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/2310.15269v1
GradSim: Gradient-Based Language Grouping for Effective Multilingual Training
Mingyang Wang, Heike Adel, Lukas Lange, Jannik Strötgen, Hinrich Schütze
cs.LG, cs.CL
Most languages of the world pose low-resource challenges to natural language processing models. With multilingual training, knowledge can be shared among languages. However, not all languages positively influence each other and it is an open research question how to select the most suitable set of languages for multili...
http://arxiv.org/pdf/2310.15269v1
2023-10-23T18:13:37
2023-10-23T18:13:37
{ "title": [ -0.02357483, 0.00009262562, -0.052581787, -0.07171631, 0.013046265, 0.00390625, -0.058410645, 0.001376152, -0.039154053, 0.03164673, 0.00831604, 0.023422241, -0.0025787354, -0.02973938, -0.012809753, -0.0049057007, -0.018051147, 0.03...
http://arxiv.org/abs/2310.15264v1
Towards Possibilities & Impossibilities of AI-generated Text Detection: A Survey
Soumya Suvra Ghosal, Souradip Chakraborty, Jonas Geiping, Furong Huang, Dinesh Manocha, Amrit Singh Bedi
cs.CL, cs.AI
Large Language Models (LLMs) have revolutionized the domain of natural language processing (NLP) with remarkable capabilities of generating human-like text responses. However, despite these advancements, several works in the existing literature have raised serious concerns about the potential misuse of LLMs such as spr...
http://arxiv.org/pdf/2310.15264v1
2023-10-23T18:11:32
2023-10-23T18:11:32
{ "title": [ -0.032714844, 0.0033073425, -0.01838684, -0.0012845993, 0.00018751621, -0.0053138733, 0.010498047, -0.033721924, -0.005970001, 0.03439331, -0.039031982, 0.015380859, 0.001001358, -0.0021705627, -0.04724121, -0.012451172, -0.024337769, ...
http://arxiv.org/abs/2310.15262v1
Data Augmentation Techniques for Machine Translation of Code-Switched Texts: A Comparative Study
Injy Hamed, Nizar Habash, Ngoc Thang Vu
cs.CL
Code-switching (CSW) text generation has been receiving increasing attention as a solution to address data scarcity. In light of this growing interest, we need more comprehensive studies comparing different augmentation approaches. In this work, we compare three popular approaches: lexical replacements, linguistic theo...
http://arxiv.org/pdf/2310.15262v1
2023-10-23T18:09:41
2023-10-23T18:09:41
{ "title": [ -0.03677368, -0.019241333, -0.012313843, -0.015899658, -0.018493652, -0.0014972687, -0.008354187, 0.0032081604, 0.004558563, 0.049591064, -0.016464233, 0.018722534, 0.033172607, -0.013641357, -0.050689697, 0.018417358, -0.019165039, ...
http://arxiv.org/abs/2310.15259v1
Reference Free Domain Adaptation for Translation of Noisy Questions with Question Specific Rewards
Baban Gain, Ramakrishna Appicharla, Soumya Chennabasavaraj, Nikesh Garera, Asif Ekbal, Muthusamy Chelliah
cs.CL, cs.AI, cs.LG
Community Question-Answering (CQA) portals serve as a valuable tool for helping users within an organization. However, making them accessible to non-English-speaking users continues to be a challenge. Translating questions can broaden the community's reach, benefiting individuals with similar inquiries in various langu...
http://arxiv.org/pdf/2310.15259v1
2023-10-23T18:08:01
2023-10-23T18:08:01
{ "title": [ -0.009735107, -0.0077323914, -0.00022172928, -0.03366089, -0.020324707, -0.009666443, -0.007408142, -0.0055007935, 0.0138168335, 0.036987305, 0.0059776306, -0.00038433075, 0.00843811, -0.022476196, -0.0446167, 0.022644043, 0.013069153, ...
http://arxiv.org/abs/2310.15258v1
Breaking the Language Barrier: Improving Cross-Lingual Reasoning with Structured Self-Attention
Negar Foroutan, Mohammadreza Banaei, Karl Aberer, Antoine Bosselut
cs.CL
In this work, we study whether multilingual language models (MultiLMs) can transfer logical reasoning abilities to other languages when they are fine-tuned for reasoning in a different language. We evaluate the cross-lingual reasoning abilities of MultiLMs in two schemes: (1) where the language of the context and the q...
http://arxiv.org/pdf/2310.15258v1
2023-10-23T18:06:38
2023-10-23T18:06:38
{ "title": [ 0.001405716, -0.00015676022, -0.032287598, -0.06939697, 0.0044288635, -0.00060892105, -0.045410156, -0.0061187744, 0.01184082, 0.029266357, -0.004383087, 0.005088806, 0.0087509155, -0.016159058, -0.036010742, 0.013931274, 0.002298355, ...
http://arxiv.org/abs/2310.15239v1
CRoW: Benchmarking Commonsense Reasoning in Real-World Tasks
Mete Ismayilzada, Debjit Paul, Syrielle Montariol, Mor Geva, Antoine Bosselut
cs.CL, cs.AI
Recent efforts in natural language processing (NLP) commonsense reasoning research have yielded a considerable number of new datasets and benchmarks. However, most of these datasets formulate commonsense reasoning challenges in artificial scenarios that are not reflective of the tasks which real-world NLP systems are d...
http://arxiv.org/pdf/2310.15239v1
2023-10-23T18:00:23
2023-10-23T18:00:23
{ "title": [ -0.004638672, -0.041381836, -0.025863647, -0.056427002, -0.01259613, -0.005870819, -0.025650024, -0.007663727, 0.022201538, 0.018692017, -0.012680054, 0.0020008087, -0.02293396, -0.027572632, -0.03604126, 0.0024337769, -0.0071372986, ...
http://arxiv.org/abs/2310.15166v1
Large Language Models are Visual Reasoning Coordinators
Liangyu Chen, Bo Li, Sheng Shen, Jingkang Yang, Chunyuan Li, Kurt Keutzer, Trevor Darrell, Ziwei Liu
cs.CV, cs.CL
Visual reasoning requires multimodal perception and commonsense cognition of the world. Recently, multiple vision-language models (VLMs) have been proposed with excellent commonsense reasoning ability in various domains. However, how to harness the collective power of these complementary VLMs is rarely explored. Existi...
http://arxiv.org/pdf/2310.15166v1
2023-10-23T17:59:31
2023-10-23T17:59:31
{ "title": [ 0.00011789799, -0.027297974, -0.027114868, -0.06976318, -0.06652832, 0.003107071, -0.07574463, 0.020111084, 0.013786316, 0.04034424, -0.04437256, -0.03668213, 0.044525146, -0.037597656, -0.0041122437, -0.012306213, 0.027832031, 0.018...
http://arxiv.org/abs/2310.15164v1
LINC: A Neurosymbolic Approach for Logical Reasoning by Combining Language Models with First-Order Logic Provers
Theo X. Olausson, Alex Gu, Benjamin Lipkin, Cedegao E. Zhang, Armando Solar-Lezama, Joshua B. Tenenbaum, Roger Levy
cs.CL, cs.AI
Logical reasoning, i.e., deductively inferring the truth value of a conclusion from a set of premises, is an important task for artificial intelligence with wide potential impacts on science, mathematics, and society. While many prompting-based strategies have been proposed to enable Large Language Models (LLMs) to do ...
http://arxiv.org/pdf/2310.15164v1
2023-10-23T17:58:40
2023-10-23T17:58:40
{ "title": [ -0.021469116, -0.016174316, -0.013061523, -0.045440674, -0.03074646, 0.0098724365, -0.062927246, -0.004886627, 0.005947113, 0.033477783, -0.0104522705, -0.017120361, 0.032043457, -0.035064697, -0.027435303, 0.0036945343, -0.016296387, ...
http://arxiv.org/abs/2310.15154v1
Linear Representations of Sentiment in Large Language Models
Curt Tigges, Oskar John Hollinsworth, Atticus Geiger, Neel Nanda
cs.LG, cs.AI, cs.CL
Sentiment is a pervasive feature in natural language text, yet it is an open question how sentiment is represented within Large Language Models (LLMs). In this study, we reveal that across a range of models, sentiment is represented linearly: a single direction in activation space mostly captures the feature across a r...
http://arxiv.org/pdf/2310.15154v1
2023-10-23T17:55:31
2023-10-23T17:55:31
{ "title": [ 0.0014257431, -0.012702942, -0.014678955, -0.06210327, -0.002708435, -0.00040578842, -0.039764404, -0.0010786057, 0.021560669, 0.03414917, -0.0309906, -0.0013237, 0.0020332336, -0.0032634735, -0.05822754, 0.012260437, -0.013847351, 0...
http://arxiv.org/abs/2310.15213v1
Function Vectors in Large Language Models
Eric Todd, Millicent L. Li, Arnab Sen Sharma, Aaron Mueller, Byron C. Wallace, David Bau
cs.CL, cs.LG
We report the presence of a simple neural mechanism that represents an input-output function as a vector within autoregressive transformer language models (LMs). Using causal mediation analysis on a diverse range of in-context-learning (ICL) tasks, we find that a small number attention heads transport a compact represe...
http://arxiv.org/pdf/2310.15213v1
2023-10-23T17:55:24
2023-10-23T17:55:24
{ "title": [ -0.011314392, -0.017715454, -0.028121948, -0.07733154, -0.00573349, 0.0070533752, -0.070007324, 0.004875183, 0.00041413307, 0.023284912, -0.021255493, -0.0033988953, 0.0070877075, -0.023773193, -0.03024292, 0.020584106, 0.00061893463, ...
http://arxiv.org/abs/2310.15151v1
Verb Conjugation in Transformers Is Determined by Linear Encodings of Subject Number
Sophie Hao, Tal Linzen
cs.CL, cs.AI, cs.LG
Deep architectures such as Transformers are sometimes criticized for having uninterpretable "black-box" representations. We use causal intervention analysis to show that, in fact, some linguistic features are represented in a linear, interpretable format. Specifically, we show that BERT's ability to conjugate verbs rel...
http://arxiv.org/pdf/2310.15151v1
2023-10-23T17:53:47
2023-10-23T17:53:47
{ "title": [ 0.019500732, -0.01473999, -0.025146484, -0.036895752, -0.032806396, -0.0033035278, -0.029815674, 0.008140564, -0.0713501, 0.049987793, -0.013114929, 0.02949524, -0.004421234, -0.04232788, 0.038970947, -0.0107803345, -0.0042381287, 0....
http://arxiv.org/abs/2310.15147v1
S3Eval: A Synthetic, Scalable, Systematic Evaluation Suite for Large Language Models
Fangyu Lei, Qian Liu, Yiming Huang, Shizhu He, Jun Zhao, Kang Liu
cs.CL
The rapid development of Large Language Models (LLMs) has led to great strides in model capabilities like reasoning and long-context understanding. However, as LLMs are able to process longer contexts, it becomes more challenging to evaluate whether they have acquired certain capabilities, since the length of text (e.g...
http://arxiv.org/pdf/2310.15147v1
2023-10-23T17:52:06
2023-10-23T17:52:06
{ "title": [ -0.036743164, -0.006160736, -0.04345703, -0.08459473, 0.018218994, 0.015266418, -0.06921387, 0.0035247803, 0.0027179718, 0.025390625, -0.0040740967, -0.0072898865, 0.013397217, -0.024780273, -0.046417236, 0.013114929, -0.021377563, 0...
http://arxiv.org/abs/2310.15141v1
SpecTr: Fast Speculative Decoding via Optimal Transport
Ziteng Sun, Ananda Theertha Suresh, Jae Hun Ro, Ahmad Beirami, Himanshu Jain, Felix Yu
cs.LG, cs.CL, cs.DS, cs.IT, math.IT
Autoregressive sampling from large language models has led to state-of-the-art results in several natural language tasks. However, autoregressive sampling generates tokens one at a time making it slow, and even prohibitive in certain tasks. One way to speed up sampling is $\textit{speculative decoding}$: use a small mo...
http://arxiv.org/pdf/2310.15141v1
2023-10-23T17:47:34
2023-10-23T17:47:34
{ "title": [ -0.05987549, -0.025283813, -0.02017212, -0.05493164, -0.011878967, 0.03036499, -0.023925781, 0.003921509, 0.012199402, 0.054626465, -0.009780884, -0.009605408, 0.012832642, -0.0053215027, -0.059814453, 0.011024475, -0.0011301041, 0.0...
http://arxiv.org/abs/2310.15140v1
AutoDAN: Automatic and Interpretable Adversarial Attacks on Large Language Models
Sicheng Zhu, Ruiyi Zhang, Bang An, Gang Wu, Joe Barrow, Zichao Wang, Furong Huang, Ani Nenkova, Tong Sun
cs.CR, cs.AI, cs.CL, cs.LG
Safety alignment of Large Language Models (LLMs) can be compromised with manual jailbreak attacks and (automatic) adversarial attacks. Recent work suggests that patching LLMs against these attacks is possible: manual jailbreak attacks are human-readable but often limited and public, making them easy to block; adversari...
http://arxiv.org/pdf/2310.15140v1
2023-10-23T17:46:07
2023-10-23T17:46:07
{ "title": [ -0.015113831, -0.015022278, -0.02835083, -0.064208984, 0.01335144, 0.0051612854, -0.040374756, 0.0053596497, -0.029388428, 0.024368286, 0.002670288, 0.018203735, -0.0087890625, -0.015129089, -0.031066895, -0.0026836395, -0.015777588, ...
http://arxiv.org/abs/2310.15135v1
Quantifying the Dialect Gap and its Correlates Across Languages
Anjali Kantharuban, Ivan Vulić, Anna Korhonen
cs.CL, cs.LG
Historically, researchers and consumers have noticed a decrease in quality when applying NLP tools to minority variants of languages (i.e. Puerto Rican Spanish or Swiss German), but studies exploring this have been limited to a select few languages. Additionally, past studies have mainly been conducted in a monolingual...
http://arxiv.org/pdf/2310.15135v1
2023-10-23T17:42:01
2023-10-23T17:42:01
{ "title": [ -0.008926392, 0.016220093, -0.039093018, -0.026931763, -0.0103302, -0.025375366, -0.033081055, -0.021316528, -0.024154663, 0.036071777, -0.035186768, -0.0022411346, 0.0015029907, -0.040100098, -0.01953125, 0.0038204193, 0.007896423, ...
http://arxiv.org/abs/2310.15129v1
Location-Aware Visual Question Generation with Lightweight Models
Nicholas Collin Suwono, Justin Chih-Yao Chen, Tun Min Hung, Ting-Hao Kenneth Huang, I-Bin Liao, Yung-Hui Li, Lun-Wei Ku, Shao-Hua Sun
cs.CL, cs.LG
This work introduces a novel task, location-aware visual question generation (LocaVQG), which aims to generate engaging questions from data relevant to a particular geographical location. Specifically, we represent such location-aware information with surrounding images and a GPS coordinate. To tackle this task, we pre...
http://arxiv.org/pdf/2310.15129v1
2023-10-23T17:33:31
2023-10-23T17:33:31
{ "title": [ 0.008148193, -0.0037326813, -0.010345459, -0.029663086, -0.011688232, 0.0077667236, -0.010566711, -0.023803711, 0.035705566, 0.01965332, -0.03414917, -0.015106201, -0.021347046, -0.018081665, -0.052947998, -0.015037537, 0.017166138, ...
http://arxiv.org/abs/2310.15127v1
Open-Ended Instructable Embodied Agents with Memory-Augmented Large Language Models
Gabriel Sarch, Yue Wu, Michael J. Tarr, Katerina Fragkiadaki
cs.AI, cs.CL, cs.LG, cs.RO
Pre-trained and frozen LLMs can effectively map simple scene re-arrangement instructions to programs over a robot's visuomotor functions through appropriate few-shot example prompting. To parse open-domain natural language and adapt to a user's idiosyncratic procedures, not known during prompt engineering time, fixed p...
http://arxiv.org/pdf/2310.15127v1
2023-10-23T17:31:55
2023-10-23T17:31:55
{ "title": [ -0.010398865, -0.012535095, -0.003358841, -0.046203613, -0.011428833, 0.008468628, -0.027832031, -0.016555786, 0.012825012, 0.028747559, -0.023834229, 0.0016555786, 0.0021572113, 0.00018930435, -0.04714966, -0.0025024414, 0.0027332306, ...
http://arxiv.org/abs/2310.15123v1
Branch-Solve-Merge Improves Large Language Model Evaluation and Generation
Swarnadeep Saha, Omer Levy, Asli Celikyilmaz, Mohit Bansal, Jason Weston, Xian Li
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) are frequently used for multi-faceted language generation and evaluation tasks that involve satisfying intricate user constraints or taking into account multiple aspects and criteria. However, their performance can fall short, due to the model's lack of coherence and inability to plan and d...
http://arxiv.org/pdf/2310.15123v1
2023-10-23T17:29:48
2023-10-23T17:29:48
{ "title": [ -0.01940918, -0.021972656, -0.046447754, -0.09423828, -0.020217896, 0.012466431, -0.09362793, 0.01979065, -0.00415802, 0.021575928, -0.031021118, 0.0072250366, 0.00920105, -0.020492554, -0.017608643, 0.006324768, -0.006462097, 0.0200...
http://arxiv.org/abs/2310.15117v1
Causal Inference Using LLM-Guided Discovery
Aniket Vashishtha, Abbavaram Gowtham Reddy, Abhinav Kumar, Saketh Bachu, Vineeth N Balasubramanian, Amit Sharma
cs.AI, cs.CL
At the core of causal inference lies the challenge of determining reliable causal graphs solely based on observational data. Since the well-known backdoor criterion depends on the graph, any errors in the graph can propagate downstream to effect inference. In this work, we initially show that complete graph information...
http://arxiv.org/pdf/2310.15117v1
2023-10-23T17:23:56
2023-10-23T17:23:56
{ "title": [ -0.0647583, -0.021652222, -0.071899414, -0.07324219, -0.021133423, 0.001282692, -0.0803833, 0.03338623, -0.025665283, -0.015487671, -0.03274536, 0.039001465, -0.046936035, -0.032104492, 0.018951416, -0.011917114, -0.008880615, 0.0397...
http://arxiv.org/abs/2310.15114v1
How To Build Competitive Multi-gender Speech Translation Models For Controlling Speaker Gender Translation
Marco Gaido, Dennis Fucci, Matteo Negri, Luisa Bentivogli
cs.CL
When translating from notional gender languages (e.g., English) into grammatical gender languages (e.g., Italian), the generated translation requires explicit gender assignments for various words, including those referring to the speaker. When the source sentence does not convey the speaker's gender, speech translation...
http://arxiv.org/pdf/2310.15114v1
2023-10-23T17:21:32
2023-10-23T17:21:32
{ "title": [ -0.016433716, 0.009178162, -0.038238525, -0.029953003, -0.006629944, -0.024780273, -0.023071289, 0.013084412, -0.04663086, 0.076416016, -0.0023040771, -0.009162903, 0.048583984, -0.02217102, -0.02822876, -0.007129669, -0.012588501, 0...
http://arxiv.org/abs/2310.15113v2
Counting the Bugs in ChatGPT's Wugs: A Multilingual Investigation into the Morphological Capabilities of a Large Language Model
Leonie Weissweiler, Valentin Hofmann, Anjali Kantharuban, Anna Cai, Ritam Dutt, Amey Hengle, Anubha Kabra, Atharva Kulkarni, Abhishek Vijayakumar, Haofei Yu, Hinrich Schütze, Kemal Oflazer, David R. Mortensen
cs.CL
Large language models (LLMs) have recently reached an impressive level of linguistic capability, prompting comparisons with human language skills. However, there have been relatively few systematic inquiries into the linguistic capabilities of the latest generation of LLMs, and those studies that do exist (i) ignore th...
http://arxiv.org/pdf/2310.15113v2
2023-10-26T11:45:28
2023-10-23T17:21:03
{ "title": [ -0.022369385, -0.011131287, -0.017608643, -0.0075302124, -0.017608643, -0.0037555695, -0.038848877, -0.014717102, -0.029418945, 0.053771973, -0.000767231, 0.018188477, 0.016326904, -0.014892578, -0.03640747, 0.019256592, -0.038391113, ...
http://arxiv.org/abs/2310.15109v1
GRENADE: Graph-Centric Language Model for Self-Supervised Representation Learning on Text-Attributed Graphs
Yichuan Li, Kaize Ding, Kyumin Lee
cs.CL
Self-supervised representation learning on text-attributed graphs, which aims to create expressive and generalizable representations for various downstream tasks, has received increasing research attention lately. However, existing methods either struggle to capture the full extent of structural context information or ...
http://arxiv.org/pdf/2310.15109v1
2023-10-23T17:18:35
2023-10-23T17:18:35
{ "title": [ -0.03253174, -0.019638062, -0.051483154, -0.046875, -0.02861023, 0.0059509277, -0.06524658, 0.023803711, -0.0048561096, 0.02468872, 0.002872467, 0.0149383545, -0.01121521, -0.026367188, -0.021972656, 0.011451721, -0.0077285767, 0.061...
http://arxiv.org/abs/2310.15100v1
LLM-in-the-loop: Leveraging Large Language Model for Thematic Analysis
Shih-Chieh Dai, Aiping Xiong, Lun-Wei Ku
cs.CL
Thematic analysis (TA) has been widely used for analyzing qualitative data in many disciplines and fields. To ensure reliable analysis, the same piece of data is typically assigned to at least two human coders. Moreover, to produce meaningful and useful analysis, human coders develop and deepen their data interpretatio...
http://arxiv.org/pdf/2310.15100v1
2023-10-23T17:05:59
2023-10-23T17:05:59
{ "title": [ -0.012023926, -0.014511108, -0.03543091, -0.045166016, -0.0012378693, -0.006210327, -0.042907715, -0.022720337, 0.015449524, 0.028457642, -0.012489319, 0.0042800903, -0.009559631, -0.020477295, -0.04559326, 0.010467529, -0.009887695, ...
http://arxiv.org/abs/2310.15080v2
Federated Learning of Large Language Models with Parameter-Efficient Prompt Tuning and Adaptive Optimization
Tianshi Che, Ji Liu, Yang Zhou, Jiaxiang Ren, Jiwen Zhou, Victor S. Sheng, Huaiyu Dai, Dejing Dou
cs.LG, cs.CL, cs.DC
Federated learning (FL) is a promising paradigm to enable collaborative model training with decentralized data. However, the training process of Large Language Models (LLMs) generally incurs the update of significant parameters, which limits the applicability of FL techniques to tackle the LLMs in real scenarios. Promp...
http://arxiv.org/pdf/2310.15080v2
2023-10-29T07:17:45
2023-10-23T16:37:59
{ "title": [ -0.028457642, 0.004901886, -0.055877686, -0.097351074, 0.012527466, 0.01600647, -0.064575195, 0.030822754, -0.017929077, 0.0309906, -0.022903442, 0.011756897, -0.008743286, -0.0016965866, -0.02557373, -0.008674622, -0.0035877228, 0.0...
http://arxiv.org/abs/2310.15079v1
Affective and Dynamic Beam Search for Story Generation
Tenghao Huang, Ehsan Qasemi, Bangzheng Li, He Wang, Faeze Brahman, Muhao Chen, Snigdha Chaturvedi
cs.CL
Storytelling's captivating potential makes it a fascinating research area, with implications for entertainment, education, therapy, and cognitive studies. In this paper, we propose Affective Story Generator (AffGen) for generating interesting narratives. AffGen introduces "intriguing twists" in narratives by employing ...
http://arxiv.org/pdf/2310.15079v1
2023-10-23T16:37:14
2023-10-23T16:37:14
{ "title": [ -0.026168823, -0.019851685, -0.011474609, -0.024429321, -0.03668213, 0.010635376, -0.01210022, -0.009963989, 0.013755798, 0.02519226, -0.02708435, -0.011756897, -0.0052223206, -0.016052246, -0.047454834, -0.007789612, -0.026351929, 0...
http://arxiv.org/abs/2310.15077v1
'Don't Get Too Technical with Me': A Discourse Structure-Based Framework for Science Journalism
Ronald Cardenas, Bingsheng Yao, Dakuo Wang, Yufang Hou
cs.CL
Science journalism refers to the task of reporting technical findings of a scientific paper as a less technical news article to the general public audience. We aim to design an automated system to support this real-world task (i.e., automatic science journalism) by 1) introducing a newly-constructed and real-world data...
http://arxiv.org/pdf/2310.15077v1
2023-10-23T16:35:05
2023-10-23T16:35:05
{ "title": [ -0.041931152, -0.019256592, -0.0440979, -0.034973145, -0.0016727448, -0.008895874, -0.012168884, -0.00007736683, -0.037353516, 0.019500732, 0.035003662, 0.012084961, -0.0670166, -0.030181885, -0.047058105, -0.008636475, -0.029205322, ...
http://arxiv.org/abs/2310.15075v1
TableQAKit: A Comprehensive and Practical Toolkit for Table-based Question Answering
Fangyu Lei, Tongxu Luo, Pengqi Yang, Weihao Liu, Hanwen Liu, Jiahe Lei, Yiming Huang, Yifan Wei, Shizhu He, Jun Zhao, Kang Liu
cs.CL
Table-based question answering (TableQA) is an important task in natural language processing, which requires comprehending tables and employing various reasoning ways to answer the questions. This paper introduces TableQAKit, the first comprehensive toolkit designed specifically for TableQA. The toolkit designs a unifi...
http://arxiv.org/pdf/2310.15075v1
2023-10-23T16:33:23
2023-10-23T16:33:23
{ "title": [ -0.025619507, -0.010787964, -0.041168213, -0.060943604, 0.014381409, 0.011383057, -0.043823242, -0.008842468, 0.01689148, 0.028900146, -0.027069092, 0.008598328, -0.00079774857, -0.000685215, -0.05960083, -0.0003414154, 0.002603531, ...
http://arxiv.org/abs/2310.15066v1
Localizing Active Objects from Egocentric Vision with Symbolic World Knowledge
Te-Lin Wu, Yu Zhou, Nanyun Peng
cs.CV, cs.CL
The ability to actively ground task instructions from an egocentric view is crucial for AI agents to accomplish tasks or assist humans virtually. One important step towards this goal is to localize and track key active objects that undergo major state change as a consequence of human actions/interactions to the environ...
http://arxiv.org/pdf/2310.15066v1
2023-10-23T16:14:05
2023-10-23T16:14:05
{ "title": [ -0.009239197, 0.009315491, -0.0015296936, 0.0027275085, -0.00041770935, -0.017868042, 0.018844604, -0.06774902, -0.00055265427, 0.037506104, -0.041046143, -0.012290955, 0.0053863525, -0.016052246, -0.056274414, -0.010932922, 0.029937744,...
http://arxiv.org/abs/2310.15061v1
The BLA Benchmark: Investigating Basic Language Abilities of Pre-Trained Multimodal Models
Xinyi Chen, Raquel Fernández, Sandro Pezzelle
cs.CL, cs.AI, cs.CV
Despite the impressive performance achieved by pre-trained language-and-vision models in downstream tasks, it remains an open question whether this reflects a proper understanding of image-text interaction. In this work, we explore to what extent they handle basic linguistic constructions -- active-passive voice, coord...
http://arxiv.org/pdf/2310.15061v1
2023-10-23T16:05:13
2023-10-23T16:05:13
{ "title": [ -0.026443481, -0.0006098747, -0.045806885, -0.055358887, -0.01360321, -0.010276794, -0.071777344, -0.013839722, 0.0021190643, 0.021911621, 0.0055618286, 0.010108948, -0.012077332, -0.029449463, -0.03173828, 0.013908386, -0.01184845, ...
http://arxiv.org/abs/2310.15055v1
Towards Conceptualization of "Fair Explanation": Disparate Impacts of anti-Asian Hate Speech Explanations on Content Moderators
Tin Nguyen, Jiannan Xu, Aayushi Roy, Hal Daumé III, Marine Carpuat
cs.CL, cs.AI, cs.HC
Recent research at the intersection of AI explainability and fairness has focused on how explanations can improve human-plus-AI task performance as assessed by fairness measures. We propose to characterize what constitutes an explanation that is itself "fair" -- an explanation that does not adversely impact specific po...
http://arxiv.org/pdf/2310.15055v1
2023-10-23T15:57:41
2023-10-23T15:57:41
{ "title": [ 0.04345703, -0.0022182465, -0.005382538, -0.064086914, 0.010917664, -0.021240234, 0.01335907, -0.0025100708, -0.041870117, 0.033813477, 0.0059051514, 0.014213562, 0.017211914, -0.038848877, -0.024993896, -0.013473511, -0.014694214, 0...
http://arxiv.org/abs/2310.15040v1
SLOG: A Structural Generalization Benchmark for Semantic Parsing
Bingzhi Li, Lucia Donatelli, Alexander Koller, Tal Linzen, Yuekun Yao, Najoung Kim
cs.CL
The goal of compositional generalization benchmarks is to evaluate how well models generalize to new complex linguistic expressions. Existing benchmarks often focus on lexical generalization, the interpretation of novel lexical items in syntactic structures familiar from training; structural generalization tasks, where...
http://arxiv.org/pdf/2310.15040v1
2023-10-23T15:39:09
2023-10-23T15:39:09
{ "title": [ -0.024230957, -0.004917145, -0.025405884, -0.094055176, 0.0059127808, 0.025253296, -0.07507324, 0.006340027, 0.0001885891, 0.021987915, -0.03237915, 0.026763916, -0.015655518, 0.004310608, -0.036834717, -0.0033721924, -0.01184082, 0....
http://arxiv.org/abs/2310.15021v1
Efficient Data Learning for Open Information Extraction with Pre-trained Language Models
Zhiyuan Fan, Shizhu He
cs.CL, cs.AI
Open Information Extraction (OpenIE) is a fundamental yet challenging task in Natural Language Processing, which involves extracting all triples (subject, predicate, object) from a given sentence. While labeling-based methods have their merits, generation-based techniques offer unique advantages, such as the ability to...
http://arxiv.org/pdf/2310.15021v1
2023-10-23T15:19:24
2023-10-23T15:19:24
{ "title": [ -0.030090332, -0.0061035156, -0.044036865, -0.08508301, 0.004245758, 0.0063438416, -0.0680542, -0.0009469986, 0.0023479462, 0.022094727, -0.026473999, 0.004966736, -0.016677856, -0.0047302246, -0.040039062, -0.0021457672, -0.003282547, ...
http://arxiv.org/abs/2310.15019v2
Meta learning with language models: Challenges and opportunities in the classification of imbalanced text
Apostol Vassilev, Honglan Jin, Munawar Hasan
cs.LG, cs.AI, cs.CL
Detecting out of policy speech (OOPS) content is important but difficult. While machine learning is a powerful tool to tackle this challenging task, it is hard to break the performance ceiling due to factors like quantity and quality limitations on training data and inconsistencies in OOPS definition and data labeling....
http://arxiv.org/pdf/2310.15019v2
2023-10-24T15:15:38
2023-10-23T15:14:55
{ "title": [ -0.02949524, 0.01158905, -0.023849487, -0.030258179, -0.010673523, 0.01876831, -0.07171631, -0.045288086, 0.036010742, 0.045715332, -0.027297974, 0.0052452087, 0.02178955, -0.007827759, -0.06616211, 0.0079422, -0.03643799, 0.08166504...
http://arxiv.org/abs/2310.15010v1
Statistical Depth for Ranking and Characterizing Transformer-Based Text Embeddings
Parker Seegmiller, Sarah Masud Preum
cs.CL
The popularity of transformer-based text embeddings calls for better statistical tools for measuring distributions of such embeddings. One such tool would be a method for ranking texts within a corpus by centrality, i.e. assigning each text a number signifying how representative that text is of the corpus as a whole. H...
http://arxiv.org/pdf/2310.15010v1
2023-10-23T15:02:44
2023-10-23T15:02:44
{ "title": [ 0.0018730164, 0.018585205, -0.035491943, -0.029037476, 0.02659607, -0.0096588135, -0.031280518, -0.02482605, 0.0042495728, 0.02508545, -0.011894226, 0.022216797, 0.0035324097, -0.010673523, -0.02960205, 0.0017299652, -0.011909485, 0....
http://arxiv.org/abs/2310.15007v1
Did the Neurons Read your Book? Document-level Membership Inference for Large Language Models
Matthieu Meeus, Shubham Jain, Marek Rei, Yves-Alexandre de Montjoye
cs.CL, cs.CR, cs.LG
With large language models (LLMs) poised to become embedded in our daily lives, questions are starting to be raised about the dataset(s) they learned from. These questions range from potential bias or misinformation LLMs could retain from their training data to questions of copyright and fair use of human-generated tex...
http://arxiv.org/pdf/2310.15007v1
2023-10-23T15:00:46
2023-10-23T15:00:46
{ "title": [ -0.024536133, -0.017532349, -0.034820557, -0.0574646, -0.016311646, 0.0021648407, -0.06781006, 0.002325058, 0.010147095, 0.004890442, -0.032989502, 0.024414062, 0.008155823, -0.005859375, -0.00982666, 0.0059928894, 0.004196167, 0.040...
http://arxiv.org/abs/2310.15004v1
When Language Models Fall in Love: Animacy Processing in Transformer Language Models
Michael Hanna, Yonatan Belinkov, Sandro Pezzelle
cs.CL
Animacy - whether an entity is alive and sentient - is fundamental to cognitive processing, impacting areas such as memory, vision, and language. However, animacy is not always expressed directly in language: in English it often manifests indirectly, in the form of selectional constraints on verbs and adjectives. This ...
http://arxiv.org/pdf/2310.15004v1
2023-10-23T14:57:52
2023-10-23T14:57:52
{ "title": [ -0.015129089, 0.0047950745, -0.021133423, -0.046936035, 0.0077285767, -0.0017986298, -0.037506104, -0.008590698, 0.020889282, 0.03289795, -0.02619934, 0.0043640137, 0.054504395, -0.022460938, -0.033691406, 0.026565552, -0.01637268, 0...
http://arxiv.org/abs/2310.14997v1
Simple Hardware-Efficient PCFGs with Independent Left and Right Productions
Wei Liu, Songlin Yang, Yoon Kim, Kewei Tu
cs.CL, cs.LG
Scaling dense PCFGs to thousands of nonterminals via a low-rank parameterization of the rule probability tensor has been shown to be beneficial for unsupervised parsing. However, PCFGs scaled this way still perform poorly as a language model, and even underperform similarly-sized HMMs. This work introduces \emph{Simple...
http://arxiv.org/pdf/2310.14997v1
2023-10-23T14:48:51
2023-10-23T14:48:51
{ "title": [ -0.014122009, -0.034576416, -0.048461914, -0.092163086, -0.013198853, 0.0019245148, -0.050994873, 0.046539307, -0.012008667, 0.062805176, -0.007572174, 0.016235352, -0.0068588257, -0.00062417984, -0.06616211, 0.0047416687, 0.001906395, ...
http://arxiv.org/abs/2310.14993v1
Understanding the Inner Workings of Language Models Through Representation Dissimilarity
Davis Brown, Charles Godfrey, Nicholas Konz, Jonathan Tu, Henry Kvinge
cs.LG, cs.AI, cs.CL
As language models are applied to an increasing number of real-world applications, understanding their inner workings has become an important issue in model trust, interpretability, and transparency. In this work we show that representation dissimilarity measures, which are functions that measure the extent to which tw...
http://arxiv.org/pdf/2310.14993v1
2023-10-23T14:46:20
2023-10-23T14:46:20
{ "title": [ -0.012992859, 0.004962921, -0.036376953, -0.064697266, -0.0008330345, -0.005580902, -0.05557251, -0.0033435822, 0.0024719238, 0.025848389, -0.03302002, -0.0006222725, 0.029083252, -0.015991211, -0.035064697, 0.0027942657, 0.0013456345, ...
http://arxiv.org/abs/2310.14985v1
LLM-Based Agent Society Investigation: Collaboration and Confrontation in Avalon Gameplay
Yihuai Lan, Zhiqiang Hu, Lei Wang, Yang Wang, Deheng Ye, Peilin Zhao, Ee-Peng Lim, Hui Xiong, Hao Wang
cs.CL
This paper aims to investigate the open research problem of uncovering the social behaviors of LLM-based agents. To achieve this goal, we adopt Avalon, a representative communication game, as the environment and use system prompts to guide LLM agents to play the game. While previous studies have conducted preliminary i...
http://arxiv.org/pdf/2310.14985v1
2023-10-23T14:35:26
2023-10-23T14:35:26
{ "title": [ -0.002439499, 0.008476257, -0.04763794, -0.017578125, -0.030899048, -0.02368164, -0.021728516, -0.0043525696, -0.023010254, -0.008682251, -0.03579712, 0.031585693, -0.038238525, -0.035980225, 0.024291992, -0.03012085, -0.024780273, 0...
http://arxiv.org/abs/2310.14981v1
Fidelity-Enriched Contrastive Search: Reconciling the Faithfulness-Diversity Trade-Off in Text Generation
Wei-Lin Chen, Cheng-Kuang Wu, Hsin-Hsi Chen, Chung-Chi Chen
cs.CL
In this paper, we address the hallucination problem commonly found in natural language generation tasks. Language models often generate fluent and convincing content but can lack consistency with the provided source, resulting in potential inaccuracies. We propose a new decoding method called Fidelity-Enriched Contrast...
http://arxiv.org/pdf/2310.14981v1
2023-10-23T14:27:45
2023-10-23T14:27:45
{ "title": [ -0.0016994476, 0.009338379, -0.022949219, -0.051330566, 0.0061836243, -0.009140015, -0.03869629, -0.034057617, 0.019973755, 0.06707764, 0.007522583, 0.021636963, 0.012466431, -0.011497498, -0.084228516, 0.013175964, -0.025405884, 0.0...
http://arxiv.org/abs/2310.14979v1
ACTOR: Active Learning with Annotator-specific Classification Heads to Embrace Human Label Variation
Xinpeng Wang, Barbara Plank
cs.CL, cs.AI, cs.LG
Label aggregation such as majority voting is commonly used to resolve annotator disagreement in dataset creation. However, this may disregard minority values and opinions. Recent studies indicate that learning from individual annotations outperforms learning from aggregated labels, though they require a considerable am...
http://arxiv.org/pdf/2310.14979v1
2023-10-23T14:26:43
2023-10-23T14:26:43
{ "title": [ -0.007637024, 0.02142334, -0.033203125, -0.044158936, 0.0030765533, -0.02859497, -0.021133423, -0.016830444, -0.012718201, 0.022003174, 0.00025439262, 0.030212402, -0.03741455, -0.031021118, -0.016342163, 0.0014686584, -0.0037193298, ...
http://arxiv.org/abs/2310.14971v1
Penalty Decoding: Well Suppress the Self-Reinforcement Effect in Open-Ended Text Generation
Wenhong Zhu, Hongkun Hao, Rui Wang
cs.CL
The decoding algorithm is critical for open-ended text generation, transforming latent representations into coherent and meaningful outputs. This paper investigates the self-reinforcement effect in text generation and the effectiveness of a repetition penalty to mitigate it. However, determining the optimal repetition ...
http://arxiv.org/pdf/2310.14971v1
2023-10-23T14:20:04
2023-10-23T14:20:04
{ "title": [ -0.0028457642, -0.0077552795, 0.002412796, -0.0022125244, -0.035308838, -0.014259338, -0.012763977, -0.043792725, -0.043395996, 0.03225708, 0.017623901, 0.023971558, -0.009613037, -0.048858643, -0.009475708, -0.004135132, -0.016845703, ...
http://arxiv.org/abs/2310.14970v1
Towards LLM-driven Dialogue State Tracking
Yujie Feng, Zexin Lu, Bo Liu, Liming Zhan, Xiao-Ming Wu
cs.CL
Dialogue State Tracking (DST) is of paramount importance in ensuring accurate tracking of user goals and system actions within task-oriented dialogue systems. The emergence of large language models (LLMs) such as GPT3 and ChatGPT has sparked considerable interest in assessing their efficacy across diverse applications....
http://arxiv.org/pdf/2310.14970v1
2023-10-23T14:15:28
2023-10-23T14:15:28
{ "title": [ -0.027053833, -0.018844604, -0.039733887, -0.07330322, -0.023834229, 0.009468079, -0.0791626, 0.009712219, 0.0015048981, 0.009902954, 0.0014295578, 0.015655518, -0.034057617, -0.035491943, -0.017166138, 0.0018882751, -0.027557373, 0....
http://arxiv.org/abs/2310.14954v2
Key Frame Mechanism For Efficient Conformer Based End-to-end Speech Recognition
Peng Fan, Changhao Shan, Sining Sun, Qing Yang, Jianwei Zhang
cs.SD, cs.CL, eess.AS
Recently, Conformer as a backbone network for end-to-end automatic speech recognition achieved state-of-the-art performance. The Conformer block leverages a self-attention mechanism to capture global information, along with a convolutional neural network to capture local information, resulting in improved performance. ...
http://arxiv.org/pdf/2310.14954v2
2023-10-28T14:38:46
2023-10-23T13:55:49
{ "title": [ -0.033050537, 0.0027179718, -0.053588867, -0.08917236, 0.021209717, 0.010803223, -0.050201416, 0.030044556, -0.015777588, 0.05105591, -0.003063202, 0.011779785, 0.008926392, -0.017303467, -0.039611816, 0.014976501, -0.017578125, 0.04...
http://arxiv.org/abs/2310.14947v1
System Combination via Quality Estimation for Grammatical Error Correction
Muhammad Reza Qorib, Hwee Tou Ng
cs.CL
Quality estimation models have been developed to assess the corrections made by grammatical error correction (GEC) models when the reference or gold-standard corrections are not available. An ideal quality estimator can be utilized to combine the outputs of multiple GEC systems by choosing the best subset of edits from...
http://arxiv.org/pdf/2310.14947v1
2023-10-23T13:46:49
2023-10-23T13:46:49
{ "title": [ -0.0077552795, 0.004737854, -0.015541077, -0.043823242, -0.009140015, -0.0040397644, -0.043670654, -0.020523071, 0.00907135, 0.039398193, -0.030349731, -0.0046195984, 0.0032176971, -0.00674057, -0.054992676, -0.009033203, -0.032318115, ...
http://arxiv.org/abs/2310.14928v1
Unveiling A Core Linguistic Region in Large Language Models
Jun Zhao, Zhihao Zhang, Yide Ma, Qi Zhang, Tao Gui, Luhui Gao, Xuanjing Huang
cs.CL
Brain localization, which describes the association between specific regions of the brain and their corresponding functions, is widely accepted in the field of cognitive science as an objective fact. Today's large language models (LLMs) possess human-level linguistic competence and can execute complex tasks requiring a...
http://arxiv.org/pdf/2310.14928v1
2023-10-23T13:31:32
2023-10-23T13:31:32
{ "title": [ -0.040985107, -0.0029678345, -0.045410156, -0.041259766, 0.0011606216, -0.013160706, -0.037841797, 0.008369446, 0.007911682, 0.040893555, -0.0060310364, 0.010513306, 0.0018968582, -0.01322937, -0.05618286, 0.023849487, -0.012931824, ...
http://arxiv.org/abs/2310.14921v1
PartialFormer: Modeling Part Instead of Whole
Tong Zheng, Bei Li, Huiwen Bao, Weiqiao Shan, Tong Xiao, Jingbo Zhu
cs.CL, cs.AI
The design choices in Transformer feed-forward neural networks have resulted in significant computational and parameter overhead. In this work, we emphasize the importance of hidden dimension in designing lightweight FFNs, a factor often overlooked in previous architectures. Guided by this principle, we introduce Parti...
http://arxiv.org/pdf/2310.14921v1
2023-10-23T13:25:54
2023-10-23T13:25:54
{ "title": [ -0.002565384, -0.020462036, -0.067993164, -0.06903076, -0.02079773, -0.0317688, -0.05316162, 0.005077362, -0.039367676, 0.02999878, 0.0068244934, -0.021133423, 0.020065308, -0.078186035, 0.007106781, -0.008880615, 0.023117065, 0.0167...
http://arxiv.org/abs/2310.14909v1
Linking Surface Facts to Large-Scale Knowledge Graphs
Gorjan Radevski, Kiril Gashteovski, Chia-Chien Hung, Carolin Lawrence, Goran Glavaš
cs.CL, cs.AI, cs.LG
Open Information Extraction (OIE) methods extract facts from natural language text in the form of ("subject"; "relation"; "object") triples. These facts are, however, merely surface forms, the ambiguity of which impedes their downstream usage; e.g., the surface phrase "Michael Jordan" may refer to either the former bas...
http://arxiv.org/pdf/2310.14909v1
2023-10-23T13:18:49
2023-10-23T13:18:49
{ "title": [ -0.021759033, -0.015579224, -0.055236816, -0.09338379, -0.0062789917, 0.009048462, -0.06213379, 0.015403748, 0.0010919571, 0.018447876, -0.036254883, 0.011779785, -0.021133423, -0.013916016, -0.0143966675, -0.009727478, 0.008598328, ...
http://arxiv.org/abs/2310.14892v3
Air-Decoding: Attribute Distribution Reconstruction for Decoding-Time Controllable Text Generation
Tianqi Zhong, Quan Wang, Jingxuan Han, Yongdong Zhang, Zhendong Mao
cs.CL
Controllable text generation (CTG) aims to generate text with desired attributes, and decoding-time-based methods have shown promising performance on this task. However, in this paper, we identify the phenomenon of Attribute Collapse for the first time. It causes the fluency of generated text to rapidly decrease when t...
http://arxiv.org/pdf/2310.14892v3
2023-11-02T03:14:47
2023-10-23T12:59:11
{ "title": [ -0.059906006, -0.005622864, -0.034973145, -0.04333496, -0.016723633, 0.021697998, -0.06335449, -0.017562866, -0.012054443, 0.055023193, 0.0013113022, 0.014961243, 0.02041626, -0.010505676, -0.050201416, 0.01953125, -0.020401001, 0.05...
http://arxiv.org/abs/2310.14883v1
Non-autoregressive Streaming Transformer for Simultaneous Translation
Zhengrui Ma, Shaolei Zhang, Shoutao Guo, Chenze Shao, Min Zhang, Yang Feng
cs.CL, cs.AI, I.2.7
Simultaneous machine translation (SiMT) models are trained to strike a balance between latency and translation quality. However, training these models to achieve high quality while maintaining low latency often leads to a tendency for aggressive anticipation. We argue that such issue stems from the autoregressive archi...
http://arxiv.org/pdf/2310.14883v1
2023-10-23T12:52:24
2023-10-23T12:52:24
{ "title": [ -0.02142334, 0.015777588, -0.04321289, -0.07562256, -0.003768921, 0.00819397, -0.05734253, 0.0021076202, -0.011306763, 0.08270264, -0.0042304993, 0.005783081, 0.031066895, -0.014152527, -0.05026245, 0.020477295, 0.023651123, 0.011871...
http://arxiv.org/abs/2310.14880v2
Can ChatGPT Perform Reasoning Using the IRAC Method in Analyzing Legal Scenarios Like a Lawyer?
Xiaoxi Kang, Lizhen Qu, Lay-Ki Soon, Adnan Trakic, Terry Yue Zhuo, Patrick Charles Emerton, Genevieve Grant
cs.CL
Large Language Models (LLMs), such as ChatGPT, have drawn a lot of attentions recently in the legal domain due to its emergent ability to tackle a variety of legal tasks. However, it is still unknown if LLMs are able to analyze a legal case and perform reasoning in the same manner as lawyers. Therefore, we constructed ...
http://arxiv.org/pdf/2310.14880v2
2023-11-03T03:50:07
2023-10-23T12:51:49
{ "title": [ -0.008399963, -0.03366089, 0.017623901, 0.010353088, -0.06390381, 0.023422241, -0.038879395, -0.02267456, -0.012931824, 0.01096344, -0.021774292, 0.0068244934, -0.024597168, -0.023864746, -0.0037384033, -0.027816772, -0.019042969, 0....
http://arxiv.org/abs/2310.14870v1
We are Who We Cite: Bridges of Influence Between Natural Language Processing and Other Academic Fields
Jan Philip Wahle, Terry Ruas, Mohamed Abdalla, Bela Gipp, Saif M. Mohammad
cs.CL, cs.DL
Natural Language Processing (NLP) is poised to substantially influence the world. However, significant progress comes hand-in-hand with substantial risks. Addressing them requires broad engagement with various fields of study. Yet, little empirical work examines the state of such engagement (past or current). In this p...
http://arxiv.org/pdf/2310.14870v1
2023-10-23T12:42:06
2023-10-23T12:42:06
{ "title": [ -0.015602112, 0.005355835, -0.047607422, -0.032714844, 0.005504608, -0.026672363, -0.0012865067, -0.0178833, -0.037872314, 0.023666382, -0.032470703, 0.010978699, 0.016143799, -0.016220093, -0.02168274, -0.017700195, -0.02078247, 0.0...
http://arxiv.org/abs/2310.14868v1
Assessing Step-by-Step Reasoning against Lexical Negation: A Case Study on Syllogism
Mengyu Ye, Tatsuki Kuribayashi, Jun Suzuki, Goro Kobayashi, Hiroaki Funayama
cs.CL
Large language models (LLMs) take advantage of step-by-step reasoning instructions, e.g., chain-of-thought (CoT) prompting. Building on this, their ability to perform CoT-style reasoning robustly is of interest from a probing perspective. In this study, we inspect the step-by-step reasoning ability of LLMs with a focus...
http://arxiv.org/pdf/2310.14868v1
2023-10-23T12:40:41
2023-10-23T12:40:41
{ "title": [ 0.00069093704, 0.0007648468, -0.018630981, -0.09069824, -0.00349617, 0.012901306, -0.053985596, -0.0017948151, -0.009727478, 0.02571106, -0.041625977, 0.0018062592, -0.008842468, -0.012702942, -0.029067993, -0.016967773, -0.010765076, ...
http://arxiv.org/abs/2310.14863v1
Paraphrase Types for Generation and Detection
Jan Philip Wahle, Bela Gipp, Terry Ruas
cs.CL
Current approaches in paraphrase generation and detection heavily rely on a single general similarity score, ignoring the intricate linguistic properties of language. This paper introduces two new tasks to address this shortcoming by considering paraphrase types - specific linguistic perturbations at particular text po...
http://arxiv.org/pdf/2310.14863v1
2023-10-23T12:32:41
2023-10-23T12:32:41
{ "title": [ -0.0025367737, -0.0061035156, -0.021102905, -0.04724121, -0.024383545, -0.0051879883, -0.038848877, -0.007537842, -0.02444458, 0.05429077, -0.01247406, -0.014167786, 0.022628784, -0.032836914, -0.015289307, 0.015304565, -0.015914917, ...
http://arxiv.org/abs/2310.14859v2
3M-TRANSFORMER: A Multi-Stage Multi-Stream Multimodal Transformer for Embodied Turn-Taking Prediction
Mehdi Fatan, Emanuele Mincato, Dimitra Pintzou, Mariella Dimiccoli
cs.CV, cs.CL
Predicting turn-taking in multiparty conversations has many practical applications in human-computer/robot interaction. However, the complexity of human communication makes it a challenging task. Recent advances have shown that synchronous multi-perspective egocentric data can significantly improve turn-taking predicti...
http://arxiv.org/pdf/2310.14859v2
2023-11-01T15:31:42
2023-10-23T12:29:10
{ "title": [ 0.014305115, 0.009407043, -0.02760315, -0.060760498, -0.015686035, -0.011764526, -0.045135498, -0.013977051, -0.0033016205, 0.037078857, -0.027664185, 0.0121154785, 0.023513794, -0.016464233, -0.021453857, -0.0063705444, -0.0026893616, ...
http://arxiv.org/abs/2310.14855v1
Contextual Refinement of Translations: Large Language Models for Sentence and Document-Level Post-Editing
Sai Koneru, Miriam Exel, Matthias Huck, Jan Niehues
cs.CL, cs.AI
Large Language Models (LLM's) have demonstrated considerable success in various Natural Language Processing tasks, but they have yet to attain state-of-the-art performance in Neural Machine Translation (NMT). Nevertheless, their significant performance in tasks demanding a broad understanding and contextual processing ...
http://arxiv.org/pdf/2310.14855v1
2023-10-23T12:22:15
2023-10-23T12:22:15
{ "title": [ -0.029251099, -0.002401352, -0.0060157776, -0.06768799, -0.000538826, 0.026519775, -0.07751465, -0.021697998, 0.017349243, 0.057617188, -0.004749298, -0.0033931732, 0.053497314, -0.018325806, -0.060302734, 0.027679443, -0.012237549, ...
http://arxiv.org/abs/2310.14853v1
Adaptive Policy with Wait-$k$ Model for Simultaneous Translation
Libo Zhao, Kai Fan, Wei Luo, Jing Wu, Shushu Wang, Ziqian Zeng, Zhongqiang Huang
cs.CL
Simultaneous machine translation (SiMT) requires a robust read/write policy in conjunction with a high-quality translation model. Traditional methods rely on either a fixed wait-$k$ policy coupled with a standalone wait-$k$ translation model, or an adaptive policy jointly trained with the translation model. In this stu...
http://arxiv.org/pdf/2310.14853v1
2023-10-23T12:16:32
2023-10-23T12:16:32
{ "title": [ -0.044311523, -0.007835388, -0.009506226, -0.05029297, -0.024856567, 0.02067566, -0.048034668, -0.0021457672, -0.05557251, 0.042175293, 0.013031006, 0.023773193, -0.031341553, -0.018829346, -0.017089844, -0.0038089752, 0.012809753, 0...
http://arxiv.org/abs/2310.14849v1
Universal Domain Adaptation for Robust Handling of Distributional Shifts in NLP
Hyuhng Joon Kim, Hyunsoo Cho, Sang-Woo Lee, Junyeob Kim, Choonghyun Park, Sang-goo Lee, Kang Min Yoo, Taeuk Kim
cs.CL
When deploying machine learning systems to the wild, it is highly desirable for them to effectively leverage prior knowledge to the unfamiliar domain while also firing alarms to anomalous inputs. In order to address these requirements, Universal Domain Adaptation (UniDA) has emerged as a novel research area in computer...
http://arxiv.org/pdf/2310.14849v1
2023-10-23T12:15:25
2023-10-23T12:15:25
{ "title": [ -0.013458252, 0.013809204, -0.036712646, -0.045806885, 0.012245178, -0.0052108765, -0.03050232, -0.015701294, 0.026824951, 0.03463745, 0.0028972626, -0.0013113022, 0.01260376, -0.015060425, -0.05178833, 0.02142334, -0.026611328, 0.05...
http://arxiv.org/abs/2310.14840v1
Transparency at the Source: Evaluating and Interpreting Language Models With Access to the True Distribution
Jaap Jumelet, Willem Zuidema
cs.CL
We present a setup for training, evaluating and interpreting neural language models, that uses artificial, language-like data. The data is generated using a massive probabilistic grammar (based on state-split PCFGs), that is itself derived from a large natural language corpus, but also provides us complete control over...
http://arxiv.org/pdf/2310.14840v1
2023-10-23T12:03:01
2023-10-23T12:03:01
{ "title": [ -0.046936035, -0.014701843, -0.04071045, -0.056793213, -0.009803772, 0.004005432, -0.076171875, -0.010154724, -0.011047363, 0.033233643, -0.007522583, -0.010360718, 0.0063171387, -0.022964478, -0.05886841, 0.005718231, -0.010238647, ...
http://arxiv.org/abs/2310.14829v1
Characterizing how 'distributional' NLP corpora distance metrics are
Samuel Ackerman, George Kour, Eitan Farchi
cs.CL, stat.AP
A corpus of vector-embedded text documents has some empirical distribution. Given two corpora, we want to calculate a single metric of distance (e.g., Mauve, Frechet Inception) between them. We describe an abstract quality, called `distributionality', of such metrics. A non-distributional metric tends to use very local...
http://arxiv.org/pdf/2310.14829v1
2023-10-23T11:48:23
2023-10-23T11:48:23
{ "title": [ -0.033294678, -0.0023422241, -0.04763794, -0.020462036, -0.007217407, -0.015411377, -0.01878357, -0.013435364, 0.007881165, 0.015945435, -0.047607422, 0.010765076, -0.00068187714, -0.0036640167, -0.03778076, -0.0064849854, -0.015365601, ...
http://arxiv.org/abs/2310.14820v1
ALCUNA: Large Language Models Meet New Knowledge
Xunjian Yin, Baizhou Huang, Xiaojun Wan
cs.CL
With the rapid development of NLP, large-scale language models (LLMs) excel in various tasks across multiple domains now. However, existing benchmarks may not adequately measure these models' capabilities, especially when faced with new knowledge. In this paper, we address the lack of benchmarks to evaluate LLMs' abili...
http://arxiv.org/pdf/2310.14820v1
2023-10-23T11:40:05
2023-10-23T11:40:05
{ "title": [ -0.044158936, -0.027923584, -0.050811768, -0.07342529, -0.0029754639, 0.0132751465, -0.06842041, 0.016967773, 0.0041770935, 0.029724121, -0.0018310547, -0.007881165, 0.017349243, -0.02230835, -0.03982544, 0.020965576, -0.011352539, 0...
http://arxiv.org/abs/2310.14819v1
Analyzing Multilingual Competency of LLMs in Multi-Turn Instruction Following: A Case Study of Arabic
Sabri Boughorbel, Majd Hawasly
cs.CL
While significant progress has been made in benchmarking Large Language Models (LLMs) across various tasks, there is a lack of comprehensive evaluation of their abilities in responding to multi-turn instructions in less-commonly tested languages like Arabic. Our paper offers a detailed examination of the proficiency of...
http://arxiv.org/pdf/2310.14819v1
2023-10-23T11:40:04
2023-10-23T11:40:04
{ "title": [ -0.023605347, -0.034942627, -0.02609253, -0.037384033, -0.04095459, 0.015838623, -0.05218506, 0.022094727, -0.026321411, 0.026641846, 0.003282547, 0.030181885, -0.041259766, -0.043762207, -0.009559631, -0.021209717, -0.019699097, 0.0...
http://arxiv.org/abs/2310.15205v2
DISC-FinLLM: A Chinese Financial Large Language Model based on Multiple Experts Fine-tuning
Wei Chen, Qiushi Wang, Zefei Long, Xianyin Zhang, Zhongtian Lu, Bingxuan Li, Siyuan Wang, Jiarong Xu, Xiang Bai, Xuanjing Huang, Zhongyu Wei
cs.CL
We propose Multiple Experts Fine-tuning Framework to build a financial large language model (LLM), DISC-FinLLM. Our methodology improves general LLMs by endowing them with multi-turn question answering abilities, domain text processing capabilities, mathematical computation skills, and retrieval-enhanced generation cap...
http://arxiv.org/pdf/2310.15205v2
2023-10-25T05:56:13
2023-10-23T11:33:41
{ "title": [ -0.058502197, -0.009986877, -0.04260254, -0.03677368, -0.017837524, 0.0031318665, -0.050201416, 0.007949829, -0.0014514923, 0.04598999, -0.00081682205, -0.023117065, 0.0051574707, -0.015144348, -0.06561279, -0.013366699, -0.022369385, ...
http://arxiv.org/abs/2310.14806v1
Leveraging Timestamp Information for Serialized Joint Streaming Recognition and Translation
Sara Papi, Peidong Wang, Junkun Chen, Jian Xue, Naoyuki Kanda, Jinyu Li, Yashesh Gaur
cs.CL, cs.SD, eess.AS
The growing need for instant spoken language transcription and translation is driven by increased global communication and cross-lingual interactions. This has made offering translations in multiple languages essential for user applications. Traditional approaches to automatic speech recognition (ASR) and speech transl...
http://arxiv.org/pdf/2310.14806v1
2023-10-23T11:00:27
2023-10-23T11:00:27
{ "title": [ -0.011116028, 0.004917145, -0.036254883, -0.101501465, 0.021972656, 0.00881958, -0.048828125, 0.018814087, -0.0038394928, 0.0335083, -0.022567749, 0.00919342, -0.021697998, -0.014030457, -0.058685303, -0.0053634644, 0.016693115, 0.01...
http://arxiv.org/abs/2310.14805v1
Cross-Modal Conceptualization in Bottleneck Models
Danis Alukaev, Semen Kiselev, Ilya Pershin, Bulat Ibragimov, Vladimir Ivanov, Alexey Kornaev, Ivan Titov
cs.CL
Concept Bottleneck Models (CBMs) assume that training examples (e.g., x-ray images) are annotated with high-level concepts (e.g., types of abnormalities), and perform classification by first predicting the concepts, followed by predicting the label relying on these concepts. The main difficulty in using CBMs comes from...
http://arxiv.org/pdf/2310.14805v1
2023-10-23T11:00:19
2023-10-23T11:00:19
{ "title": [ -0.009269714, 0.022506714, -0.028778076, -0.026885986, -0.008842468, -0.020950317, -0.041229248, -0.04562378, 0.011543274, 0.036315918, -0.03201294, 0.017471313, -0.01940918, -0.011352539, -0.03555298, 0.00006866455, -0.0088272095, 0...
http://arxiv.org/abs/2310.14804v1
Large Language Models can Share Images, Too!
Young-Jun Lee, Jonghwan Hyeon, Ho-Jin Choi
cs.CV, cs.AI, cs.CL
This paper explores the image-sharing capability of Large Language Models (LLMs), such as InstructGPT, ChatGPT, and GPT-4, in a zero-shot setting, without the help of visual foundation models. Inspired by the two-stage process of image-sharing in human dialogues, we propose a two-stage framework that allows LLMs to pre...
http://arxiv.org/pdf/2310.14804v1
2023-10-23T10:59:21
2023-10-23T10:59:21
{ "title": [ 0.040924072, -0.0137786865, 0.008720398, -0.06518555, -0.024749756, 0.007095337, -0.038269043, -0.009208679, 0.011749268, 0.021453857, -0.024887085, 0.013122559, -0.0051612854, -0.03652954, 0.019470215, -0.011940002, 0.021224976, 0.0...
http://arxiv.org/abs/2310.14799v1
Cross-lingual Prompting: Improving Zero-shot Chain-of-Thought Reasoning across Languages
Libo Qin, Qiguang Chen, Fuxuan Wei, Shijue Huang, Wanxiang Che
cs.CL, cs.AI
Chain-of-thought (CoT) is capable of eliciting models to explicitly generate reasoning paths, thus promoting reasoning accuracy and attracting increasing attention. Specifically, zero-shot CoT achieves remarkable improvements in a wide range of reasoning tasks by simply instructing the LLM with the prompt "Let's think ...
http://arxiv.org/pdf/2310.14799v1
2023-10-23T10:56:03
2023-10-23T10:56:03
{ "title": [ 0.008117676, -0.037078857, -0.011680603, -0.06652832, -0.020431519, -0.0050621033, -0.05215454, 0.0017824173, 0.012664795, 0.012619019, -0.01713562, 0.013046265, 0.008041382, -0.028701782, -0.020126343, 0.012565613, 0.00085401535, 0....
http://arxiv.org/abs/2310.14793v1
What do Deck Chairs and Sun Hats Have in Common? Uncovering Shared Properties in Large Concept Vocabularies
Amit Gajbhiye, Zied Bouraoui, Na Li, Usashi Chatterjee, Luis Espinosa Anke, Steven Schockaert
cs.CL, cs.AI, cs.LG
Concepts play a central role in many applications. This includes settings where concepts have to be modelled in the absence of sentence context. Previous work has therefore focused on distilling decontextualised concept embeddings from language models. But concepts can be modelled from different perspectives, whereas c...
http://arxiv.org/pdf/2310.14793v1
2023-10-23T10:53:25
2023-10-23T10:53:25
{ "title": [ 0.016799927, 0.0022792816, -0.036895752, -0.025802612, 0.0013065338, -0.038726807, 0.002210617, -0.02619934, -0.004558563, 0.0013408661, -0.05899048, 0.0390625, -0.015548706, -0.008651733, -0.0008468628, -0.029663086, -0.008918762, 0...
http://arxiv.org/abs/2310.14777v1
Geographical Erasure in Language Generation
Pola Schwöbel, Jacek Golebiowski, Michele Donini, Cédric Archambeau, Danish Pruthi
cs.CL, cs.LG
Large language models (LLMs) encode vast amounts of world knowledge. However, since these models are trained on large swaths of internet data, they are at risk of inordinately capturing information about dominant groups. This imbalance can propagate into generated language. In this work, we study and operationalise a f...
http://arxiv.org/pdf/2310.14777v1
2023-10-23T10:26:14
2023-10-23T10:26:14
{ "title": [ -0.042999268, -0.005168915, -0.041412354, -0.023391724, -0.024841309, 0.0034542084, -0.029388428, 0.000702858, -0.011871338, 0.05657959, -0.004234314, -0.028152466, 0.025177002, -0.023406982, -0.043304443, 0.017486572, -0.019439697, ...
http://arxiv.org/abs/2310.14771v1
Evaluating the Knowledge Base Completion Potential of GPT
Blerta Veseli, Simon Razniewski, Jan-Christoph Kalo, Gerhard Weikum
cs.CL, cs.AI
Structured knowledge bases (KBs) are an asset for search engines and other applications, but are inevitably incomplete. Language models (LMs) have been proposed for unsupervised knowledge base completion (KBC), yet, their ability to do this at scale and with high accuracy remains an open question. Prior experimental st...
http://arxiv.org/pdf/2310.14771v1
2023-10-23T10:15:13
2023-10-23T10:15:13
{ "title": [ 0.0011320114, 0.01802063, -0.034240723, -0.06011963, 0.008651733, 0.007484436, -0.05114746, -0.034179688, -0.0037670135, 0.020843506, -0.052734375, 0.026138306, -0.030212402, -0.006816864, -0.034179688, 0.0050582886, -0.00680542, 0.0...
http://arxiv.org/abs/2310.14757v1
SuperTweetEval: A Challenging, Unified and Heterogeneous Benchmark for Social Media NLP Research
Dimosthenis Antypas, Asahi Ushio, Francesco Barbieri, Leonardo Neves, Kiamehr Rezaee, Luis Espinosa-Anke, Jiaxin Pei, Jose Camacho-Collados
cs.CL
Despite its relevance, the maturity of NLP for social media pales in comparison with general-purpose models, metrics and benchmarks. This fragmented landscape makes it hard for the community to know, for instance, given a task, which is the best performing model and how it compares with others. To alleviate this issue,...
http://arxiv.org/pdf/2310.14757v1
2023-10-23T09:48:25
2023-10-23T09:48:25
{ "title": [ 0.006587982, 0.0024776459, -0.028381348, -0.057861328, 0.014015198, 0.0046577454, -0.027542114, -0.019226074, 0.00090026855, 0.032592773, -0.032714844, 0.046875, -0.011306763, 0.00869751, -0.033569336, 0.010856628, -0.00969696, 0.047...
http://arxiv.org/abs/2310.14747v2
MCC-KD: Multi-CoT Consistent Knowledge Distillation
Hongzhan Chen, Siyue Wu, Xiaojun Quan, Rui Wang, Ming Yan, Ji Zhang
cs.CL
Large language models (LLMs) have showcased remarkable capabilities in complex reasoning through chain of thought (CoT) prompting. Recently, there has been a growing interest in transferring these reasoning abilities from LLMs to smaller models. However, achieving both the diversity and consistency in rationales presen...
http://arxiv.org/pdf/2310.14747v2
2023-10-24T01:43:33
2023-10-23T09:32:53
{ "title": [ -0.059051514, -0.018966675, -0.080566406, -0.101989746, 0.0038471222, 0.02017212, -0.09863281, 0.021392822, 0.03062439, 0.023361206, -0.031280518, -0.014503479, 0.019058228, -0.02168274, -0.037597656, -0.0071983337, 0.0034389496, 0.0...
http://arxiv.org/abs/2310.14735v2
Unleashing the potential of prompt engineering in Large Language Models: a comprehensive review
Banghao Chen, Zhaofeng Zhang, Nicolas Langrené, Shengxin Zhu
cs.CL, cs.AI, I.2.7
This paper delves into the pivotal role of prompt engineering in unleashing the capabilities of Large Language Models (LLMs). Prompt engineering is the process of structuring input text for LLMs and is a technique integral to optimizing the efficacy of LLMs. This survey elucidates foundational principles of prompt engi...
http://arxiv.org/pdf/2310.14735v2
2023-10-27T14:22:43
2023-10-23T09:15:18
{ "title": [ -0.021255493, -0.019805908, -0.037139893, -0.06524658, -0.004825592, -0.008201599, -0.060302734, 0.0124435425, 0.00089263916, 0.023208618, 0.018096924, -0.00023460388, 0.01675415, -0.044647217, -0.02709961, 0.029083252, -0.017593384, ...
http://arxiv.org/abs/2310.14732v1
Generating Prototypes for Contradiction Detection Using Large Language Models and Linguistic Rules
Maren Pielka, Svetlana Schmidt, Rafet Sifa
cs.CL, cs.AI
We introduce a novel data generation method for contradiction detection, which leverages the generative power of large language models as well as linguistic rules. Our vision is to provide a condensed corpus of prototypical contradictions, allowing for in-depth linguistic analysis as well as efficient language model fi...
http://arxiv.org/pdf/2310.14732v1
2023-10-23T09:07:27
2023-10-23T09:07:27
{ "title": [ -0.0058670044, -0.005584717, -0.03375244, -0.054992676, -0.0022239685, 0.00944519, -0.06185913, -0.022979736, -0.0043525696, 0.007549286, -0.009742737, -0.020645142, 0.0027503967, -0.028915405, -0.03277588, -0.021453857, -0.033691406, ...
http://arxiv.org/abs/2310.14724v2
A Survey on LLM-generated Text Detection: Necessity, Methods, and Future Directions
Junchao Wu, Shu Yang, Runzhe Zhan, Yulin Yuan, Derek F. Wong, Lidia S. Chao
cs.CL, cs.AI
The powerful ability to understand, follow, and generate complex language emerging from large language models (LLMs) makes LLM-generated text flood many areas of our daily lives at an incredible speed and is widely accepted by humans. As LLMs continue to expand, there is an imperative need to develop detectors that can...
http://arxiv.org/pdf/2310.14724v2
2023-10-24T11:31:38
2023-10-23T09:01:13
{ "title": [ -0.05734253, 0.0012388229, -0.03451538, -0.00497818, -0.020767212, -0.00081825256, -0.050994873, -0.026321411, -0.022338867, 0.020599365, -0.029190063, 0.036987305, -0.01626587, -0.013298035, -0.019851685, -0.010894775, -0.029815674, ...
http://arxiv.org/abs/2310.14709v1
Once Upon a $\textit{Time}$ in $\textit{Graph}$: Relative-Time Pretraining for Complex Temporal Reasoning
Sen Yang, Xin Li, Lidong Bing, Wai Lam
cs.CL
Our physical world is constantly evolving over time, rendering challenges for pre-trained language models to understand and reason over the temporal contexts of texts. Existing work focuses on strengthening the direct association between a piece of text and its time-stamp. However, the knowledge-time association is usu...
http://arxiv.org/pdf/2310.14709v1
2023-10-23T08:49:00
2023-10-23T08:49:00
{ "title": [ 0.0309906, -0.019577026, -0.033569336, -0.061920166, -0.0011177063, -0.03353882, -0.026290894, -0.0033874512, -0.016464233, 0.013542175, -0.008895874, 0.024124146, 0.018371582, -0.03967285, 0.012229919, 0.007881165, 0.0067214966, 0.0...
http://arxiv.org/abs/2310.14708v1
Strong and Efficient Baselines for Open Domain Conversational Question Answering
Andrei C. Coman, Gianni Barlacchi, Adrià de Gispert
cs.CL
Unlike the Open Domain Question Answering (ODQA) setting, the conversational (ODConvQA) domain has received limited attention when it comes to reevaluating baselines for both efficiency and effectiveness. In this paper, we study the State-of-the-Art (SotA) Dense Passage Retrieval (DPR) retriever and Fusion-in-Decoder (...
http://arxiv.org/pdf/2310.14708v1
2023-10-23T08:48:14
2023-10-23T08:48:14
{ "title": [ 0.0028324127, 0.004524231, -0.02885437, -0.085876465, 0.021987915, 0.0065956116, -0.04525757, 0.008972168, 0.02822876, 0.03152466, -0.021148682, 0.0029315948, 0.010437012, 0.0012178421, -0.0574646, 0.009231567, -0.008880615, 0.020553...
http://arxiv.org/abs/2310.14703v1
The continued usefulness of vocabulary tests for evaluating large language models
Gonzalo Martínez, Javier Conde, Elena Merino-Gómez, Beatriz Bermúdez-Margaretto, José Alberto Hernández, Pedro Reviriego, Marc Brysbaert
cs.CL
In their seminal article on semantic vectors, Landauer and Dumain (1997) proposed testing the quality of AI language models with a challenging vocabulary test. We show that their Test of English as a Foreign Language (TOEFL) test remains informative for contemporary major language models, since none of the models was p...
http://arxiv.org/pdf/2310.14703v1
2023-10-23T08:45:12
2023-10-23T08:45:12
{ "title": [ -0.03656006, -0.014503479, -0.039123535, -0.048858643, -0.01940918, -0.016693115, -0.04849243, -0.0039711, 0.011833191, 0.023010254, -0.03250122, -0.0058135986, -0.01348114, -0.021697998, -0.05307007, 0.0026340485, -0.012260437, 0.03...
http://arxiv.org/abs/2310.14696v1
Tree of Clarifications: Answering Ambiguous Questions with Retrieval-Augmented Large Language Models
Gangwoo Kim, Sungdong Kim, Byeongguk Jeon, Joonsuk Park, Jaewoo Kang
cs.CL
Questions in open-domain question answering are often ambiguous, allowing multiple interpretations. One approach to handling them is to identify all possible interpretations of the ambiguous question (AQ) and to generate a long-form answer addressing them all, as suggested by Stelmakh et al., (2022). While it provides ...
http://arxiv.org/pdf/2310.14696v1
2023-10-23T08:42:49
2023-10-23T08:42:49
{ "title": [ -0.020965576, -0.009162903, -0.031829834, -0.06323242, -0.0079422, 0.016174316, -0.054626465, -0.0032539368, 0.021865845, 0.01499939, -0.025344849, -0.012008667, -0.00497818, -0.011222839, -0.06726074, -0.011695862, -0.011161804, 0.0...
http://arxiv.org/abs/2310.14687v1
API-Assisted Code Generation for Question Answering on Varied Table Structures
Yihan Cao, Shuyi Chen, Ryan Liu, Zhiruo Wang, Daniel Fried
cs.CL, cs.AI
A persistent challenge to table question answering (TableQA) by generating executable programs has been adapting to varied table structures, typically requiring domain-specific logical forms. In response, this paper introduces a unified TableQA framework that: (1) provides a unified representation for structured tables...
http://arxiv.org/pdf/2310.14687v1
2023-10-23T08:26:28
2023-10-23T08:26:28
{ "title": [ -0.0052223206, -0.027008057, -0.042877197, -0.07165527, -0.008865356, -0.0067977905, -0.041656494, 0.038360596, 0.015991211, 0.032562256, -0.035614014, 0.000975132, -0.0031929016, 0.0072288513, -0.072509766, -0.011871338, 0.0065612793, ...
http://arxiv.org/abs/2310.14684v1
SpEL: Structured Prediction for Entity Linking
Hassan S. Shavarani, Anoop Sarkar
cs.CL
Entity linking is a prominent thread of research focused on structured data creation by linking spans of text to an ontology or knowledge source. We revisit the use of structured prediction for entity linking which classifies each individual input token as an entity, and aggregates the token predictions. Our system, ca...
http://arxiv.org/pdf/2310.14684v1
2023-10-23T08:24:35
2023-10-23T08:24:35
{ "title": [ -0.0012769699, -0.0061454773, -0.043640137, -0.119628906, 0.026992798, 0.013244629, -0.047088623, 0.013160706, 0.003255844, 0.020767212, 0.0018615723, 0.0027122498, 0.026779175, -0.026351929, -0.013183594, 0.014022827, -0.02041626, 0...
http://arxiv.org/abs/2310.14676v1
Pre-Trained Language Models Augmented with Synthetic Scanpaths for Natural Language Understanding
Shuwen Deng, Paul Prasse, David R. Reich, Tobias Scheffer, Lena A. Jäger
cs.CL
Human gaze data offer cognitive information that reflects natural language comprehension. Indeed, augmenting language models with human scanpaths has proven beneficial for a range of NLP tasks, including language understanding. However, the applicability of this approach is hampered because the abundance of text corpor...
http://arxiv.org/pdf/2310.14676v1
2023-10-23T08:15:38
2023-10-23T08:15:38
{ "title": [ -0.016433716, -0.019454956, -0.005176544, -0.075805664, -0.025985718, 0.010757446, -0.057281494, 0.0073394775, 0.00995636, 0.011260986, -0.03805542, -0.0037765503, -0.012336731, -0.00970459, -0.043060303, -0.010414124, -0.010734558, ...
http://arxiv.org/abs/2310.14670v2
Dataset Bias Mitigation in Multiple-Choice Visual Question Answering and Beyond
Zhecan Wang, Long Chen, Haoxuan You, Keyang Xu, Yicheng He, Wenhao Li, Noel Codella, Kai-Wei Chang, Shih-Fu Chang
cs.CV, cs.AI, cs.CL, cs.LG, cs.MM
Vision-language (VL) understanding tasks evaluate models' comprehension of complex visual scenes through multiple-choice questions. However, we have identified two dataset biases that models can exploit as shortcuts to resolve various VL tasks correctly without proper understanding. The first type of dataset bias is \e...
http://arxiv.org/pdf/2310.14670v2
2023-10-31T20:49:11
2023-10-23T08:09:42
{ "title": [ 0.0029277802, 0.005748749, -0.030456543, -0.018798828, -0.0046195984, -0.022857666, 0.0007367134, -0.016983032, 0.04019165, 0.011505127, -0.04763794, 0.0053215027, -0.016098022, -0.0027885437, -0.055725098, -0.012031555, 0.0037078857, ...
http://arxiv.org/abs/2310.14663v1
DPP-TTS: Diversifying prosodic features of speech via determinantal point processes
Seongho Joo, Hyukhun Koh, Kyomin Jung
eess.AS, cs.CL
With the rapid advancement in deep generative models, recent neural Text-To-Speech(TTS) models have succeeded in synthesizing human-like speech. There have been some efforts to generate speech with various prosody beyond monotonous prosody patterns. However, previous works have several limitations. First, typical TTS m...
http://arxiv.org/pdf/2310.14663v1
2023-10-23T07:59:46
2023-10-23T07:59:46
{ "title": [ 0.0021820068, 0.019088745, -0.041534424, -0.064575195, 0.017456055, -0.00970459, -0.041229248, -0.0008749962, -0.009605408, 0.04611206, 0.0007171631, -0.023742676, 0.05105591, -0.027923584, -0.0335083, 0.0046157837, -0.0024490356, 0....
http://arxiv.org/abs/2310.14657v1
Reasoning about Ambiguous Definite Descriptions
Stefan F. Schouten, Peter Bloem, Ilia Markov, Piek Vossen
cs.CL, cs.AI
Natural language reasoning plays an increasingly important role in improving language models' ability to solve complex language understanding tasks. An interesting use case for reasoning is the resolution of context-dependent ambiguity. But no resources exist to evaluate how well Large Language Models can use explicit ...
http://arxiv.org/pdf/2310.14657v1
2023-10-23T07:52:38
2023-10-23T07:52:38
{ "title": [ -0.029083252, -0.001115799, -0.037963867, -0.072143555, -0.003616333, 0.0061454773, -0.042663574, 0.0022201538, -0.0049324036, 0.021057129, -0.020980835, -0.0044784546, -0.02545166, -0.02468872, -0.014343262, -0.009788513, -0.0154800415,...
http://arxiv.org/abs/2310.14654v2
SPRING-INX: A Multilingual Indian Language Speech Corpus by SPRING Lab, IIT Madras
Nithya R, Malavika S, Jordan F, Arjun Gangwar, Metilda N J, S Umesh, Rithik Sarab, Akhilesh Kumar Dubey, Govind Divakaran, Samudra Vijaya K, Suryakanth V Gangashetty
cs.CL, eess.AS
India is home to a multitude of languages of which 22 languages are recognised by the Indian Constitution as official. Building speech based applications for the Indian population is a difficult problem owing to limited data and the number of languages and accents to accommodate. To encourage the language technology co...
http://arxiv.org/pdf/2310.14654v2
2023-10-24T06:03:14
2023-10-23T07:50:10
{ "title": [ -0.04058838, -0.012626648, -0.12512207, -0.0070724487, -0.023223877, -0.04232788, -0.03652954, 0.040374756, 0.024429321, 0.036834717, -0.042053223, -0.0010643005, 0.037078857, -0.019210815, 0.007335663, -0.01586914, -0.016815186, 0.0...
http://arxiv.org/abs/2310.14644v1
Multilingual k-Nearest-Neighbor Machine Translation
David Stap, Christof Monz
cs.CL
k-nearest-neighbor machine translation has demonstrated remarkable improvements in machine translation quality by creating a datastore of cached examples. However, these improvements have been limited to high-resource language pairs, with large datastores, and remain a challenge for low-resource languages. In this pape...
http://arxiv.org/pdf/2310.14644v1
2023-10-23T07:35:37
2023-10-23T07:35:37
{ "title": [ -0.045043945, -0.022979736, 0.0064430237, -0.041046143, -0.01777649, 0.02999878, -0.05645752, -0.017303467, 0.0067863464, 0.0546875, -0.007820129, 0.0053367615, 0.02468872, -0.013381958, -0.039886475, 0.023147583, 0.0044937134, 0.029...
http://arxiv.org/abs/2310.14633v2
Extending Input Contexts of Language Models through Training on Segmented Sequences
Petros Karypis, Julian McAuley, George Karypis
cs.CL, cs.LG
Effectively training language models on long inputs poses many technical challenges. As a cost consideration, languages models are pretrained on a fixed sequence length before being adapted to longer sequences. We explore various methods for adapting models to longer inputs by training on segmented sequences and an int...
http://arxiv.org/pdf/2310.14633v2
2023-10-28T04:14:38
2023-10-23T07:13:31
{ "title": [ 0.010375977, -0.009757996, -0.024169922, -0.07385254, -0.00982666, -0.0047721863, -0.048828125, 0.009803772, -0.007987976, 0.03918457, -0.019012451, 0.018341064, -0.00082826614, -0.012901306, -0.03363037, 0.007106781, -0.018234253, 0...
http://arxiv.org/abs/2310.14628v1
Plan, Verify and Switch: Integrated Reasoning with Diverse X-of-Thoughts
Tengxiao Liu, Qipeng Guo, Yuqing Yang, Xiangkun Hu, Yue Zhang, Xipeng Qiu, Zheng Zhang
cs.CL
As large language models (LLMs) have shown effectiveness with different prompting methods, such as Chain of Thought, Program of Thought, we find that these methods have formed a great complementarity to each other on math reasoning tasks. In this work, we propose XoT, an integrated problem solving framework by promptin...
http://arxiv.org/pdf/2310.14628v1
2023-10-23T07:02:20
2023-10-23T07:02:20
{ "title": [ -0.008079529, -0.018371582, -0.040405273, -0.06695557, 0.020126343, 0.00023257732, -0.014160156, -0.0061569214, -0.007282257, 0.0024032593, -0.033355713, 0.006385803, -0.021438599, -0.017318726, -0.01676941, -0.010620117, -0.0034389496, ...
http://arxiv.org/abs/2310.14627v1
CrisisMatch: Semi-Supervised Few-Shot Learning for Fine-Grained Disaster Tweet Classification
Henry Peng Zou, Yue Zhou, Cornelia Caragea, Doina Caragea
cs.CL, cs.LG
The shared real-time information about natural disasters on social media platforms like Twitter and Facebook plays a critical role in informing volunteers, emergency managers, and response organizations. However, supervised learning models for monitoring disaster events require large amounts of annotated data, making t...
http://arxiv.org/pdf/2310.14627v1
2023-10-23T07:01:09
2023-10-23T07:01:09
{ "title": [ 0.0062713623, -0.0055656433, -0.030548096, -0.066101074, 0.0024700165, -0.0013837814, -0.02532959, 0.018341064, 0.0013875961, 0.032196045, -0.05795288, 0.034210205, -0.010238647, 0.0009613037, -0.044647217, -0.030960083, 0.0011501312, ...
http://arxiv.org/abs/2310.14626v1
Conversational Recommender System and Large Language Model Are Made for Each Other in E-commerce Pre-sales Dialogue
Yuanxing Liu, Wei-Nan Zhang, Yifan Chen, Yuchi Zhang, Haopeng Bai, Fan Feng, Hengbin Cui, Yongbin Li, Wanxiang Che
cs.CL, cs.IR
E-commerce pre-sales dialogue aims to understand and elicit user needs and preferences for the items they are seeking so as to provide appropriate recommendations. Conversational recommender systems (CRSs) learn user representation and provide accurate recommendations based on dialogue context, but rely on external kno...
http://arxiv.org/pdf/2310.14626v1
2023-10-23T07:00:51
2023-10-23T07:00:51
{ "title": [ 0.004512787, -0.023376465, -0.025848389, -0.06976318, -0.038848877, 0.0032958984, -0.068359375, 0.024520874, -0.024169922, 0.013748169, -0.015777588, -0.016860962, 0.007335663, -0.043395996, 0.02885437, -0.03050232, 0.009216309, 0.05...
http://arxiv.org/abs/2310.14623v1
CoF-CoT: Enhancing Large Language Models with Coarse-to-Fine Chain-of-Thought Prompting for Multi-domain NLU Tasks
Hoang H. Nguyen, Ye Liu, Chenwei Zhang, Tao Zhang, Philip S. Yu
cs.CL, cs.LG
While Chain-of-Thought prompting is popular in reasoning tasks, its application to Large Language Models (LLMs) in Natural Language Understanding (NLU) is under-explored. Motivated by multi-step reasoning of LLMs, we propose Coarse-to-Fine Chain-of-Thought (CoF-CoT) approach that breaks down NLU tasks into multiple rea...
http://arxiv.org/pdf/2310.14623v1
2023-10-23T06:54:51
2023-10-23T06:54:51
{ "title": [ 0.0011529922, -0.023513794, -0.0146102905, -0.06304932, -0.025650024, -0.00079631805, -0.041534424, 0.017745972, -0.011421204, 0.01612854, -0.031341553, 0.01991272, 0.024642944, -0.018295288, -0.00680542, -0.004081726, -0.016540527, ...
http://arxiv.org/abs/2310.14614v1
Efficient Cross-Task Prompt Tuning for Few-Shot Conversational Emotion Recognition
Yige Xu, Zhiwei Zeng, Zhiqi Shen
cs.CL
Emotion Recognition in Conversation (ERC) has been widely studied due to its importance in developing emotion-aware empathetic machines. The rise of pre-trained language models (PLMs) has further pushed the limit of ERC performance. However, most recent works on ERC using PLMs are heavily data-driven, and requires fine...
http://arxiv.org/pdf/2310.14614v1
2023-10-23T06:46:03
2023-10-23T06:46:03
{ "title": [ -0.0014762878, -0.016143799, -0.020095825, -0.043426514, -0.025558472, -0.006008148, -0.04812622, 0.0016784668, 0.03604126, 0.015823364, -0.032806396, -0.004295349, 0.019454956, -0.026733398, -0.04547119, -0.0030555725, 0.000992775, ...
http://arxiv.org/abs/2310.14610v1
That was the last straw, we need more: Are Translation Systems Sensitive to Disambiguating Context?
Jaechan Lee, Alisa Liu, Orevaoghene Ahia, Hila Gonen, Noah A. Smith
cs.CL
The translation of ambiguous text presents a challenge for translation systems, as it requires using the surrounding context to disambiguate the intended meaning as much as possible. While prior work has studied ambiguities that result from different grammatical features of the source and target language, we study sema...
http://arxiv.org/pdf/2310.14610v1
2023-10-23T06:38:49
2023-10-23T06:38:49
{ "title": [ 0.00069618225, 0.02279663, -0.015975952, -0.046203613, -0.0020866394, 0.0025253296, -0.03994751, -0.030410767, -0.024795532, 0.0713501, -0.039245605, -0.012916565, 0.043395996, -0.015655518, -0.049468994, -0.04058838, 0.013084412, 0....
http://arxiv.org/abs/2310.14609v1
Long Short-Term Planning for Conversational Recommendation Systems
Xian Li, Hongguang Shi, Yunfei Wang, Yeqin Zhang, Xubin Li, Cam-Tu Nguyen
cs.CL
In Conversational Recommendation Systems (CRS), the central question is how the conversational agent can naturally ask for user preferences and provide suitable recommendations. Existing works mainly follow the hierarchical architecture, where a higher policy decides whether to invoke the conversation module (to ask qu...
http://arxiv.org/pdf/2310.14609v1
2023-10-23T06:34:39
2023-10-23T06:34:39
{ "title": [ 0.0135650635, -0.004558563, -0.0135650635, -0.06317139, 0.0021266937, 0.0002039671, -0.021591187, 0.0019330978, -0.010070801, 0.027175903, -0.026016235, 0.020324707, -0.024276733, -0.015327454, -0.02406311, -0.017501831, -0.0033073425, ...
http://arxiv.org/abs/2310.14607v1
Investigating the Fairness of Large Language Models for Predictions on Tabular Data
Yanchen Liu, Srishti Gautam, Jiaqi Ma, Himabindu Lakkaraju
cs.CL, cs.LG
Recent literature has suggested the potential of using large language models (LLMs) to make predictions for tabular tasks. However, LLMs have been shown to exhibit harmful social biases that reflect the stereotypes and inequalities present in the society. To this end, as well as the widespread use of tabular data in ma...
http://arxiv.org/pdf/2310.14607v1
2023-10-23T06:31:28
2023-10-23T06:31:28
{ "title": [ -0.030914307, -0.010276794, -0.06323242, -0.066467285, -0.011245728, -0.005382538, -0.07537842, 0.009147644, 0.012214661, 0.03466797, -0.035125732, -0.021194458, 0.032287598, -0.015823364, -0.049957275, 0.0007777214, -0.020874023, 0....
http://arxiv.org/abs/2310.14605v1
M2DF: Multi-grained Multi-curriculum Denoising Framework for Multimodal Aspect-based Sentiment Analysis
Fei Zhao, Chunhui Li, Zhen Wu, Yawen Ouyang, Jianbing Zhang, Xinyu Dai
cs.CL, cs.MM
Multimodal Aspect-based Sentiment Analysis (MABSA) is a fine-grained Sentiment Analysis task, which has attracted growing research interests recently. Existing work mainly utilizes image information to improve the performance of MABSA task. However, most of the studies overestimate the importance of images since there ...
http://arxiv.org/pdf/2310.14605v1
2023-10-23T06:22:39
2023-10-23T06:22:39
{ "title": [ -0.008560181, 0.0039100647, -0.04522705, -0.058532715, 0.014160156, -0.0028533936, -0.021652222, 0.005050659, 0.04324341, 0.04751587, 0.0018587112, -0.005050659, 0.02456665, -0.016464233, -0.06536865, 0.020584106, -0.023117065, 0.021...
http://arxiv.org/abs/2310.14602v2
Generative Pre-trained Transformer for Vietnamese Community-based COVID-19 Question Answering
Tam Minh Vo, Khiem Vinh Tran
cs.CL
Recent studies have provided empirical evidence of the wide-ranging potential of Generative Pre-trained Transformer (GPT), a pretrained language model, in the field of natural language processing. GPT has been effectively employed as a decoder within state-of-the-art (SOTA) question answering systems, yielding exceptio...
http://arxiv.org/pdf/2310.14602v2
2023-10-31T09:15:51
2023-10-23T06:14:07
{ "title": [ 0.049713135, -0.017501831, -0.031402588, -0.09741211, -0.0027980804, -0.0074691772, -0.030685425, 0.021316528, -0.011489868, 0.042755127, -0.040649414, 0.011161804, 0.035308838, -0.023147583, -0.018447876, -0.018478394, 0.013450623, ...
http://arxiv.org/abs/2310.14599v1
Prefix-Tuning Based Unsupervised Text Style Transfer
Huiyu Mai, Wenhao Jiang, Zhihong Deng
cs.CL, cs.AI
Unsupervised text style transfer aims at training a generative model that can alter the style of the input sentence while preserving its content without using any parallel data. In this paper, we employ powerful pre-trained large language models and present a new prefix-tuning-based method for unsupervised text style t...
http://arxiv.org/pdf/2310.14599v1
2023-10-23T06:13:08
2023-10-23T06:13:08
{ "title": [ -0.04888916, -0.003921509, -0.034179688, -0.02406311, -0.0135269165, 0.011291504, -0.03741455, -0.008728027, -0.008148193, 0.050079346, 0.008834839, 0.033416748, 0.021820068, -0.01386261, -0.015060425, 0.02494812, -0.0037631989, 0.02...
http://arxiv.org/abs/2310.14596v1
Learning to Correct Noisy Labels for Fine-Grained Entity Typing via Co-Prediction Prompt Tuning
Minghao Tang, Yongquan He, Yongxiu Xu, Hongbo Xu, Wenyuan Zhang, Yang Lin
cs.CL, cs.AI
Fine-grained entity typing (FET) is an essential task in natural language processing that aims to assign semantic types to entities in text. However, FET poses a major challenge known as the noise labeling problem, whereby current methods rely on estimating noise distribution to identify noisy labels but are confused b...
http://arxiv.org/pdf/2310.14596v1
2023-10-23T06:04:07
2023-10-23T06:04:07
{ "title": [ -0.021377563, 0.0058250427, -0.019256592, -0.05545044, 0.0154953, 0.010307312, -0.04522705, -0.008163452, -0.0001333952, 0.03213501, -0.03189087, 0.020690918, 0.032928467, -0.009567261, -0.02078247, -0.007446289, -0.022979736, 0.0391...