 0.012046814, -0.078125, 0.045043945, -0.002998352, 0.03378296, -0.014144897, 0.009689331, 0.022155762, -0.015586853, -0.03515625, 0.015686035, -0.0124053955, 0.05191...
http://arxiv.org/abs/2310.15743v1
RAPL: A Relation-Aware Prototype Learning Approach for Few-Shot Document-Level Relation Extraction
Shiao Meng, Xuming Hu, Aiwei Liu, Shu'ang Li, Fukun Ma, Yawen Yang, Lijie Wen
cs.CL, 68T50, I.2.7
How to identify semantic relations among entities in a document when only a few labeled documents are available? Few-shot document-level relation extraction (FSDLRE) is crucial for addressing the pervasive data scarcity problem in real-world scenarios. Metric-based meta-learning is an effective framework widely adopted...
http://arxiv.org/pdf/2310.15743v1
2023-10-24T11:35:23
2023-10-24T11:35:23
{ "title": [ 0.0033855438, -0.013450623, -0.032806396, -0.07318115, -0.009468079, 0.0046653748, -0.04748535, 0.016021729, 0.027893066, -0.001206398, -0.025497437, 0.0040397644, -0.015808105, -0.0126571655, -0.03253174, -0.01727295, -0.003118515, ...
http://arxiv.org/abs/2310.15724v1
Variator: Accelerating Pre-trained Models with Plug-and-Play Compression Modules
Chaojun Xiao, Yuqi Luo, Wenbin Zhang, Pengle Zhang, Xu Han, Yankai Lin, Zhengyan Zhang, Ruobing Xie, Zhiyuan Liu, Maosong Sun, Jie Zhou
cs.CL
Pre-trained language models (PLMs) have achieved remarkable results on NLP tasks but at the expense of huge parameter sizes and the consequent computational costs. In this paper, we propose Variator, a parameter-efficient acceleration method that enhances computational efficiency through plug-and-play compression plugi...
http://arxiv.org/pdf/2310.15724v1
2023-10-24T11:00:07
2023-10-24T11:00:07
{ "title": [ -0.015075684, -0.011192322, -0.05303955, -0.08300781, 0.018600464, -0.00084638596, -0.048309326, -0.0047416687, -0.004802704, 0.019927979, 0.000018298626, 0.01008606, -0.018432617, -0.03201294, -0.020965576, 0.009391785, 0.021484375, ...
http://arxiv.org/abs/2310.15722v1
Re-Temp: Relation-Aware Temporal Representation Learning for Temporal Knowledge Graph Completion
Kunze Wang, Soyeon Caren Han, Josiah Poon
cs.CL
Temporal Knowledge Graph Completion (TKGC) under the extrapolation setting aims to predict the missing entity from a fact in the future, posing a challenge that aligns more closely with real-world prediction problems. Existing research mostly encodes entities and relations using sequential graph neural networks applied...
http://arxiv.org/pdf/2310.15722v1
2023-10-24T10:58:33
2023-10-24T10:58:33
{ "title": [ 0.011634827, -0.01637268, -0.075805664, -0.12298584, 0.016830444, -0.010261536, -0.057922363, 0.046722412, 0.020431519, 0.038635254, -0.025634766, 0.02041626, 0.010505676, 0.0016708374, -0.049346924, 0.0066566467, -0.0029144287, 0.03...
http://arxiv.org/abs/2310.15720v2
Ensemble of Task-Specific Language Models for Brain Encoding
Arvindh Arun, Jerrin John, Sanjai Kumaran
cs.CL, cs.NE
Language models have been shown to be rich enough to encode fMRI activations of certain Regions of Interest in our Brains. Previous works have explored transfer learning from representations learned for popular natural language processing tasks for predicting brain responses. In our work, we improve the performance of ...
http://arxiv.org/pdf/2310.15720v2
2023-11-09T07:03:54
2023-10-24T10:52:41
{ "title": [ -0.035614014, 0.0016841888, -0.04135132, -0.016311646, -0.009605408, -0.009765625, -0.042053223, -0.01133728, 0.005355835, 0.047180176, 0.018692017, -0.01776123, 0.025634766, -0.037261963, -0.05078125, 0.040252686, -0.00087976456, 0....
http://arxiv.org/abs/2310.15702v1
Enhancing Biomedical Lay Summarisation with External Knowledge Graphs
Tomas Goldsack, Zhihao Zhang, Chen Tang, Carolina Scarton, Chenghua Lin
cs.CL
Previous approaches for automatic lay summarisation are exclusively reliant on the source article that, given it is written for a technical audience (e.g., researchers), is unlikely to explicitly define all technical concepts or state all of the background information that is relevant for a lay audience. We address thi...
http://arxiv.org/pdf/2310.15702v1
2023-10-24T10:25:21
2023-10-24T10:25:21
{ "title": [ -0.01965332, -0.015045166, -0.044006348, -0.100097656, -0.018112183, 0.014251709, -0.07318115, 0.021499634, -0.007850647, 0.039978027, -0.0047569275, 0.012161255, -0.008682251, -0.0079574585, -0.042785645, -0.000770092, 0.00831604, 0...
http://arxiv.org/abs/2310.15694v4
COPF: Continual Learning Human Preference through Optimal Policy Fitting
Han Zhang, Lin Gui, Yuanzhao Zhai, Hui Wang, Yu Lei, Ruifeng Xu
cs.LG, cs.CL
The technique of Reinforcement Learning from Human Feedback (RLHF) is a commonly employed method to improve pre-trained Language Models (LM), enhancing their ability to conform to human preferences. Nevertheless, the current RLHF-based LMs necessitate full retraining each time novel queries or feedback are introduced, ...
http://arxiv.org/pdf/2310.15694v4
2023-10-28T02:00:39
2023-10-24T10:05:32
{ "title": [ 0.010498047, 0.0030269623, -0.012840271, -0.038391113, 0.015914917, -0.018539429, 0.009063721, -0.028381348, -0.027023315, 0.0024032593, -0.028518677, 0.019973755, -0.039093018, -0.020355225, -0.009819031, -0.012176514, -0.0068626404, ...
http://arxiv.org/abs/2310.15693v1
Towards Automated Recipe Genre Classification using Semi-Supervised Learning
Nazmus Sakib, G. M. Shahariar, Md. Mohsinul Kabir, Md. Kamrul Hasan, Hasan Mahmud
cs.CL, cs.AI
Sharing cooking recipes is a great way to exchange culinary ideas and provide instructions for food preparation. However, categorizing raw recipes found online into appropriate food genres can be challenging due to a lack of adequate labeled data. In this study, we present a dataset named the ``Assorted, Archetypal, an...
http://arxiv.org/pdf/2310.15693v1
2023-10-24T10:03:27
2023-10-24T10:03:27
{ "title": [ -0.0115356445, 0.015151978, -0.06359863, -0.045318604, 0.03152466, -0.02633667, -0.02609253, 0.0021076202, -0.011672974, 0.034973145, 0.014625549, -0.0002527237, 0.010063171, -0.03945923, -0.026184082, 0.0073280334, -0.011367798, 0.0...
http://arxiv.org/abs/2310.15689v1
Creating a silver standard for patent simplification
Silvia Casola, Alberto Lavelli, Horacio Saggion
cs.CL
Patents are legal documents that aim at protecting inventions on the one hand and at making technical knowledge circulate on the other. Their complex style -- a mix of legal, technical, and extremely vague language -- makes their content hard to access for humans and machines and poses substantial challenges to the inf...
http://arxiv.org/pdf/2310.15689v1
2023-10-24T10:00:56
2023-10-24T10:00:56
{ "title": [ 0.005832672, 0.015312195, -0.072265625, -0.02911377, -0.01184845, -0.034301758, -0.034179688, 0.0018539429, -0.041229248, 0.05456543, -0.016784668, 0.00630188, 0.019317627, -0.033599854, 0.011787415, -0.0101623535, 0.025131226, -0.01...
http://arxiv.org/abs/2310.15684v1
Improving Biomedical Abstractive Summarisation with Knowledge Aggregation from Citation Papers
Chen Tang, Shun Wang, Tomas Goldsack, Chenghua Lin
cs.CL, cs.AI
Abstracts derived from biomedical literature possess distinct domain-specific characteristics, including specialised writing styles and biomedical terminologies, which necessitate a deep understanding of the related literature. As a result, existing language models struggle to generate technical summaries that are on p...
http://arxiv.org/pdf/2310.15684v1
2023-10-24T09:56:46
2023-10-24T09:56:46
{ "title": [ -0.012931824, 0.012496948, -0.05203247, -0.08917236, 0.008239746, 0.007247925, -0.0602417, 0.007423401, 0.009727478, 0.04901123, -0.002811432, -0.004322052, 0.009780884, -0.011924744, -0.07287598, 0.00566864, 0.004219055, 0.029373169...
http://arxiv.org/abs/2310.15683v1
Prevalence and prevention of large language model use in crowd work
Veniamin Veselovsky, Manoel Horta Ribeiro, Philip Cozzolino, Andrew Gordon, David Rothschild, Robert West
cs.CL
We show that the use of large language models (LLMs) is prevalent among crowd workers, and that targeted mitigation strategies can significantly reduce, but not eliminate, LLM use. On a text summarization task where workers were not directed in any way regarding their LLM use, the estimated prevalence of LLM use was ar...
http://arxiv.org/pdf/2310.15683v1
2023-10-24T09:52:09
2023-10-24T09:52:09
{ "title": [ -0.005760193, 0.008171082, -0.056518555, -0.05432129, 0.015457153, -0.017959595, -0.03414917, 0.007850647, -0.029129028, 0.03881836, -0.021255493, 0.0059051514, 0.021942139, -0.018432617, -0.00856781, 0.0054473877, -0.0012903214, 0.0...
http://arxiv.org/abs/2310.15672v1
How Much Context Does My Attention-Based ASR System Need?
Robert Flynn, Anton Ragni
cs.CL, cs.SD, eess.AS
For the task of speech recognition, the use of more than 30 seconds of acoustic context during training is uncommon, and under-investigated in literature. In this work, we examine the effect of scaling the sequence length used to train/evaluate (dense-attention based) acoustic and language models on speech recognition ...
http://arxiv.org/pdf/2310.15672v1
2023-10-24T09:31:03
2023-10-24T09:31:03
{ "title": [ 0.033935547, 0.0021705627, -0.023239136, -0.049438477, -0.029129028, -0.012901306, -0.025314331, 0.0067596436, -0.031677246, 0.05102539, -0.06652832, -0.015914917, 0.020889282, -0.0034046173, 0.005493164, -0.033843994, 0.034423828, 0...
http://arxiv.org/abs/2310.15664v1
Expression Syntax Information Bottleneck for Math Word Problems
Jing Xiong, Chengming Li, Min Yang, Xiping Hu, Bin Hu
cs.CL
Math Word Problems (MWP) aims to automatically solve mathematical questions given in texts. Previous studies tend to design complex models to capture additional information in the original text so as to enable the model to gain more comprehensive features. In this paper, we turn our attention in the opposite direction,...
http://arxiv.org/pdf/2310.15664v1
2023-10-24T09:23:57
2023-10-24T09:23:57
{ "title": [ -0.018539429, 0.000063717365, -0.039398193, -0.012802124, -0.037475586, -0.031433105, -0.015594482, 0.0039711, -0.01071167, 0.0335083, -0.052764893, 0.0021076202, 0.0126571655, -0.024536133, 0.0037670135, -0.012191772, 0.01777649, 0....
http://arxiv.org/abs/2310.15654v1
A Survey on Detection of LLMs-Generated Content
Xianjun Yang, Liangming Pan, Xuandong Zhao, Haifeng Chen, Linda Petzold, William Yang Wang, Wei Cheng
cs.CL, cs.AI, cs.CY, cs.HC, cs.LG
The burgeoning capabilities of advanced large language models (LLMs) such as ChatGPT have led to an increase in synthetic content generation with implications across a variety of sectors, including media, cybersecurity, public discourse, and education. As such, the ability to detect LLMs-generated content has become of...
http://arxiv.org/pdf/2310.15654v1
2023-10-24T09:10:26
2023-10-24T09:10:26
{ "title": [ -0.015304565, 0.014373779, -0.03366089, -0.025939941, -0.000647068, -0.0043144226, -0.046142578, -0.029815674, -0.033416748, 0.010818481, -0.009613037, 0.036895752, -0.027450562, -0.03878784, 0.0019397736, -0.015159607, -0.029052734, ...
http://arxiv.org/abs/2310.15638v1
CoAnnotating: Uncertainty-Guided Work Allocation between Human and Large Language Models for Data Annotation
Minzhi Li, Taiwei Shi, Caleb Ziems, Min-Yen Kan, Nancy F. Chen, Zhengyuan Liu, Diyi Yang
cs.CL
Annotated data plays a critical role in Natural Language Processing (NLP) in training models and evaluating their performance. Given recent developments in Large Language Models (LLMs), models such as ChatGPT demonstrate zero-shot capability on many text-annotation tasks, comparable with or even exceeding human annotat...
http://arxiv.org/pdf/2310.15638v1
2023-10-24T08:56:49
2023-10-24T08:56:49
{ "title": [ -0.024597168, -0.0039863586, -0.043914795, -0.074035645, -0.011756897, -0.004714966, -0.071899414, -0.006614685, 0.010002136, 0.034942627, -0.012023926, 0.0015487671, 0.00057792664, -0.017410278, -0.05050659, 0.0035915375, -0.0025691986,...
http://arxiv.org/abs/2310.15636v1
Career Path Prediction using Resume Representation Learning and Skill-based Matching
Jens-Joris Decorte, Jeroen Van Hautte, Johannes Deleu, Chris Develder, Thomas Demeester
cs.CL, cs.AI
The impact of person-job fit on job satisfaction and performance is widely acknowledged, which highlights the importance of providing workers with next steps at the right time in their career. This task of predicting the next step in a career is known as career path prediction, and has diverse applications such as turn...
http://arxiv.org/pdf/2310.15636v1
2023-10-24T08:56:06
2023-10-24T08:56:06
{ "title": [ 0.022506714, -0.009521484, -0.01474762, -0.041931152, -0.0037384033, -0.011634827, -0.0011739731, -0.015090942, -0.0041236877, 0.029159546, -0.001496315, -0.008216858, 0.010246277, -0.023284912, -0.03665161, 0.003370285, -0.012886047, ...
http://arxiv.org/abs/2310.15632v1
Tips for making the most of 64-bit architectures in langage design, libraries or garbage collection
Benoît Sonntag, Dominique Colnet
cs.CL
The 64-bit architectures that have become standard today offer unprecedented low-level programming possibilities. For the first time in the history of computing, the size of address registers far exceeded the physical capacity of their bus.After a brief reminder of the possibilities offered by the small size of address...
http://arxiv.org/pdf/2310.15632v1
2023-10-24T08:54:23
2023-10-24T08:54:23
{ "title": [ -0.021987915, -0.018600464, -0.0463562, -0.007083893, -0.034210205, -0.015113831, 0.014533997, 0.024963379, -0.033172607, 0.07342529, -0.0043411255, 0.045074463, -0.022628784, -0.007423401, -0.035217285, -0.015357971, 0.032684326, 0....
http://arxiv.org/abs/2310.15612v3
Machine Translation for Nko: Tools, Corpora and Baseline Results
Moussa Koulako Bala Doumbouya, Baba Mamadi Diané, Solo Farabado Cissé, Djibrila Diané, Abdoulaye Sow, Séré Moussa Doumbouya, Daouda Bangoura, Fodé Moriba Bayo, Ibrahima Sory 2. Condé, Kalo Mory Diané, Chris Piech, Christopher Manning
cs.CL, cs.CY, cs.HC, cs.LG, I.2.6; I.2.7
Currently, there is no usable machine translation system for Nko, a language spoken by tens of millions of people across multiple West African countries, which holds significant cultural and educational value. To address this issue, we present a set of tools, resources, and baseline results aimed towards the developm...
http://arxiv.org/pdf/2310.15612v3
2023-11-15T08:47:28
2023-10-24T08:27:56
{ "title": [ -0.06536865, -0.013954163, -0.0062446594, -0.046295166, -0.027557373, 0.037384033, -0.07348633, -0.019744873, -0.00522995, 0.06793213, 0.007385254, 0.005886078, 0.05041504, -0.02960205, -0.046081543, 0.03866577, -0.015098572, 0.03930...
http://arxiv.org/abs/2310.15602v1
MUSER: A Multi-View Similar Case Retrieval Dataset
Qingquan Li, Yiran Hu, Feng Yao, Chaojun Xiao, Zhiyuan Liu, Maosong Sun, Weixing Shen
cs.CL
Similar case retrieval (SCR) is a representative legal AI application that plays a pivotal role in promoting judicial fairness. However, existing SCR datasets only focus on the fact description section when judging the similarity between cases, ignoring other valuable sections (e.g., the court's opinion) that can provi...
http://arxiv.org/pdf/2310.15602v1
2023-10-24T08:17:11
2023-10-24T08:17:11
{ "title": [ 0.010635376, 0.009773254, -0.048736572, -0.080444336, 0.025421143, -0.0066223145, -0.045440674, -0.019851685, 0.015365601, 0.036499023, -0.040374756, 0.02507019, 0.005607605, -0.011726379, -0.031158447, 0.011054993, -0.014419556, 0.0...
http://arxiv.org/abs/2310.15594v1
Retrieval-based Knowledge Transfer: An Effective Approach for Extreme Large Language Model Compression
Jiduan Liu, Jiahao Liu, Qifan Wang, Jingang Wang, Xunliang Cai, Dongyan Zhao, Ran Lucien Wang, Rui Yan
cs.CL, cs.AI
Large-scale pre-trained language models (LLMs) have demonstrated exceptional performance in various natural language processing (NLP) tasks. However, the massive size of these models poses huge challenges for their deployment in real-world applications. While numerous model compression techniques have been proposed, mo...
http://arxiv.org/pdf/2310.15594v1
2023-10-24T07:58:20
2023-10-24T07:58:20
{ "title": [ -0.03869629, -0.02078247, -0.048828125, -0.06713867, 0.0026779175, 0.013847351, -0.059387207, 0.019226074, 0.0107040405, 0.03717041, -0.008178711, -0.0017242432, 0.007820129, -0.025726318, -0.051574707, 0.009178162, 0.019317627, 0.00...
http://arxiv.org/abs/2310.15587v1
ScanDL: A Diffusion Model for Generating Synthetic Scanpaths on Texts
Lena S. Bolliger, David R. Reich, Patrick Haller, Deborah N. Jakobi, Paul Prasse, Lena A. Jäger
cs.CL
Eye movements in reading play a crucial role in psycholinguistic research studying the cognitive mechanisms underlying human language processing. More recently, the tight coupling between eye movements and cognition has also been leveraged for language-related machine learning tasks such as the interpretability, enhanc...
http://arxiv.org/pdf/2310.15587v1
2023-10-24T07:52:19
2023-10-24T07:52:19
{ "title": [ -0.022506714, -0.02418518, -0.010025024, -0.033996582, -0.017501831, 0.013458252, -0.01979065, -0.015777588, 0.010978699, 0.023117065, -0.025161743, 0.029678345, -0.029251099, -0.0030879974, -0.046295166, -0.0031642914, -0.030410767, ...
http://arxiv.org/abs/2310.15585v1
Multimodal Representations for Teacher-Guided Compositional Visual Reasoning
Wafa Aissa, Marin Ferecatu, Michel Crucianu
cs.CL, cs.CV, cs.LG
Neural Module Networks (NMN) are a compelling method for visual question answering, enabling the translation of a question into a program consisting of a series of reasoning sub-tasks that are sequentially executed on the image to produce an answer. NMNs provide enhanced explainability compared to integrated models, al...
http://arxiv.org/pdf/2310.15585v1
2023-10-24T07:51:08
2023-10-24T07:51:08
{ "title": [ 0.001115799, -0.014915466, -0.02178955, -0.06542969, -0.013397217, -0.0043296814, -0.049560547, 0.004722595, 0.018539429, 0.01448822, -0.02798462, 0.0018110275, 0.0019302368, -0.018310547, -0.021133423, -0.01890564, -0.011924744, 0.0...
http://arxiv.org/abs/2310.15577v1
CONTRASTE: Supervised Contrastive Pre-training With Aspect-based Prompts For Aspect Sentiment Triplet Extraction
Rajdeep Mukherjee, Nithish Kannen, Saurabh Kumar Pandey, Pawan Goyal
cs.CL, cs.AI, I.2.7
Existing works on Aspect Sentiment Triplet Extraction (ASTE) explicitly focus on developing more efficient fine-tuning techniques for the task. Instead, our motivation is to come up with a generic approach that can improve the downstream performances of multiple ABSA tasks simultaneously. Towards this, we present CONTR...
http://arxiv.org/pdf/2310.15577v1
2023-10-24T07:40:09
2023-10-24T07:40:09
{ "title": [ 0.03543091, 0.0002667904, -0.013534546, -0.09124756, 0.0035514832, -0.005290985, -0.040802002, -0.007461548, 0.060638428, 0.052124023, -0.029785156, 0.008415222, 0.0062561035, 0.0059661865, -0.09442139, 0.013908386, -0.0066490173, 0....
http://arxiv.org/abs/2310.15575v1
POE: Process of Elimination for Multiple Choice Reasoning
Chenkai Ma, Xinya Du
cs.CL
Language models (LMs) are capable of conducting in-context learning for multiple choice reasoning tasks, but the options in these tasks are treated equally. As humans often first eliminate wrong options before picking the final correct answer, we argue a similar two-step strategy can make LMs better at these tasks. To ...
http://arxiv.org/pdf/2310.15575v1
2023-10-24T07:38:43
2023-10-24T07:38:43
{ "title": [ -0.010101318, -0.004840851, -0.049194336, -0.015899658, 0.011810303, -0.024429321, 0.030273438, -0.013053894, 0.015556335, 0.0032920837, -0.014175415, -0.013465881, -0.004512787, -0.040924072, -0.023788452, 0.002193451, -0.014312744, ...
http://arxiv.org/abs/2310.15572v1
Natural Language Processing for Drug Discovery Knowledge Graphs: promises and pitfalls
J. Charles G. Jeynes, Tim James, Matthew Corney
cs.CL
Building and analysing knowledge graphs (KGs) to aid drug discovery is a topical area of research. A salient feature of KGs is their ability to combine many heterogeneous data sources in a format that facilitates discovering connections. The utility of KGs has been exemplified in areas such as drug repurposing, with in...
http://arxiv.org/pdf/2310.15572v1
2023-10-24T07:35:24
2023-10-24T07:35:24
{ "title": [ -0.047180176, -0.023880005, -0.044952393, -0.044921875, -0.03189087, 0.013450623, -0.053741455, 0.0143966675, -0.009986877, 0.012687683, -0.03488159, 0.0077934265, -0.032440186, -0.008285522, -0.039093018, -0.021377563, -0.007411957, ...
http://arxiv.org/abs/2310.15571v1
Visually Grounded Continual Language Learning with Selective Specialization
Kyra Ahrens, Lennart Bengtson, Jae Hee Lee, Stefan Wermter
cs.CL
A desirable trait of an artificial agent acting in the visual world is to continually learn a sequence of language-informed tasks while striking a balance between sufficiently specializing in each task and building a generalized knowledge for transfer. Selective specialization, i.e., a careful selection of model compon...
http://arxiv.org/pdf/2310.15571v1
2023-10-24T07:35:23
2023-10-24T07:35:23
{ "title": [ 0.00086688995, 0.0076904297, -0.020401001, -0.068847656, -0.008781433, -0.014541626, -0.03024292, 0.0026474, -0.008598328, 0.030273438, -0.011627197, 0.022857666, -0.01878357, -0.017532349, -0.045074463, 0.00011610985, -0.007461548, ...
http://arxiv.org/abs/2310.15569v1
MuLMS: A Multi-Layer Annotated Text Corpus for Information Extraction in the Materials Science Domain
Timo Pierre Schrader, Matteo Finco, Stefan Grünewald, Felix Hildebrand, Annemarie Friedrich
cs.CL
Keeping track of all relevant recent publications and experimental results for a research area is a challenging task. Prior work has demonstrated the efficacy of information extraction models in various scientific areas. Recently, several datasets have been released for the yet understudied materials science domain. Ho...
http://arxiv.org/pdf/2310.15569v1
2023-10-24T07:23:46
2023-10-24T07:23:46
{ "title": [ -0.0592041, -0.008979797, -0.059509277, -0.04034424, -0.00484848, 0.008232117, -0.0524292, 0.0093688965, 0.015060425, 0.029403687, 0.0027809143, 0.005203247, 0.008972168, -0.031036377, -0.037994385, 0.025726318, -0.025024414, 0.04257...
http://arxiv.org/abs/2310.15556v2
TCRA-LLM: Token Compression Retrieval Augmented Large Language Model for Inference Cost Reduction
Junyi Liu, Liangzhi Li, Tong Xiang, Bowen Wang, Yiming Qian
cs.CL, cs.IR
Since ChatGPT released its API for public use, the number of applications built on top of commercial large language models (LLMs) increase exponentially. One popular usage of such models is leveraging its in-context learning ability and generating responses given user queries leveraging knowledge obtained by retrieval ...
http://arxiv.org/pdf/2310.15556v2
2023-10-25T07:50:52
2023-10-24T06:56:38
{ "title": [ -0.058135986, -0.038116455, -0.067871094, -0.05871582, -0.023361206, 0.0008177757, -0.074645996, 0.021316528, 0.0032253265, 0.023071289, -0.010520935, -0.009162903, 0.012413025, -0.03527832, -0.029403687, 0.004787445, -0.008705139, 0...
http://arxiv.org/abs/2310.15552v1
Unveiling Multilinguality in Transformer Models: Exploring Language Specificity in Feed-Forward Networks
Sunit Bhattacharya, Ondrej Bojar
cs.CL
Recent research suggests that the feed-forward module within Transformers can be viewed as a collection of key-value memories, where the keys learn to capture specific patterns from the input based on the training examples. The values then combine the output from the 'memories' of the keys to generate predictions about...
http://arxiv.org/pdf/2310.15552v1
2023-10-24T06:45:00
2023-10-24T06:45:00
{ "title": [ -0.010139465, 0.008018494, -0.039154053, -0.025115967, -0.0009355545, -0.030410767, -0.046203613, -0.026138306, -0.027648926, 0.043121338, 0.016983032, 0.022506714, 0.005744934, -0.03665161, -0.014091492, 0.013618469, -0.014671326, 0...
http://arxiv.org/abs/2310.18359v1
DeSIQ: Towards an Unbiased, Challenging Benchmark for Social Intelligence Understanding
Xiao-Yu Guo, Yuan-Fang Li, Gholamreza Haffari
cs.CL, cs.AI
Social intelligence is essential for understanding and reasoning about human expressions, intents and interactions. One representative benchmark for its study is Social Intelligence Queries (Social-IQ), a dataset of multiple-choice questions on videos of complex social interactions. We define a comprehensive methodolog...
http://arxiv.org/pdf/2310.18359v1
2023-10-24T06:21:34
2023-10-24T06:21:34
{ "title": [ -0.0043640137, -0.0020065308, -0.014724731, -0.02406311, 0.011985779, -0.014595032, -0.00447464, -0.050628662, 0.0052986145, 0.0064353943, -0.017501831, 0.011604309, -0.029083252, -0.020812988, -0.04156494, 0.00017905235, -0.011856079, ...
http://arxiv.org/abs/2310.15541v1
Improving Language Models Meaning Understanding and Consistency by Learning Conceptual Roles from Dictionary
Myeongjun Erik Jang, Thomas Lukasiewicz
cs.CL
The non-humanlike behaviour of contemporary pre-trained language models (PLMs) is a leading cause undermining their trustworthiness. A striking phenomenon of such faulty behaviours is the generation of inconsistent predictions, which produces logically contradictory results, such as generating different predictions for...
http://arxiv.org/pdf/2310.15541v1
2023-10-24T06:15:15
2023-10-24T06:15:15
{ "title": [ -0.031402588, -0.0052757263, -0.050567627, -0.056030273, -0.014076233, -0.0060653687, -0.062683105, 0.014556885, -0.002714157, 0.030731201, -0.029510498, -0.016998291, 0.033416748, -0.024627686, -0.03353882, 0.003490448, -0.016738892, ...
http://arxiv.org/abs/2310.15539v1
SteloCoder: a Decoder-Only LLM for Multi-Language to Python Code Translation
Jialing Pan, Adrien Sadé, Jin Kim, Eric Soriano, Guillem Sole, Sylvain Flamant
cs.CL, cs.AI
With the recent focus on Large Language Models (LLMs), both StarCoder (Li et al., 2023) and Code Llama (Rozi\`ere et al., 2023) have demonstrated remarkable performance in code generation. However, there is still a need for improvement in code translation functionality with efficient training techniques. In response to...
http://arxiv.org/pdf/2310.15539v1
2023-10-24T06:04:28
2023-10-24T06:04:28
{ "title": [ -0.054382324, -0.035705566, -0.056030273, -0.026229858, -0.047180176, -0.005680084, -0.0814209, 0.004497528, 0.013076782, 0.03967285, 0.012062073, 0.010398865, 0.03552246, -0.048431396, -0.009117126, 0.03451538, 0.00819397, 0.0251617...
http://arxiv.org/abs/2310.15517v1
MarkQA: A large scale KBQA dataset with numerical reasoning
Xiang Huang, Sitao Cheng, Yuheng Bao, Shanshan Huang, Yuzhong Qu
cs.CL
While question answering over knowledge bases (KBQA) has shown progress in addressing factoid questions, KBQA with numerical reasoning remains relatively unexplored. In this paper, we focus on the complex numerical reasoning in KBQA and propose a new task, NR-KBQA, which necessitates the ability to perform both multi-h...
http://arxiv.org/pdf/2310.15517v1
2023-10-24T04:50:59
2023-10-24T04:50:59
{ "title": [ -0.042388916, -0.022491455, -0.06262207, -0.060333252, 0.000035643578, -0.0067710876, -0.05831909, -0.009262085, 0.01789856, 0.030303955, 0.010185242, -0.028686523, 0.0044174194, -0.042144775, -0.0463562, 0.008331299, 0.0051994324, 0...
http://arxiv.org/abs/2310.15515v1
Fighting Fire with Fire: The Dual Role of LLMs in Crafting and Detecting Elusive Disinformation
Jason Lucas, Adaku Uchendu, Michiharu Yamashita, Jooyoung Lee, Shaurya Rohatgi, Dongwon Lee
cs.CL
Recent ubiquity and disruptive impacts of large language models (LLMs) have raised concerns about their potential to be misused (.i.e, generating large-scale harmful and misleading content). To combat this emerging risk of LLMs, we propose a novel "Fighting Fire with Fire" (F3) strategy that harnesses modern LLMs' gene...
http://arxiv.org/pdf/2310.15515v1
2023-10-24T04:50:29
2023-10-24T04:50:29
{ "title": [ -0.00021982193, 0.00932312, -0.043945312, -0.014190674, -0.007423401, -0.028320312, -0.012123108, -0.019363403, -0.06109619, -0.01096344, -0.03869629, 0.052612305, -0.024307251, -0.049591064, 0.055267334, -0.041259766, -0.023330688, ...
http://arxiv.org/abs/2310.15513v1
A Joint Matrix Factorization Analysis of Multilingual Representations
Zheng Zhao, Yftah Ziser, Bonnie Webber, Shay B. Cohen
cs.CL
We present an analysis tool based on joint matrix factorization for comparing latent representations of multilingual and monolingual models. An alternative to probing, this tool allows us to analyze multiple sets of representations in a joint manner. Using this tool, we study to what extent and how morphosyntactic feat...
http://arxiv.org/pdf/2310.15513v1
2023-10-24T04:43:45
2023-10-24T04:43:45
{ "title": [ -0.0037078857, -0.008590698, -0.022613525, -0.06958008, 0.010955811, 0.0056495667, -0.06060791, -0.0031032562, -0.012145996, 0.05117798, 0.02331543, 0.008766174, 0.0042304993, -0.032104492, -0.02822876, 0.020645142, -0.014587402, 0.0...
http://arxiv.org/abs/2310.15511v1
KITAB: Evaluating LLMs on Constraint Satisfaction for Information Retrieval
Marah I Abdin, Suriya Gunasekar, Varun Chandrasekaran, Jerry Li, Mert Yuksekgonul, Rahee Ghosh Peshawaria, Ranjita Naik, Besmira Nushi
cs.LG, cs.AI, cs.CL, cs.IR, I.2.7
We study the ability of state-of-the art models to answer constraint satisfaction queries for information retrieval (e.g., 'a list of ice cream shops in San Diego'). In the past, such queries were considered to be tasks that could only be solved via web-search or knowledge bases. More recently, large language models (L...
http://arxiv.org/pdf/2310.15511v1
2023-10-24T04:40:38
2023-10-24T04:40:38
{ "title": [ -0.04849243, -0.014221191, -0.03527832, -0.019012451, -0.007633209, 0.0015602112, -0.020706177, -0.011650085, 0.017868042, 0.02154541, -0.03302002, 0.0052948, 0.010375977, -0.01108551, -0.036834717, 0.002664566, -0.01737976, 0.012001...
http://arxiv.org/abs/2310.15494v2
TRAMS: Training-free Memory Selection for Long-range Language Modeling
Haofei Yu, Cunxiang Wang, Yue Zhang, Wei Bi
cs.CL
The Transformer architecture is crucial for numerous AI models, but it still faces challenges in long-range language modeling. Though several specific transformer architectures have been designed to tackle issues of long-range dependencies, existing methods like Transformer-XL are plagued by a high percentage of ineffe...
http://arxiv.org/pdf/2310.15494v2
2023-11-05T06:47:31
2023-10-24T03:42:49
{ "title": [ -0.017593384, -0.014968872, -0.04034424, -0.06781006, -0.0052261353, -0.000467062, -0.053009033, 0.021743774, 0.0022449493, 0.036712646, -0.00667572, 0.010536194, 0.00040340424, -0.016433716, -0.051971436, 0.01713562, -0.014633179, 0...
http://arxiv.org/abs/2310.15484v1
NuTrea: Neural Tree Search for Context-guided Multi-hop KGQA
Hyeong Kyu Choi, Seunghun Lee, Jaewon Chu, Hyunwoo J. Kim
cs.CL, cs.AI
Multi-hop Knowledge Graph Question Answering (KGQA) is a task that involves retrieving nodes from a knowledge graph (KG) to answer natural language questions. Recent GNN-based approaches formulate this task as a KG path searching problem, where messages are sequentially propagated from the seed node towards the answer ...
http://arxiv.org/pdf/2310.15484v1
2023-10-24T03:24:15
2023-10-24T03:24:15
{ "title": [ -0.06414795, -0.043273926, -0.02142334, 0.006290436, -0.054351807, 0.022354126, -0.03616333, -0.002817154, -0.0006828308, 0.06890869, -0.012573242, -0.014907837, 0.023742676, -0.023101807, -0.059509277, -0.004638672, -0.0038471222, 0...
http://arxiv.org/abs/2310.15477v1
CRaSh: Clustering, Removing, and Sharing Enhance Fine-tuning without Full Large Language Model
Kaiyan Zhang, Ning Ding, Biqing Qi, Xuekai Zhu, Xinwei Long, Bowen Zhou
cs.CL
Instruction tuning has recently been recognized as an effective way of aligning Large Language Models (LLMs) to enhance their generalization ability across various tasks. However, when tuning publicly accessible, centralized LLMs with private instruction data, privacy concerns are inevitable. While direct transfer of p...
http://arxiv.org/pdf/2310.15477v1
2023-10-24T03:08:58
2023-10-24T03:08:58
{ "title": [ 0.012123108, -0.019165039, -0.026931763, -0.06854248, -0.02268982, 0.003944397, -0.024871826, 0.03262329, -0.019866943, 0.026000977, -0.020187378, 0.009124756, 0.014152527, -0.02078247, -0.006526947, -0.012527466, 0.009979248, 0.0164...
http://arxiv.org/abs/2310.18358v1
A Communication Theory Perspective on Prompting Engineering Methods for Large Language Models
Yuanfeng Song, Yuanqin He, Xuefang Zhao, Hanlin Gu, Di Jiang, Haijun Yang, Lixin Fan, Qiang Yang
cs.CL, cs.AI
The springing up of Large Language Models (LLMs) has shifted the community from single-task-orientated natural language processing (NLP) research to a holistic end-to-end multi-task learning paradigm. Along this line of research endeavors in the area, LLM-based prompting methods have attracted much attention, partially...
http://arxiv.org/pdf/2310.18358v1
2023-10-24T03:05:21
2023-10-24T03:05:21
{ "title": [ -0.034484863, -0.01411438, -0.022415161, -0.045135498, -0.008674622, -0.0018062592, -0.058013916, 0.005622864, 0.006793976, 0.01159668, -0.0023536682, -0.011383057, 0.007587433, -0.040985107, -0.016143799, 0.008514404, -0.009880066, ...
http://arxiv.org/abs/2310.15470v1
Continual Event Extraction with Semantic Confusion Rectification
Zitao Wang, Xinyi Wang, Wei Hu
cs.CL, cs.LG
We study continual event extraction, which aims to extract incessantly emerging event information while avoiding forgetting. We observe that the semantic confusion on event types stems from the annotations of the same text being updated over time. The imbalance between event types even aggravates this issue. This paper...
http://arxiv.org/pdf/2310.15470v1
2023-10-24T02:48:50
2023-10-24T02:48:50
{ "title": [ -0.003929138, 0.013702393, -0.053619385, -0.07281494, 0.031341553, -0.012565613, -0.04824829, -0.018112183, 0.036499023, 0.040130615, 0.007858276, -0.011161804, 0.021728516, -0.029083252, -0.052124023, 0.022262573, -0.006088257, 0.01...
http://arxiv.org/abs/2310.15469v1
The Janus Interface: How Fine-Tuning in Large Language Models Amplifies the Privacy Risks
Xiaoyi Chen, Siyuan Tang, Rui Zhu, Shijun Yan, Lei Jin, Zihao Wang, Liya Su, XiaoFeng Wang, Haixu Tang
cs.CR, cs.CL
The era post-2018 marked the advent of Large Language Models (LLMs), with innovations such as OpenAI's ChatGPT showcasing prodigious linguistic prowess. As the industry galloped toward augmenting model parameters and capitalizing on vast swaths of human language data, security and privacy challenges also emerged. Forem...
http://arxiv.org/pdf/2310.15469v1
2023-10-24T02:48:19
2023-10-24T02:48:19
{ "title": [ -0.015853882, -0.017242432, -0.054382324, -0.06890869, 0.008216858, -0.02470398, -0.03326416, 0.015396118, -0.039916992, 0.026275635, -0.017501831, 0.027114868, 0.00027418137, -0.016921997, -0.010917664, -0.004886627, -0.012649536, 0...
http://arxiv.org/abs/2310.15464v1
Interpreting Answers to Yes-No Questions in User-Generated Content
Shivam Mathur, Keun Hee Park, Dhivya Chinnappa, Saketh Kotamraju, Eduardo Blanco
cs.CL
Interpreting answers to yes-no questions in social media is difficult. Yes and no keywords are uncommon, and the few answers that include them are rarely to be interpreted what the keywords suggest. In this paper, we present a new corpus of 4,442 yes-no question-answer pairs from Twitter. We discuss linguistic characte...
http://arxiv.org/pdf/2310.15464v1
2023-10-24T02:27:06
2023-10-24T02:27:06
{ "title": [ -0.005722046, 0.0008530617, -0.0014572144, -0.03225708, -0.008155823, -0.010551453, 0.020248413, -0.024154663, -0.021194458, 0.039031982, -0.009277344, 0.009712219, -0.029144287, -0.016082764, -0.06237793, -0.0340271, 0.0026893616, 0...
http://arxiv.org/abs/2310.15461v1
Facilitating Self-Guided Mental Health Interventions Through Human-Language Model Interaction: A Case Study of Cognitive Restructuring
Ashish Sharma, Kevin Rushton, Inna Wanyin Lin, Theresa Nguyen, Tim Althoff
cs.HC, cs.CL
Self-guided mental health interventions, such as "do-it-yourself" tools to learn and practice coping strategies, show great promise to improve access to mental health care. However, these interventions are often cognitively demanding and emotionally triggering, creating accessibility barriers that limit their wide-scal...
http://arxiv.org/pdf/2310.15461v1
2023-10-24T02:23:34
2023-10-24T02:23:34
{ "title": [ -0.027862549, 0.0032787323, -0.046325684, -0.04434204, 0.0057411194, -0.011909485, -0.055389404, -0.01374054, -0.011703491, 0.040100098, 0.013442993, 0.0011892319, 0.0016946793, -0.046966553, -0.03515625, 0.010063171, 0.009651184, 0....
http://arxiv.org/abs/2310.15439v1
K-HATERS: A Hate Speech Detection Corpus in Korean with Target-Specific Ratings
Chaewon Park, Soohwan Kim, Kyubyong Park, Kunwoo Park
cs.CL, cs.SI
Numerous datasets have been proposed to combat the spread of online hate. Despite these efforts, a majority of these resources are English-centric, primarily focusing on overt forms of hate. This research gap calls for developing high-quality corpora in diverse languages that also encapsulate more subtle hate expressio...
http://arxiv.org/pdf/2310.15439v1
2023-10-24T01:20:05
2023-10-24T01:20:05
{ "title": [ -0.007129669, -0.0026130676, -0.02154541, -0.036468506, -0.0012845993, -0.009132385, -0.025238037, -0.0135269165, 0.0014667511, 0.027908325, -0.030761719, 0.010559082, 0.015792847, -0.0093307495, -0.04397583, -0.010047913, -0.025131226, ...
http://arxiv.org/abs/2310.18357v1
Leveraging Large Language Models for Enhanced Product Descriptions in eCommerce
Jianghong Zhou, Bo Liu, Jhalak Nilesh Acharya Yao Hong, Kuang-chih Lee, Musen Wen
cs.CL, I.2.7; H.3.3; H.3.5; K.4.4
In the dynamic field of eCommerce, the quality and comprehensiveness of product descriptions are pivotal for enhancing search visibility and customer engagement. Effective product descriptions can address the 'cold start' problem, align with market trends, and ultimately lead to increased click-through rates. Tradition...
http://arxiv.org/pdf/2310.18357v1
2023-10-24T00:55:14
2023-10-24T00:55:14
{ "title": [ -0.03604126, -0.027267456, -0.05831909, -0.07232666, -0.00042891502, -0.0077056885, -0.058166504, 0.018814087, -0.010513306, 0.045654297, -0.01689148, -0.005126953, 0.022644043, -0.0049819946, -0.043792725, 0.0053520203, 0.0015573502, ...
http://arxiv.org/abs/2310.15431v2
What Makes it Ok to Set a Fire? Iterative Self-distillation of Contexts and Rationales for Disambiguating Defeasible Social and Moral Situations
Kavel Rao, Liwei Jiang, Valentina Pyatkin, Yuling Gu, Niket Tandon, Nouha Dziri, Faeze Brahman, Yejin Choi
cs.CL
Moral or ethical judgments rely heavily on the specific contexts in which they occur. Understanding varying shades of defeasible contextualizations (i.e., additional information that strengthens or attenuates the moral acceptability of an action) is critical to accurately represent the subtlety and intricacy of grounde...
http://arxiv.org/pdf/2310.15431v2
2023-11-01T04:39:14
2023-10-24T00:51:29
{ "title": [ 0.027313232, 0.023925781, -0.06915283, -0.049346924, 0.019180298, -0.05065918, -0.017044067, 0.00011843443, -0.00009012222, 0.0076789856, -0.05404663, 0.011039734, -0.0001411438, -0.020431519, 0.022903442, -0.045715332, 0.00096178055, ...
http://arxiv.org/abs/2310.15429v1
Beyond Sentiment: Leveraging Topic Metrics for Political Stance Classification
Weihong Qi
cs.CL, cs.CY
Sentiment analysis, widely critiqued for capturing merely the overall tone of a corpus, falls short in accurately reflecting the latent structures and political stances within texts. This study introduces topic metrics, dummy variables converted from extracted topics, as both an alternative and complement to sentiment ...
http://arxiv.org/pdf/2310.15429v1
2023-10-24T00:50:33
2023-10-24T00:50:33
{ "title": [ 0.011680603, 0.0048065186, -0.032806396, -0.044128418, 0.01928711, -0.025909424, -0.0059890747, -0.0262146, 0.010879517, 0.013809204, -0.037109375, 0.018661499, -0.03945923, -0.00006580353, -0.042755127, -0.018096924, -0.012184143, 0...
http://arxiv.org/abs/2310.18356v2
LoRAShear: Efficient Large Language Model Structured Pruning and Knowledge Recovery
Tianyi Chen, Tianyu Ding, Badal Yadav, Ilya Zharkov, Luming Liang
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have transformed the landscape of artificial intelligence, while their enormous size presents significant challenges in terms of computational costs. We introduce LoRAShear, a novel efficient approach to structurally prune LLMs and recover knowledge. Given general LLMs, LoRAShear at first c...
http://arxiv.org/pdf/2310.18356v2
2023-10-31T04:21:33
2023-10-24T00:47:26
{ "title": [ -0.04385376, -0.036071777, -0.05444336, -0.09136963, -0.006095886, 0.014945984, -0.072021484, 0.04736328, 0.00504303, 0.020935059, -0.017745972, 0.010520935, 0.015686035, -0.01864624, -0.029327393, 0.0017004013, -0.0004079342, 0.0184...
http://arxiv.org/abs/2310.15425v1
The Mason-Alberta Phonetic Segmenter: A forced alignment system based on deep neural networks and interpolation
Matthew C. Kelley, Scott James Perry, Benjamin V. Tucker
eess.AS, cs.CL, cs.LG, cs.SD
Forced alignment systems automatically determine boundaries between segments in speech data, given an orthographic transcription. These tools are commonplace in phonetics to facilitate the use of speech data that would be infeasible to manually transcribe and segment. In the present paper, we describe a new neural netw...
http://arxiv.org/pdf/2310.15425v1
2023-10-24T00:43:54
2023-10-24T00:43:54
{ "title": [ -0.01335144, 0.014419556, -0.05923462, -0.05911255, 0.0013799667, -0.017501831, -0.051879883, 0.002670288, -0.021621704, 0.04748535, -0.0023422241, -0.007865906, 0.027359009, -0.035125732, -0.018234253, 0.0002565384, -0.005771637, 0....
http://arxiv.org/abs/2310.15421v3
FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions
Hyunwoo Kim, Melanie Sclar, Xuhui Zhou, Ronan Le Bras, Gunhee Kim, Yejin Choi, Maarten Sap
cs.CL, cs.AI
Theory of mind (ToM) evaluations currently focus on testing models using passive narratives that inherently lack interactivity. We introduce FANToM, a new benchmark designed to stress-test ToM within information-asymmetric conversational contexts via question answering. Our benchmark draws upon important theoretical re...
http://arxiv.org/pdf/2310.15421v3
2023-10-31T17:58:30
2023-10-24T00:24:11
{ "title": [ -0.016464233, -0.021331787, -0.01651001, -0.028015137, -0.0052871704, -0.014450073, 0.004386902, -0.007575989, 0.004585266, 0.0027599335, -0.019592285, -0.005542755, 0.0027656555, -0.028320312, -0.037475586, -0.015533447, -0.01751709, ...
http://arxiv.org/abs/2310.15420v1
Let the Pretrained Language Models "Imagine" for Short Texts Topic Modeling
Pritom Saha Akash, Jie Huang, Kevin Chen-Chuan Chang
cs.CL
Topic models are one of the compelling methods for discovering latent semantics in a document collection. However, it assumes that a document has sufficient co-occurrence information to be effective. However, in short texts, co-occurrence information is minimal, which results in feature sparsity in document representat...
http://arxiv.org/pdf/2310.15420v1
2023-10-24T00:23:30
2023-10-24T00:23:30
{ "title": [ -0.023590088, -0.016204834, 0.0044136047, 0.03491211, -0.05960083, -0.0032787323, -0.011268616, -0.028015137, -0.016860962, 0.033050537, -0.011184692, 0.02204895, 0.0184021, -0.0385437, 0.019058228, -0.008041382, -0.0034503937, 0.020...
http://arxiv.org/abs/2310.15415v1
Mind the Gap Between Conversations for Improved Long-Term Dialogue Generation
Qiang Zhang, Jason Naradowsky, Yusuke Miyao
cs.CL, cs.AI, cs.HC
Knowing how to end and resume conversations over time is a natural part of communication, allowing for discussions to span weeks, months, or years. The duration of gaps between conversations dictates which topics are relevant and which questions to ask, and dialogue systems which do not explicitly model time may genera...
http://arxiv.org/pdf/2310.15415v1
2023-10-24T00:12:38
2023-10-24T00:12:38
{ "title": [ 0.015563965, -0.0034885406, -0.023223877, -0.062683105, -0.009437561, -0.019714355, -0.040802002, 0.000995636, 0.0022296906, 0.011123657, -0.043640137, -0.0030269623, -0.0029010773, -0.03665161, 0.0048675537, -0.021011353, -0.013336182, ...
http://arxiv.org/abs/2310.15405v1
GPT-4 as an Effective Zero-Shot Evaluator for Scientific Figure Captions
Ting-Yao Hsu, Chieh-Yang Huang, Ryan Rossi, Sungchul Kim, C. Lee Giles, Ting-Hao K. Huang
cs.CL
There is growing interest in systems that generate captions for scientific figures. However, assessing these systems output poses a significant challenge. Human evaluation requires academic expertise and is costly, while automatic evaluation depends on often low-quality author-written captions. This paper investigates ...
http://arxiv.org/pdf/2310.15405v1
2023-10-23T23:24:57
2023-10-23T23:24:57
{ "title": [ 0.02520752, 0.0041160583, -0.0019931793, -0.013755798, -0.015335083, -0.02519226, 0.03137207, -0.026657104, 0.024505615, 0.05493164, 0.013328552, -0.008666992, 0.013587952, -0.03289795, -0.060913086, 0.028076172, -0.007205963, 0.0396...
http://arxiv.org/abs/2310.15398v1
"One-size-fits-all"? Observations and Expectations of NLG Systems Across Identity-Related Language Features
Li Lucy, Su Lin Blodgett, Milad Shokouhi, Hanna Wallach, Alexandra Olteanu
cs.CL, cs.HC
Fairness-related assumptions about what constitutes appropriate NLG system behaviors range from invariance, where systems are expected to respond identically to social groups, to adaptation, where responses should instead vary across them. We design and conduct five case studies, in which we perturb different types of ...
http://arxiv.org/pdf/2310.15398v1
2023-10-23T23:00:34
2023-10-23T23:00:34
{ "title": [ -0.005794525, -0.008140564, -0.022827148, -0.054107666, -0.0027637482, -0.0087890625, -0.022399902, 0.008720398, -0.010612488, 0.03894043, -0.012077332, 0.0072898865, 0.025146484, -0.025268555, -0.041931152, 0.006149292, -0.0181427, ...
http://arxiv.org/abs/2310.15393v1
DoGE: Domain Reweighting with Generalization Estimation
Simin Fan, Matteo Pagliardini, Martin Jaggi
cs.LG, cs.AI, cs.CL
The coverage and composition of the pretraining data corpus significantly impacts the generalization ability of large language models. Conventionally, the pretraining corpus is composed of various source domains (e.g. CommonCrawl, Wikipedia, Github etc.) according to certain sampling probabilities (domain weights). How...
http://arxiv.org/pdf/2310.15393v1
2023-10-23T22:51:58
2023-10-23T22:51:58
{ "title": [ -0.031951904, -0.015266418, -0.04714966, -0.057647705, -0.0015249252, -0.0048561096, -0.049072266, -0.03778076, 0.013755798, 0.024291992, -0.007534027, 0.007232666, -0.0023231506, -0.03189087, -0.028076172, 0.016799927, -0.008552551, ...
http://arxiv.org/abs/2310.15389v1
Irreducible Curriculum for Language Model Pretraining
Simin Fan, Martin Jaggi
cs.CL, cs.AI, cs.LG
Automatic data selection and curriculum design for training large language models is challenging, with only a few existing methods showing improvements over standard training. Furthermore, current schemes focus on domain-level selection, overlooking the more fine-grained contributions of each individual training point....
http://arxiv.org/pdf/2310.15389v1
2023-10-23T22:41:33
2023-10-23T22:41:33
{ "title": [ -0.00983429, -0.011810303, -0.043121338, -0.060943604, -0.035339355, -0.0087509155, -0.05090332, 0.006515503, -0.033447266, 0.027053833, 0.0020389557, 0.0042152405, -0.010498047, -0.05218506, 0.015258789, 0.0015516281, -0.010940552, ...
http://arxiv.org/abs/2310.15383v1
GD-COMET: A Geo-Diverse Commonsense Inference Model
Mehar Bhatia, Vered Shwartz
cs.CL
With the increasing integration of AI into everyday life, it's becoming crucial to design AI systems that serve users from diverse backgrounds by making them culturally aware. In this paper, we present GD-COMET, a geo-diverse version of the COMET commonsense inference model. GD-COMET goes beyond Western commonsense kno...
http://arxiv.org/pdf/2310.15383v1
2023-10-23T22:03:56
2023-10-23T22:03:56
{ "title": [ 0.005847931, -0.020828247, -0.02418518, -0.070617676, -0.0072898865, -0.0006713867, -0.053375244, -0.009803772, 0.032226562, 0.012809753, -0.0284729, -0.016723633, -0.022613525, -0.02003479, -0.03753662, -0.0018291473, -0.00070619583, ...
http://arxiv.org/abs/2310.18355v1
Health Disparities through Generative AI Models: A Comparison Study Using A Domain Specific large language model
Yohn Jairo Parra Bautista, Vinicious Lima, Carlos Theran, Richard Alo
cs.CL, cs.AI, cs.LG
Health disparities are differences in health outcomes and access to healthcare between different groups, including racial and ethnic minorities, low-income people, and rural residents. An artificial intelligence (AI) program called large language models (LLMs) can understand and generate human language, improving healt...
http://arxiv.org/pdf/2310.18355v1
2023-10-23T21:24:05
2023-10-23T21:24:05
{ "title": [ 0.01448822, 0.0039749146, -0.03970337, -0.070495605, 0.015823364, -0.0151901245, -0.047821045, -0.012710571, 0.006225586, 0.015670776, -0.015068054, 0.016143799, -0.003786087, -0.033569336, -0.027053833, -0.0050239563, -0.01789856, 0...
http://arxiv.org/abs/2310.15373v1
Semantic Data Management in Data Lakes
Sayed Hoseini, Johannes Theissen-Lipp, Christoph Quix
cs.DB, cs.AI, cs.CL, cs.HC
In recent years, data lakes emerged as away to manage large amounts of heterogeneous data for modern data analytics. One way to prevent data lakes from turning into inoperable data swamps is semantic data management. Some approaches propose the linkage of metadata to knowledge graphs based on the Linked Data principles...
http://arxiv.org/pdf/2310.15373v1
2023-10-23T21:16:50
2023-10-23T21:16:50
{ "title": [ -0.03463745, -0.0046806335, -0.08734131, -0.09301758, 0.0023593903, -0.0020923615, -0.055999756, 0.036865234, -0.0077400208, 0.055236816, -0.04019165, -0.006134033, 0.010498047, -0.0049476624, -0.044189453, -0.0033664703, -0.0018863678, ...
http://arxiv.org/abs/2310.15372v1
EpiK-Eval: Evaluation for Language Models as Epistemic Models
Gabriele Prato, Jerry Huang, Prasannna Parthasarathi, Shagun Sodhani, Sarath Chandar
cs.CL, cs.AI
In the age of artificial intelligence, the role of large language models (LLMs) is becoming increasingly central. Despite their growing prevalence, their capacity to consolidate knowledge from different training documents - a crucial ability in numerous applications - remains unexplored. This paper presents the first s...
http://arxiv.org/pdf/2310.15372v1
2023-10-23T21:15:54
2023-10-23T21:15:54
{ "title": [ -0.037200928, -0.011047363, -0.040496826, -0.075683594, 0.005794525, 0.017089844, -0.064697266, -0.005908966, 0.009490967, 0.0104522705, -0.011962891, -0.024612427, -0.00062704086, -0.029388428, -0.05227661, 0.000015676022, -0.028747559,...
http://arxiv.org/abs/2310.15355v1
Why LLMs Hallucinate, and How to Get (Evidential) Closure: Perceptual, Intensional, and Extensional Learning for Faithful Natural Language Generation
Adam Bouyamourn
cs.CL, cs.AI, I.2.7
We show that LLMs hallucinate because their output is not constrained to be synonymous with claims for which they have evidence: a condition that we call evidential closure. Information about the truth or falsity of sentences is not statistically identified in the standard neural probabilistic language model setup, and...
http://arxiv.org/pdf/2310.15355v1
2023-10-23T20:35:52
2023-10-23T20:35:52
{ "title": [ -0.038360596, 0.0009522438, -0.057128906, -0.006931305, -0.041015625, -0.0093688965, -0.049346924, -0.009605408, -0.03149414, 0.036865234, -0.055633545, 0.037841797, 0.0029258728, -0.009063721, -0.010627747, -0.019165039, -0.011657715, ...
http://arxiv.org/abs/2310.18354v1
A Review of Reinforcement Learning for Natural Language Processing, and Applications in Healthcare
Ying Liu, Haozhu Wang, Huixue Zhou, Mingchen Li, Yu Hou, Sicheng Zhou, Fang Wang, Rama Hoetzlein, Rui Zhang
cs.CL
Reinforcement learning (RL) has emerged as a powerful approach for tackling complex medical decision-making problems such as treatment planning, personalized medicine, and optimizing the scheduling of surgeries and appointments. It has gained significant attention in the field of Natural Language Processing (NLP) due t...
http://arxiv.org/pdf/2310.18354v1
2023-10-23T20:26:15
2023-10-23T20:26:15
{ "title": [ -0.007896423, -0.0036258698, -0.030639648, -0.04977417, 0.017303467, -0.010658264, -0.023223877, -0.011955261, -0.056762695, 0.01802063, -0.02154541, 0.019836426, -0.036376953, -0.027755737, -0.008605957, -0.029953003, -0.008712769, ...
http://arxiv.org/abs/2310.15337v1
Moral Foundations of Large Language Models
Marwa Abdulhai, Gregory Serapio-Garcia, Clément Crepy, Daria Valter, John Canny, Natasha Jaques
cs.AI, cs.CL, cs.CY
Moral foundations theory (MFT) is a psychological assessment tool that decomposes human moral reasoning into five factors, including care/harm, liberty/oppression, and sanctity/degradation (Graham et al., 2009). People vary in the weight they place on these dimensions when making moral decisions, in part due to their c...
http://arxiv.org/pdf/2310.15337v1
2023-10-23T20:05:37
2023-10-23T20:05:37
{ "title": [ -0.021560669, -0.0061416626, -0.03704834, -0.06274414, 0.0023097992, -0.0007224083, -0.04208374, 0.013572693, -0.00466156, 0.032226562, -0.03173828, -0.0064582825, -0.010124207, -0.0057754517, -0.04812622, -0.011680603, -0.017105103, ...
http://arxiv.org/abs/2310.15326v1
Specialist or Generalist? Instruction Tuning for Specific NLP Tasks
Chufan Shi, Yixuan Su, Cheng Yang, Yujiu Yang, Deng Cai
cs.CL
The potential of large language models (LLMs) to simultaneously perform a wide range of natural language processing (NLP) tasks has been the subject of extensive research. Although instruction tuning has proven to be a data-efficient method for transforming LLMs into such generalist models, their performance still lags...
http://arxiv.org/pdf/2310.15326v1
2023-10-23T19:46:48
2023-10-23T19:46:48
{ "title": [ -0.03201294, -0.01612854, -0.0287323, -0.032043457, -0.031982422, -0.020233154, -0.033813477, 0.0029850006, -0.020584106, 0.024871826, -0.037475586, 0.02406311, -0.0206604, -0.0022068024, -0.026397705, -0.014518738, -0.013847351, 0.0...
http://arxiv.org/abs/2310.15325v1
LXMERT Model Compression for Visual Question Answering
Maryam Hashemi, Ghazaleh Mahmoudi, Sara Kodeiri, Hadi Sheikhi, Sauleh Eetemadi
cs.CV, cs.CL, cs.LG
Large-scale pretrained models such as LXMERT are becoming popular for learning cross-modal representations on text-image pairs for vision-language tasks. According to the lottery ticket hypothesis, NLP and computer vision models contain smaller subnetworks capable of being trained in isolation to full performance. In t...
http://arxiv.org/pdf/2310.15325v1
2023-10-23T19:46:41
2023-10-23T19:46:41
{ "title": [ -0.021316528, -0.014190674, -0.026550293, -0.047698975, -0.018188477, 0.008911133, -0.050689697, -0.004096985, 0.024261475, 0.023208618, -0.01751709, -0.012825012, -0.0048179626, -0.01373291, -0.051940918, 0.000995636, 0.03189087, -0...
http://arxiv.org/abs/2310.15319v1
Hallucination Detection for Grounded Instruction Generation
Lingjun Zhao, Khanh Nguyen, Hal Daumé III
cs.CL, cs.AI, cs.LG
We investigate the problem of generating instructions to guide humans to navigate in simulated residential environments. A major issue with current models is hallucination: they generate references to actions or objects that are inconsistent with what a human follower would perform or encounter along the described path...
http://arxiv.org/pdf/2310.15319v1
2023-10-23T19:36:28
2023-10-23T19:36:28
{ "title": [ -0.011985779, 0.017700195, -0.026855469, -0.016067505, 0.003768921, -0.013832092, -0.0055732727, -0.036102295, -0.0074310303, 0.028823853, -0.028671265, 0.027801514, -0.004699707, -0.01361084, -0.025360107, 0.0026741028, -0.025421143, ...
http://arxiv.org/abs/2310.15317v1
Exploring the Potential of Large Language Models in Generating Code-Tracing Questions for Introductory Programming Courses
Aysa Xuemo Fan, Ranran Haoran Zhang, Luc Paquette, Rui Zhang
cs.CL, cs.CY
In this paper, we explore the application of large language models (LLMs) for generating code-tracing questions in introductory programming courses. We designed targeted prompts for GPT4, guiding it to generate code-tracing questions based on code snippets and descriptions. We established a set of human evaluation metr...
http://arxiv.org/pdf/2310.15317v1
2023-10-23T19:35:01
2023-10-23T19:35:01
{ "title": [ -0.01725769, -0.053741455, 0.0043182373, -0.032958984, -0.040985107, 0.010177612, -0.026992798, 0.011680603, 0.002878189, 0.0034999847, -0.031677246, 0.010688782, -0.015975952, -0.02420044, -0.0030784607, -0.027282715, 0.0029182434, ...
http://arxiv.org/abs/2310.15316v1
Probing Representations for Document-level Event Extraction
Barry Wang, Xinya Du, Claire Cardie
cs.CL
The probing classifiers framework has been employed for interpreting deep neural network models for a variety of natural language processing (NLP) applications. Studies, however, have largely focused on sentencelevel NLP tasks. This work is the first to apply the probing paradigm to representations learned for document...
http://arxiv.org/pdf/2310.15316v1
2023-10-23T19:33:04
2023-10-23T19:33:04
{ "title": [ 0.008636475, 0.002696991, -0.023254395, -0.08380127, 0.0158844, -0.0023078918, -0.037719727, -0.010887146, 0.016159058, 0.024337769, -0.007369995, 0.0072631836, -0.000018119812, -0.022003174, -0.037384033, 0.016021729, -0.016189575, ...
http://arxiv.org/abs/2310.15302v1
Toward a Critical Toponymy Framework for Named Entity Recognition: A Case Study of Airbnb in New York City
Mikael Brunila, Jack LaViolette, Sky CH-Wang, Priyanka Verma, Clara Féré, Grant McKenzie
cs.CL, cs.AI, cs.CY
Critical toponymy examines the dynamics of power, capital, and resistance through place names and the sites to which they refer. Studies here have traditionally focused on the semantic content of toponyms and the top-down institutional processes that produce them. However, they have generally ignored the ways in which ...
http://arxiv.org/pdf/2310.15302v1
2023-10-23T19:09:07
2023-10-23T19:09:07
{ "title": [ 0.01524353, -0.0039100647, -0.011260986, -0.034088135, 0.0022125244, -0.010826111, -0.025039673, -0.038116455, -0.011955261, 0.034423828, -0.039276123, 0.022964478, 0.002506256, -0.009063721, -0.034210205, -0.021408081, 0.0007715225, ...
http://arxiv.org/abs/2310.15298v2
TaskDiff: A Similarity Metric for Task-Oriented Conversations
Ankita Bhaumik, Praveen Venkateswaran, Yara Rizk, Vatche Isahagian
cs.CL, cs.AI
The popularity of conversational digital assistants has resulted in the availability of large amounts of conversational data which can be utilized for improved user experience and personalized response generation. Building these assistants using popular large language models like ChatGPT also require additional emphasi...
http://arxiv.org/pdf/2310.15298v2
2023-10-25T06:10:07
2023-10-23T19:03:35
{ "title": [ -0.034332275, -0.004787445, -0.056854248, -0.051330566, 0.019973755, -0.0053482056, -0.04977417, -0.00907135, 0.021987915, 0.03237915, 0.00026655197, 0.011619568, 0.0066452026, -0.019058228, -0.046295166, 0.0065345764, -0.0020370483, ...
http://arxiv.org/abs/2310.15296v1
DeTiME: Diffusion-Enhanced Topic Modeling using Encoder-decoder based LLM
Weijie Xu, Wenxiang Hu, Fanyou Wu, Srinivasan Sengamedu
cs.CL, cs.AI, 68T50, I.2.7
In the burgeoning field of natural language processing, Neural Topic Models (NTMs) and Large Language Models (LLMs) have emerged as areas of significant research interest. Despite this, NTMs primarily utilize contextual embeddings from LLMs, which are not optimal for clustering or capable for topic generation. Our stud...
http://arxiv.org/pdf/2310.15296v1
2023-10-23T19:03:04
2023-10-23T19:03:04
{ "title": [ -0.03201294, -0.0019512177, -0.05291748, -0.025283813, -0.009307861, 0.0014944077, -0.026535034, -0.022720337, 0.028503418, 0.016448975, -0.007827759, 0.04324341, 0.0002641678, -0.031829834, -0.0071525574, 0.030426025, -0.017669678, ...
http://arxiv.org/abs/2310.15294v1
Adaptive End-to-End Metric Learning for Zero-Shot Cross-Domain Slot Filling
Yuanjun Shi, Linzhi Wu, Minglai Shao
cs.CL
Recently slot filling has witnessed great development thanks to deep learning and the availability of large-scale annotated data. However, it poses a critical challenge to handle a novel domain whose samples are never seen during training. The recognition performance might be greatly degraded due to severe domain shift...
http://arxiv.org/pdf/2310.15294v1
2023-10-23T19:01:16
2023-10-23T19:01:16
{ "title": [ 0.006652832, -0.0116119385, -0.01889038, -0.049957275, 0.00409317, -0.008407593, -0.006877899, -0.0015211105, 0.001244545, 0.03717041, -0.0149383545, 0.012130737, -0.0008378029, -0.019363403, -0.01512146, 0.00095939636, -0.0064735413, ...
http://arxiv.org/abs/2310.15285v1
On the Dimensionality of Sentence Embeddings
Hongwei Wang, Hongming Zhang, Dong Yu
cs.CL
Learning sentence embeddings is a fundamental problem in natural language processing. While existing research primarily focuses on enhancing the quality of sentence embeddings, the exploration of sentence embedding dimensions is limited. Here we present a comprehensive and empirical analysis of the dimensionality of se...
http://arxiv.org/pdf/2310.15285v1
2023-10-23T18:51:00
2023-10-23T18:51:00
{ "title": [ -0.0154800415, 0.00039935112, -0.034179688, -0.0491333, 0.010169983, 0.0035934448, -0.033599854, -0.003604889, 0.02204895, 0.013298035, -0.023101807, 0.00592041, 0.005191803, -0.009635925, -0.043121338, 0.008903503, -0.02243042, 0.04...
http://arxiv.org/abs/2310.15276v1
Efficient Algorithms for Recognizing Weighted Tree-Adjoining Languages
Alexandra Butoi, Tim Vieira, Ryan Cotterell, David Chiang
cs.CL, cs.FL
The class of tree-adjoining languages can be characterized by various two-level formalisms, consisting of a context-free grammar (CFG) or pushdown automaton (PDA) controlling another CFG or PDA. These four formalisms are equivalent to tree-adjoining grammars (TAG), linear indexed grammars (LIG), pushdown-adjoining auto...
http://arxiv.org/pdf/2310.15276v1
2023-10-23T18:26:00
2023-10-23T18:26:00
{ "title": [ -0.010765076, -0.01084137, -0.022216797, -0.11071777, 0.006980896, 0.022460938, -0.07336426, 0.023864746, -0.00605011, 0.01889038, -0.01537323, -0.0066070557, -0.03564453, -0.013671875, -0.051696777, -0.025390625, -0.012893677, 0.033...