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/2311.06647v1
Robust Text Classification: Analyzing Prototype-Based Networks
Zhivar Sourati, Darshan Deshpande, Filip Ilievski, Kiril Gashteovski, Sascha Saralajew
cs.CL
Downstream applications often require text classification models to be accurate, robust, and interpretable. While the accuracy of the stateof-the-art language models approximates human performance, they are not designed to be interpretable and often exhibit a drop in performance on noisy data. The family of PrototypeBa...
http://arxiv.org/pdf/2311.06647v1
2023-11-11T19:34:06
2023-11-11T19:34:06
{ "title": [ -0.017669678, 0.0074424744, -0.037628174, -0.029937744, 0.015197754, -0.0068473816, -0.02178955, -0.0256958, -0.021362305, 0.013252258, -0.016998291, 0.040039062, -0.03933716, -0.01285553, -0.037719727, -0.020355225, -0.026153564, 0....
http://arxiv.org/abs/2311.06622v1
TrainerAgent: Customizable and Efficient Model Training through LLM-Powered Multi-Agent System
Haoyuan Li, Hao Jiang, Tianke Zhang, Zhelun Yu, Aoxiong Yin, Hao Cheng, Siming Fu, Yuhao Zhang, Wanggui He
cs.AI, cs.CL
Training AI models has always been challenging, especially when there is a need for custom models to provide personalized services. Algorithm engineers often face a lengthy process to iteratively develop models tailored to specific business requirements, making it even more difficult for non-experts. The quest for high...
http://arxiv.org/pdf/2311.06622v1
2023-11-11T17:39:24
2023-11-11T17:39:24
{ "title": [ -0.045135498, -0.0067596436, -0.047454834, -0.046661377, -0.012939453, -0.00020885468, -0.056427002, -0.006313324, -0.018630981, 0.019180298, -0.024749756, 0.027038574, -0.019515991, -0.020874023, -0.008201599, -0.00466156, -0.026657104,...
http://arxiv.org/abs/2311.06612v1
PerceptionGPT: Effectively Fusing Visual Perception into LLM
Renjie Pi, Lewei Yao, Jiahui Gao, Jipeng Zhang, Tong Zhang
cs.CV, cs.CL
The integration of visual inputs with large language models (LLMs) has led to remarkable advancements in multi-modal capabilities, giving rise to visual large language models (VLLMs). However, effectively harnessing VLLMs for intricate visual perception tasks remains a challenge. In this paper, we present a novel end-t...
http://arxiv.org/pdf/2311.06612v1
2023-11-11T16:59:20
2023-11-11T16:59:20
{ "title": [ -0.020996094, -0.013633728, -0.034057617, -0.011497498, -0.051605225, -0.0149002075, -0.04486084, -0.03314209, 0.02432251, 0.0067100525, -0.060272217, 0.0018043518, 0.018936157, -0.034301758, -0.004764557, -0.019699097, -0.008110046, ...
http://arxiv.org/abs/2311.06607v1
Monkey: Image Resolution and Text Label Are Important Things for Large Multi-modal Models
Zhang Li, Biao Yang, Qiang Liu, Zhiyin Ma, Shuo Zhang, Jingxu Yang, Yabo Sun, Yuliang Liu, Xiang Bai
cs.CV, cs.AI, cs.CL
Large Multimodal Models have demonstrated impressive capabilities in understanding general vision-language tasks. However, due to the limitation of supported input resolution (e.g., 448 x 448) as well as the inexhaustive description of the training image-text pair, these models often encounter challenges when dealing w...
http://arxiv.org/pdf/2311.06607v1
2023-11-11T16:37:41
2023-11-11T16:37:41
{ "title": [ -0.0077209473, 0.003320694, -0.052612305, -0.03378296, -0.025100708, -0.04095459, -0.05328369, -0.008468628, -0.010932922, 0.036956787, -0.020050049, 0.0026493073, 0.038024902, -0.04421997, 0.013572693, -0.01411438, -0.00063991547, 0...
http://arxiv.org/abs/2311.06602v1
BizBench: A Quantitative Reasoning Benchmark for Business and Finance
Rik Koncel-Kedziorski, Michael Krumdick, Viet Lai, Varshini Reddy, Charles Lovering, Chris Tanner
cs.CL
As large language models (LLMs) impact a growing number of complex domains, it is becoming increasingly important to have fair, accurate, and rigorous evaluation benchmarks. Evaluating the reasoning skills required for business and financial NLP stands out as a particularly difficult challenge. We introduce BizBench, a...
http://arxiv.org/pdf/2311.06602v1
2023-11-11T16:16:11
2023-11-11T16:16:11
{ "title": [ -0.024719238, -0.014053345, -0.043426514, -0.03729248, 0.013389587, -0.01789856, -0.018737793, -0.015838623, -0.011116028, 0.030761719, -0.037506104, 0.004310608, -0.0033016205, -0.013656616, -0.03778076, -0.029632568, -0.008903503, ...
http://arxiv.org/abs/2311.06595v2
From Classification to Generation: Insights into Crosslingual Retrieval Augmented ICL
Xiaoqian Li, Ercong Nie, Sheng Liang
cs.CL
The remarkable ability of Large Language Models (LLMs) to understand and follow instructions has sometimes been limited by their in-context learning (ICL) performance in low-resource languages. To address this, we introduce a novel approach that leverages cross-lingual retrieval-augmented in-context learning (CREA-ICL)...
http://arxiv.org/pdf/2311.06595v2
2023-11-14T15:14:06
2023-11-11T15:40:21
{ "title": [ -0.023651123, -0.009391785, -0.03567505, -0.047790527, -0.0025959015, 0.007896423, -0.03793335, 0.0010519028, 0.019958496, 0.036834717, -0.012786865, 0.0061416626, 0.012130737, -0.02166748, -0.028915405, 0.011810303, -0.030929565, 0....
http://arxiv.org/abs/2311.06555v1
Heuristics-Driven Link-of-Analogy Prompting: Enhancing Large Language Models for Document-Level Event Argument Extraction
Hanzhang Zhou, Junlang Qian, Zijian Feng, Hui Lu, Zixiao Zhu, Kezhi Mao
cs.CL, cs.AI
In this study, we investigate in-context learning (ICL) in document-level event argument extraction (EAE). The paper identifies key challenges in this problem, including example selection, context length limitation, abundance of event types, and the limitation of Chain-of-Thought (CoT) prompting in non-reasoning tasks....
http://arxiv.org/pdf/2311.06555v1
2023-11-11T12:05:01
2023-11-11T12:05:01
{ "title": [ 0.026885986, -0.011833191, -0.006717682, -0.10253906, -0.006828308, 0.012649536, -0.061523438, -0.011314392, 0.015823364, 0.02708435, 0.029464722, -0.00233078, 0.029205322, -0.035095215, -0.022079468, 0.034240723, -0.00472641, 0.0151...
http://arxiv.org/abs/2311.06549v1
Zero-Shot Cross-Lingual Sentiment Classification under Distribution Shift: an Exploratory Study
Maarten De Raedt, Semere Kiros Bitew, Fréderic Godin, Thomas Demeester, Chris Develder
cs.CL
The brittleness of finetuned language model performance on out-of-distribution (OOD) test samples in unseen domains has been well-studied for English, yet is unexplored for multi-lingual models. Therefore, we study generalization to OOD test data specifically in zero-shot cross-lingual transfer settings, analyzing perf...
http://arxiv.org/pdf/2311.06549v1
2023-11-11T11:56:56
2023-11-11T11:56:56
{ "title": [ 0.0051651, -0.010017395, -0.021957397, -0.02861023, -0.0018548965, -0.013908386, -0.018310547, -0.014564514, 0.036346436, 0.048034668, -0.012573242, -0.0071640015, 0.0057754517, -0.0053520203, -0.07556152, 0.005432129, -0.009063721, ...
http://arxiv.org/abs/2311.06534v1
Enhancing Public Understanding of Court Opinions with Automated Summarizers
Elliott Ash, Aniket Kesari, Suresh Naidu, Lena Song, Dominik Stammbach
cs.CL
Written judicial opinions are an important tool for building public trust in court decisions, yet they can be difficult for non-experts to understand. We present a pipeline for using an AI assistant to generate simplified summaries of judicial opinions. These are more accessible to the public and more easily understood...
http://arxiv.org/pdf/2311.06534v1
2023-11-11T11:05:27
2023-11-11T11:05:27
{ "title": [ 0.018951416, -0.0033168793, -0.015914917, -0.06762695, -0.01121521, -0.008644104, -0.029769897, -0.007411957, 0.000330925, 0.0496521, -0.059417725, 0.019119263, 0.0061683655, 0.0126953125, -0.06286621, -0.001033783, 0.0031147003, 0.0...
http://arxiv.org/abs/2311.06532v1
Added Toxicity Mitigation at Inference Time for Multimodal and Massively Multilingual Translation
Marta R. Costa-jussà, David Dale, Maha Elbayad, Bokai Yu
cs.CL, I.2.7
Added toxicity in the context of translation refers to the fact of producing a translation output with more toxicity than there exists in the input. In this paper, we present MinTox which is a novel pipeline to identify added toxicity and mitigate this issue which works at inference time. MinTox uses a toxicity detecti...
http://arxiv.org/pdf/2311.06532v1
2023-11-11T11:02:02
2023-11-11T11:02:02
{ "title": [ -0.024368286, -0.005382538, -0.025360107, -0.09362793, -0.0109939575, 0.010421753, -0.07714844, 0.014274597, -0.02545166, 0.05041504, -0.0013036728, 0.026626587, 0.030227661, -0.027435303, -0.0007944107, 0.019210815, 0.0019245148, 0....
http://arxiv.org/abs/2311.06530v1
How ChatGPT is Solving Vulnerability Management Problem
Peiyu Liu, Junming Liu, Lirong Fu, Kangjie Lu, Yifan Xia, Xuhong Zhang, Wenzhi Chen, Haiqin Weng, Shouling Ji, Wenhai Wang
cs.SE, cs.AI, cs.CL, cs.CR
Recently, ChatGPT has attracted great attention from the code analysis domain. Prior works show that ChatGPT has the capabilities of processing foundational code analysis tasks, such as abstract syntax tree generation, which indicates the potential of using ChatGPT to comprehend code syntax and static behaviors. Howeve...
http://arxiv.org/pdf/2311.06530v1
2023-11-11T11:01:13
2023-11-11T11:01:13
{ "title": [ 0.021270752, -0.0057296753, -0.01701355, -0.019683838, -0.03253174, -0.005695343, -0.009559631, -0.00091314316, -0.052825928, 0.048095703, 0.01828003, 0.0057868958, -0.013046265, -0.031433105, -0.0016860962, -0.019363403, -0.011512756, ...
http://arxiv.org/abs/2311.06518v1
Minimum Description Length Hopfield Networks
Matan Abudy, Nur Lan, Emmanuel Chemla, Roni Katzir
cs.LG, cs.CL
Associative memory architectures are designed for memorization but also offer, through their retrieval method, a form of generalization to unseen inputs: stored memories can be seen as prototypes from this point of view. Focusing on Modern Hopfield Networks (MHN), we show that a large memorization capacity undermines t...
http://arxiv.org/pdf/2311.06518v1
2023-11-11T09:23:54
2023-11-11T09:23:54
{ "title": [ -0.056671143, -0.021957397, -0.046936035, -0.005004883, -0.021865845, -0.0027656555, -0.018081665, -0.0075263977, -0.014320374, 0.029800415, -0.0074653625, 0.0071868896, -0.0104599, -0.021636963, -0.024856567, 0.0044021606, -0.0018186569...
http://arxiv.org/abs/2311.06513v2
Step by Step to Fairness: Attributing Societal Bias in Task-oriented Dialogue Systems
Hsuan Su, Rebecca Qian, Chinnadhurai Sankar, Shahin Shayandeh, Shang-Tse Chen, Hung-yi Lee, Daniel M. Bikel
cs.CL, cs.AI
Recent works have shown considerable improvements in task-oriented dialogue (TOD) systems by utilizing pretrained large language models (LLMs) in an end-to-end manner. However, the biased behavior of each component in a TOD system and the error propagation issue in the end-to-end framework can lead to seriously biased ...
http://arxiv.org/pdf/2311.06513v2
2023-11-14T07:30:37
2023-11-11T09:06:15
{ "title": [ 0.010047913, -0.0064315796, -0.04107666, -0.04937744, 0.0181427, -0.027618408, 0.006008148, 0.006954193, -0.0041389465, 0.023284912, -0.033447266, -0.00831604, 0.005016327, -0.014556885, -0.044281006, -0.027313232, -0.03277588, 0.047...
http://arxiv.org/abs/2311.06503v1
Knowledgeable Preference Alignment for LLMs in Domain-specific Question Answering
Yichi Zhang, Zhuo Chen, Yin Fang, Lei Cheng, Yanxi Lu, Fangming Li, Wen Zhang, Huajun Chen
cs.CL, cs.AI
Recently, the development of large language models (LLMs) has attracted wide attention in academia and industry. Deploying LLMs to real scenarios is one of the key directions in the current Internet industry. In this paper, we present a novel pipeline to apply LLMs for domain-specific question answering (QA) that incor...
http://arxiv.org/pdf/2311.06503v1
2023-11-11T07:56:40
2023-11-11T07:56:40
{ "title": [ -0.021911621, -0.0046577454, -0.053375244, -0.05505371, -0.011169434, -0.0023899078, -0.04055786, 0.011497498, 0.014221191, 0.024978638, -0.030776978, 0.006931305, -0.0007328987, -0.011756897, -0.04067993, -0.0027389526, -0.016845703, ...
http://arxiv.org/abs/2311.06493v1
L3 Ensembles: Lifelong Learning Approach for Ensemble of Foundational Language Models
Aidin Shiri, Kaushik Roy, Amit Sheth, Manas Gaur
cs.CL
Fine-tuning pre-trained foundational language models (FLM) for specific tasks is often impractical, especially for resource-constrained devices. This necessitates the development of a Lifelong Learning (L3) framework that continuously adapts to a stream of Natural Language Processing (NLP) tasks efficiently. We propose...
http://arxiv.org/pdf/2311.06493v1
2023-11-11T06:59:50
2023-11-11T06:59:50
{ "title": [ -0.033203125, 0.0035362244, -0.06842041, -0.07220459, 0.014221191, -0.0053634644, -0.05886841, 0.021652222, 0.023864746, 0.036712646, -0.051330566, 0.014656067, 0.009010315, -0.0061798096, -0.055755615, 0.0059928894, -0.0035152435, 0...
http://arxiv.org/abs/2311.06467v1
Adaptive Language-based Mental Health Assessment with Item-Response Theory
Vasudha Varadarajan, Sverker Sikström, Oscar N. E. Kjell, H. Andrew Schwartz
cs.CL, cs.AI
Mental health issues widely vary across individuals - the manifestations of signs and symptoms can be fairly heterogeneous. Recently, language-based depression and anxiety assessments have shown promise for capturing this heterogeneous nature by evaluating a patient's own language, but such approaches require a large s...
http://arxiv.org/pdf/2311.06467v1
2023-11-11T03:37:17
2023-11-11T03:37:17
{ "title": [ -0.024780273, -0.011741638, -0.037078857, -0.052124023, 0.013305664, -0.003414154, -0.016662598, 0.012588501, 0.0029506683, 0.031707764, 0.006340027, -0.007965088, -0.014808655, -0.04159546, -0.0368042, 0.008911133, 0.015464783, 0.00...
http://arxiv.org/abs/2311.06453v2
DocGen: Generating Detailed Parameter Docstrings in Python
Vatsal Venkatkrishna, Durga Shree Nagabushanam, Emmanuel Iko-Ojo Simon, Melina Vidoni
cs.SE, cs.CL
Documentation debt hinders the effective utilization of open-source software. Although code summarization tools have been helpful for developers, most would prefer a detailed account of each parameter in a function rather than a high-level summary. However, generating such a summary is too intricate for a single genera...
http://arxiv.org/pdf/2311.06453v2
2023-11-14T05:14:31
2023-11-11T01:14:37
{ "title": [ -0.00084495544, -0.028060913, -0.047790527, -0.068481445, -0.02633667, -0.008712769, -0.04598999, 0.01966858, -0.046325684, 0.054260254, 0.00843811, -0.0023860931, 0.033355713, -0.045776367, -0.0054969788, -0.0059928894, 0.013053894, ...
http://arxiv.org/abs/2311.06446v1
THOS: A Benchmark Dataset for Targeted Hate and Offensive Speech
Saad Almohaimeed, Saleh Almohaimeed, Ashfaq Ali Shafin, Bogdan Carbunar, Ladislau Bölöni
cs.CL, cs.AI
Detecting harmful content on social media, such as Twitter, is made difficult by the fact that the seemingly simple yes/no classification conceals a significant amount of complexity. Unfortunately, while several datasets have been collected for training classifiers in hate and offensive speech, there is a scarcity of d...
http://arxiv.org/pdf/2311.06446v1
2023-11-11T00:30:31
2023-11-11T00:30:31
{ "title": [ 0.013282776, 0.0047912598, -0.04736328, -0.06213379, 0.029647827, -0.021850586, -0.002588272, -0.0026245117, -0.002609253, 0.030899048, -0.015670776, 0.024917603, -0.0026340485, -0.00894928, -0.048339844, -0.000081539154, -0.0071868896, ...
http://arxiv.org/abs/2311.06440v1
Separating the Wheat from the Chaff with BREAD: An open-source benchmark and metrics to detect redundancy in text
Isaac Caswell, Lisa Wang, Isabel Papadimitriou
cs.CL, cs.LG
Data quality is a problem that perpetually resurfaces throughout the field of NLP, regardless of task, domain, or architecture, and remains especially severe for lower-resource languages. A typical and insidious issue, affecting both training data and model output, is data that is repetitive and dominated by linguistic...
http://arxiv.org/pdf/2311.06440v1
2023-11-11T00:11:50
2023-11-11T00:11:50
{ "title": [ -0.035247803, 0.043701172, -0.036743164, -0.047698975, 0.031036377, 0.004096985, -0.051757812, -0.04421997, -0.0051498413, 0.045013428, -0.022766113, 0.017227173, 0.03479004, -0.018341064, -0.03945923, 0.0017738342, -0.0019416809, 0....
http://arxiv.org/abs/2311.06427v1
ChatGPT Prompting Cannot Estimate Predictive Uncertainty in High-Resource Languages
Martino Pelucchi, Matias Valdenegro-Toro
cs.CL, cs.LG
ChatGPT took the world by storm for its impressive abilities. Due to its release without documentation, scientists immediately attempted to identify its limits, mainly through its performance in natural language processing (NLP) tasks. This paper aims to join the growing literature regarding ChatGPT's abilities by focu...
http://arxiv.org/pdf/2311.06427v1
2023-11-10T23:25:34
2023-11-10T23:25:34
{ "title": [ -0.026565552, -0.028335571, -0.008552551, -0.028579712, -0.061706543, 0.013572693, -0.04901123, -0.002779007, 0.0023059845, 0.004966736, -0.023803711, -0.00056648254, -0.01184082, -0.046783447, -0.00491333, 0.0005760193, 0.008598328, ...
http://arxiv.org/abs/2311.06414v1
Knowledge Graphs are not Created Equal: Exploring the Properties and Structure of Real KGs
Nedelina Teneva, Estevam Hruschka
cs.LG, cs.CL
Despite the recent popularity of knowledge graph (KG) related tasks and benchmarks such as KG embeddings, link prediction, entity alignment and evaluation of the reasoning abilities of pretrained language models as KGs, the structure and properties of real KGs are not well studied. In this paper, we perform a large sca...
http://arxiv.org/pdf/2311.06414v1
2023-11-10T22:18:09
2023-11-10T22:18:09
{ "title": [ -0.05279541, -0.022277832, -0.04840088, -0.0418396, -0.04272461, 0.0033359528, -0.054229736, -0.0028133392, -0.02029419, 0.029876709, -0.005382538, -0.005306244, -0.019088745, -0.025436401, -0.043121338, -0.0121154785, 0.019363403, 0...
http://arxiv.org/abs/2311.06411v1
Analyzing Modular Approaches for Visual Question Decomposition
Apoorv Khandelwal, Ellie Pavlick, Chen Sun
cs.CV, cs.CL
Modular neural networks without additional training have recently been shown to surpass end-to-end neural networks on challenging vision-language tasks. The latest such methods simultaneously introduce LLM-based code generation to build programs and a number of skill-specific, task-oriented modules to execute them. In ...
http://arxiv.org/pdf/2311.06411v1
2023-11-10T22:14:26
2023-11-10T22:14:26
{ "title": [ 0.013999939, 0.0031147003, -0.029922485, -0.066467285, 0.008430481, 0.0026721954, -0.026138306, 0.0018386841, 0.015052795, 0.023788452, -0.013366699, 0.033416748, -0.020462036, -0.012245178, -0.045959473, 0.004562378, 0.016494751, 0....
http://arxiv.org/abs/2311.06401v2
Autoregressive Language Models For Estimating the Entropy of Epic EHR Audit Logs
Benjamin C. Warner, Thomas Kannampallil, Seunghwan Kim
cs.CL, cs.IT, math.IT
EHR audit logs are a highly granular stream of events that capture clinician activities, and is a significant area of interest for research in characterizing clinician workflow on the electronic health record (EHR). Existing techniques to measure the complexity of workflow through EHR audit logs (audit logs) involve ti...
http://arxiv.org/pdf/2311.06401v2
2023-11-14T02:01:23
2023-11-10T21:32:34
{ "title": [ -0.052368164, 0.009391785, -0.050201416, -0.040985107, -0.0032348633, 0.021347046, -0.0713501, -0.013763428, 0.00033211708, 0.06021118, -0.023376465, 0.008377075, 0.037994385, -0.013023376, -0.06097412, 0.008605957, 0.006401062, 0.02...
http://arxiv.org/abs/2311.06383v1
Distilling Large Language Models using Skill-Occupation Graph Context for HR-Related Tasks
Pouya Pezeshkpour, Hayate Iso, Thom Lake, Nikita Bhutani, Estevam Hruschka
cs.CL, cs.LG
Numerous HR applications are centered around resumes and job descriptions. While they can benefit from advancements in NLP, particularly large language models, their real-world adoption faces challenges due to absence of comprehensive benchmarks for various HR tasks, and lack of smaller models with competitive capabili...
http://arxiv.org/pdf/2311.06383v1
2023-11-10T20:25:42
2023-11-10T20:25:42
{ "title": [ -0.0036563873, -0.0072669983, -0.04953003, -0.07397461, 0.0066223145, -0.011245728, -0.0647583, -0.0016803741, -0.01259613, 0.042388916, 0.00730896, -0.0067977905, 0.021820068, -0.021438599, -0.022872925, 0.010261536, -0.015296936, 0...
http://arxiv.org/abs/2311.06382v1
Transfer Learning for Structured Pruning under Limited Task Data
Lucio Dery, David Grangier, Awni Hannun
cs.CL, cs.LG
Large, pre-trained models are problematic to use in resource constrained applications. Fortunately, task-aware structured pruning methods offer a solution. These approaches reduce model size by dropping structural units like layers and attention heads in a manner that takes into account the end-task. However, these pru...
http://arxiv.org/pdf/2311.06382v1
2023-11-10T20:23:35
2023-11-10T20:23:35
{ "title": [ -0.022567749, -0.00012302399, -0.04852295, -0.0491333, 0.0029182434, -0.0052223206, -0.03164673, 0.006958008, 0.008331299, -0.010848999, -0.024978638, 0.019348145, -0.034210205, -0.014167786, -0.01461792, -0.018692017, 0.008460999, 0...
http://arxiv.org/abs/2311.06379v1
DeMuX: Data-efficient Multilingual Learning
Simran Khanuja, Srinivas Gowriraj, Lucio Dery, Graham Neubig
cs.CL
We consider the task of optimally fine-tuning pre-trained multilingual models, given small amounts of unlabelled target data and an annotation budget. In this paper, we introduce DEMUX, a framework that prescribes the exact data-points to label from vast amounts of unlabelled multilingual data, having unknown degrees o...
http://arxiv.org/pdf/2311.06379v1
2023-11-10T20:09:08
2023-11-10T20:09:08
{ "title": [ -0.024505615, -0.019805908, -0.045288086, -0.07489014, 0.016479492, 0.00042414665, -0.041809082, 0.006351471, 0.00969696, 0.024841309, -0.016036987, 0.003446579, -0.0019254684, -0.012390137, -0.033050537, 0.0066375732, -0.013801575, ...
http://arxiv.org/abs/2311.06377v1
Heaps' Law in GPT-Neo Large Language Model Emulated Corpora
Uyen Lai, Gurjit S. Randhawa, Paul Sheridan
cs.CL
Heaps' law is an empirical relation in text analysis that predicts vocabulary growth as a function of corpus size. While this law has been validated in diverse human-authored text corpora, its applicability to large language model generated text remains unexplored. This study addresses this gap, focusing on the emulati...
http://arxiv.org/pdf/2311.06377v1
2023-11-10T20:07:32
2023-11-10T20:07:32
{ "title": [ -0.029769897, -0.019134521, -0.02784729, -0.0074310303, -0.046569824, -0.00737381, -0.031204224, 0.01574707, 0.0075035095, 0.036712646, -0.06323242, -0.035369873, 0.041778564, -0.025650024, -0.029220581, -0.03262329, -0.009109497, 0....
http://arxiv.org/abs/2311.06364v1
Relation Extraction in underexplored biomedical domains: A diversity-optimised sampling and synthetic data generation approach
Maxime Delmas, Magdalena Wysocka, André Freitas
cs.CL
The sparsity of labelled data is an obstacle to the development of Relation Extraction models and the completion of databases in various biomedical areas. While being of high interest in drug-discovery, the natural-products literature, reporting the identification of potential bioactive compounds from organisms, is a c...
http://arxiv.org/pdf/2311.06364v1
2023-11-10T19:36:00
2023-11-10T19:36:00
{ "title": [ -0.014274597, -0.002243042, -0.06628418, -0.0871582, 0.016296387, -0.0032463074, -0.051940918, 0.0030593872, 0.021499634, 0.0065727234, -0.0032806396, -0.00970459, -0.033813477, -0.026733398, -0.05633545, -0.0010271072, -0.012992859, ...
http://arxiv.org/abs/2311.06362v1
Word Definitions from Large Language Models
Yunting Yin, Steven Skiena
cs.CL
Dictionary definitions are historically the arbitrator of what words mean, but this primacy has come under threat by recent progress in NLP, including word embeddings and generative models like ChatGPT. We present an exploratory study of the degree of alignment between word definitions from classical dictionaries and t...
http://arxiv.org/pdf/2311.06362v1
2023-11-10T19:27:20
2023-11-10T19:27:20
{ "title": [ -0.03439331, -0.008361816, -0.035003662, -0.05996704, 0.009162903, 0.005420685, -0.05581665, -0.0044021606, 0.010604858, 0.024459839, -0.027450562, -0.0055160522, 0.004131317, -0.013648987, -0.033416748, 0.01612854, -0.0043754578, 0....
http://arxiv.org/abs/2311.06345v1
Schema Graph-Guided Prompt for Multi-Domain Dialogue State Tracking
Ruolin Su, Ting-Wei Wu, Biing-Hwang Juang
cs.CL
Tracking dialogue states is an essential topic in task-oriented dialogue systems, which involve filling in the necessary information in pre-defined slots corresponding to a schema. While general pre-trained language models have been shown effective in slot-filling, their performance is limited when applied to specific ...
http://arxiv.org/pdf/2311.06345v1
2023-11-10T19:00:02
2023-11-10T19:00:02
{ "title": [ 0.0036964417, -0.024505615, -0.047180176, -0.08062744, -0.015350342, -0.0071640015, -0.053253174, 0.01940918, -0.0070495605, -0.0039043427, 0.017868042, -0.0061683655, -0.002828598, -0.05682373, 0.0068092346, -0.0027866364, -0.013183594,...
http://arxiv.org/abs/2311.06243v1
Parameter-Efficient Orthogonal Finetuning via Butterfly Factorization
Weiyang Liu, Zeju Qiu, Yao Feng, Yuliang Xiu, Yuxuan Xue, Longhui Yu, Haiwen Feng, Zhen Liu, Juyeon Heo, Songyou Peng, Yandong Wen, Michael J. Black, Adrian Weller, Bernhard Schölkopf
cs.LG, cs.AI, cs.CL, cs.CV
Large foundation models are becoming ubiquitous, but training them from scratch is prohibitively expensive. Thus, efficiently adapting these powerful models to downstream tasks is increasingly important. In this paper, we study a principled finetuning paradigm -- Orthogonal Finetuning (OFT) -- for downstream task adapt...
http://arxiv.org/pdf/2311.06243v1
2023-11-10T18:59:54
2023-11-10T18:59:54
{ "title": [ 0.0001847744, -0.0074653625, -0.01209259, -0.10675049, 0.006252289, 0.022827148, -0.04449463, 0.0005841255, -0.028533936, 0.04449463, 0.011726379, 0.028579712, -0.010803223, -0.020339966, -0.022827148, 0.008201599, 0.013282776, 0.002...
http://arxiv.org/abs/2311.06239v1
Argumentation Element Annotation Modeling using XLNet
Christopher Ormerod, Amy Burkhardt, Mackenzie Young, Sue Lottridge
cs.CL, cs.LG
This study demonstrates the effectiveness of XLNet, a transformer-based language model, for annotating argumentative elements in persuasive essays. XLNet's architecture incorporates a recurrent mechanism that allows it to model long-term dependencies in lengthy texts. Fine-tuned XLNet models were applied to three datas...
http://arxiv.org/pdf/2311.06239v1
2023-11-10T18:55:23
2023-11-10T18:55:23
{ "title": [ 0.0061798096, -0.006866455, -0.028778076, -0.05166626, -0.022460938, -0.0060424805, -0.058441162, -0.015731812, -0.012504578, 0.03338623, 0.00491333, -0.016921997, 0.027923584, -0.043914795, -0.038208008, -0.0003759861, -0.027389526, ...
http://arxiv.org/abs/2311.06237v2
Summon a Demon and Bind it: A Grounded Theory of LLM Red Teaming in the Wild
Nanna Inie, Jonathan Stray, Leon Derczynski
cs.CL, cs.CR, cs.HC
Engaging in the deliberate generation of abnormal outputs from large language models (LLMs) by attacking them is a novel human activity. This paper presents a thorough exposition of how and why people perform such attacks. Using a formal qualitative methodology, we interviewed dozens of practitioners from a broad range...
http://arxiv.org/pdf/2311.06237v2
2023-11-13T17:00:40
2023-11-10T18:52:58
{ "title": [ -0.03955078, 0.0014505386, -0.036987305, 0.012420654, -0.03756714, -0.015670776, -0.019897461, -0.024047852, -0.040618896, 0.005382538, -0.011360168, 0.022125244, -0.03387451, -0.048187256, 0.011505127, -0.03656006, -0.0035362244, 0....
http://arxiv.org/abs/2311.06233v1
Data Contamination Quiz: A Tool to Detect and Estimate Contamination in Large Language Models
Shahriar Golchin, Mihai Surdeanu
cs.CL, cs.AI, cs.LG
We propose the Data Contamination Quiz, a simple and effective approach to detect data contamination in large language models (LLMs) and estimate the amount of it. Specifically, we frame data contamination detection as a series of multiple-choice questions. We devise a quiz format wherein three perturbed versions of ea...
http://arxiv.org/pdf/2311.06233v1
2023-11-10T18:48:58
2023-11-10T18:48:58
{ "title": [ -0.0042915344, 0.0035629272, -0.053222656, -0.08642578, 0.020111084, -0.011009216, -0.061157227, -0.0056877136, -0.0087890625, 0.028381348, -0.02897644, -0.002878189, 0.012634277, -0.024597168, -0.040985107, -0.0048942566, -0.02381897, ...
http://arxiv.org/abs/2311.06221v1
A Comparison of Lexicon-Based and ML-Based Sentiment Analysis: Are There Outlier Words?
Siddhant Jaydeep Mahajani, Shashank Srivastava, Alan F. Smeaton
cs.CL
Lexicon-based approaches to sentiment analysis of text are based on each word or lexical entry having a pre-defined weight indicating its sentiment polarity. These are usually manually assigned but the accuracy of these when compared against machine leaning based approaches to computing sentiment, are not known. It may...
http://arxiv.org/pdf/2311.06221v1
2023-11-10T18:21:50
2023-11-10T18:21:50
{ "title": [ -0.04083252, 0.0024738312, -0.027770996, -0.021774292, -0.004119873, 0.0008201599, -0.019577026, -0.01889038, -0.0053253174, 0.038757324, -0.05645752, 0.022323608, -0.008544922, 0.0069084167, -0.049468994, -0.022644043, -0.028427124, ...
http://arxiv.org/abs/2311.06217v1
MultiIoT: Towards Large-scale Multisensory Learning for the Internet of Things
Shentong Mo, Paul Pu Liang, Russ Salakhutdinov, Louis-Philippe Morency
cs.LG, cs.AI, cs.CL, cs.CV, cs.MM
The Internet of Things (IoT), the network integrating billions of smart physical devices embedded with sensors, software, and communication technologies for the purpose of connecting and exchanging data with other devices and systems, is a critical and rapidly expanding component of our modern world. The IoT ecosystem ...
http://arxiv.org/pdf/2311.06217v1
2023-11-10T18:13:08
2023-11-10T18:13:08
{ "title": [ -0.01890564, -0.004535675, -0.04135132, -0.02708435, 0.020217896, -0.010940552, 0.0028743744, -0.006187439, 0.009254456, 0.038909912, -0.029449463, -0.0027599335, 0.031280518, -0.013626099, -0.040405273, 0.0022735596, 0.0020618439, 0...
http://arxiv.org/abs/2311.06204v1
BanglaBait: Semi-Supervised Adversarial Approach for Clickbait Detection on Bangla Clickbait Dataset
Md. Motahar Mahtab, Monirul Haque, Mehedi Hasan, Farig Sadeque
cs.CL, cs.AI, I.2.7
Intentionally luring readers to click on a particular content by exploiting their curiosity defines a title as clickbait. Although several studies focused on detecting clickbait titles in English articles, low resource language like Bangla has not been given adequate attention. To tackle clickbait titles in Bangla, we ...
http://arxiv.org/pdf/2311.06204v1
2023-11-10T17:38:46
2023-11-10T17:38:46
{ "title": [ 0.013267517, -0.004009247, 0.0004439354, -0.03225708, 0.013244629, -0.013839722, 0.005466461, -0.014411926, -0.0309906, 0.032196045, 0.009307861, 0.02015686, -0.021972656, -0.03451538, -0.026870728, -0.015258789, -0.029571533, 0.0543...
http://arxiv.org/abs/2311.06329v1
A Survey of AI Text-to-Image and AI Text-to-Video Generators
Aditi Singh
cs.CV, cs.AI, cs.CL, cs.LG, eess.IV
Text-to-Image and Text-to-Video AI generation models are revolutionary technologies that use deep learning and natural language processing (NLP) techniques to create images and videos from textual descriptions. This paper investigates cutting-edge approaches in the discipline of Text-to-Image and Text-to-Video AI gener...
http://arxiv.org/pdf/2311.06329v1
2023-11-10T17:33:58
2023-11-10T17:33:58
{ "title": [ -0.014533997, -0.004398346, -0.023849487, -0.022232056, -0.014953613, -0.015419006, 0.004611969, -0.012893677, -0.0043144226, 0.058929443, -0.03643799, 0.012107849, 0.022750854, -0.012763977, -0.03567505, -0.002773285, 0.0074920654, ...
http://arxiv.org/abs/2311.06189v1
Syntax-semantics interface: an algebraic model
Matilde Marcolli, Robert C. Berwick, Noam Chomsky
cs.CL, math.LO, math.QA, math.RA, 91F20, 16T05, 18C50
We extend our formulation of Merge and Minimalism in terms of Hopf algebras to an algebraic model of a syntactic-semantic interface. We show that methods adopted in the formulation of renormalization (extraction of meaningful physical values) in theoretical physics are relevant to describe the extraction of meaning fro...
http://arxiv.org/pdf/2311.06189v1
2023-11-10T17:12:09
2023-11-10T17:12:09
{ "title": [ -0.029327393, -0.01701355, -0.034362793, -0.083984375, 0.011634827, 0.02218628, -0.051849365, 0.009796143, 0.008407593, 0.02859497, -0.007583618, -0.014907837, -0.01991272, -0.025344849, -0.06463623, 0.0015621185, -0.010276794, 0.015...
http://arxiv.org/abs/2311.06158v1
Language Models can be Logical Solvers
Jiazhan Feng, Ruochen Xu, Junheng Hao, Hiteshi Sharma, Yelong Shen, Dongyan Zhao, Weizhu Chen
cs.CL, cs.AI
Logical reasoning is a fundamental aspect of human intelligence and a key component of tasks like problem-solving and decision-making. Recent advancements have enabled Large Language Models (LLMs) to potentially exhibit reasoning capabilities, but complex logical reasoning remains a challenge. The state-of-the-art, sol...
http://arxiv.org/pdf/2311.06158v1
2023-11-10T16:23:50
2023-11-10T16:23:50
{ "title": [ -0.008277893, -0.02154541, 0.004814148, -0.05117798, -0.05621338, 0.022384644, -0.11047363, -0.0032196045, -0.007835388, 0.050750732, -0.009849548, -0.031036377, 0.013366699, -0.029922485, -0.013397217, -0.0012207031, 0.0017251968, 0...
http://arxiv.org/abs/2311.06121v1
Is it indeed bigger better? The comprehensive study of claim detection LMs applied for disinformation tackling
Martin Hyben, Sebastian Kula, Ivan Srba, Robert Moro, Jakub Simko
cs.CL
This study compares the performance of (1) fine-tuned models and (2) extremely large language models on the task of check-worthy claim detection. For the purpose of the comparison we composed a multilingual and multi-topical dataset comprising texts of various sources and styles. Building on this, we performed a benchm...
http://arxiv.org/pdf/2311.06121v1
2023-11-10T15:36:35
2023-11-10T15:36:35
{ "title": [ 0.031341553, 0.007648468, -0.03640747, -0.02998352, 0.022567749, -0.046447754, 0.015342712, -0.02180481, -0.050048828, 0.000063598156, -0.04663086, 0.0236969, 0.0057296753, -0.03277588, 0.038879395, -0.044067383, -0.01725769, 0.05508...
http://arxiv.org/abs/2311.06102v1
Making LLMs Worth Every Penny: Resource-Limited Text Classification in Banking
Lefteris Loukas, Ilias Stogiannidis, Odysseas Diamantopoulos, Prodromos Malakasiotis, Stavros Vassos
cs.CL, cs.AI
Standard Full-Data classifiers in NLP demand thousands of labeled examples, which is impractical in data-limited domains. Few-shot methods offer an alternative, utilizing contrastive learning techniques that can be effective with as little as 20 examples per class. Similarly, Large Language Models (LLMs) like GPT-4 can...
http://arxiv.org/pdf/2311.06102v1
2023-11-10T15:10:36
2023-11-10T15:10:36
{ "title": [ -0.043548584, -0.00447464, -0.052978516, -0.028030396, -0.016983032, -0.017105103, -0.039489746, -0.011039734, -0.004219055, 0.02708435, -0.075927734, 0.033172607, -0.0060920715, 0.0025615692, -0.0463562, -0.03994751, -0.029083252, 0...
http://arxiv.org/abs/2311.06062v1
Practical Membership Inference Attacks against Fine-tuned Large Language Models via Self-prompt Calibration
Wenjie Fu, Huandong Wang, Chen Gao, Guanghua Liu, Yong Li, Tao Jiang
cs.CL, cs.CR, cs.LG
Membership Inference Attacks (MIA) aim to infer whether a target data record has been utilized for model training or not. Prior attempts have quantified the privacy risks of language models (LMs) via MIAs, but there is still no consensus on whether existing MIA algorithms can cause remarkable privacy leakage on practic...
http://arxiv.org/pdf/2311.06062v1
2023-11-10T13:55:05
2023-11-10T13:55:05
{ "title": [ -0.033416748, -0.011833191, -0.050109863, -0.06311035, 0.009017944, -0.0034770966, -0.045410156, 0.012313843, -0.029312134, 0.022216797, 0.0023460388, 0.024917603, -0.014083862, -0.019592285, -0.016662598, -0.0021190643, 0.0034446716, ...
http://arxiv.org/abs/2311.07598v1
Multi-Label Topic Model for Financial Textual Data
Moritz Scherrmann
q-fin.ST, cs.CL, cs.LG
This paper presents a multi-label topic model for financial texts like ad-hoc announcements, 8-K filings, finance related news or annual reports. I train the model on a new financial multi-label database consisting of 3,044 German ad-hoc announcements that are labeled manually using 20 predefined, economically motivate...
http://arxiv.org/pdf/2311.07598v1
2023-11-10T12:56:07
2023-11-10T12:56:07
{ "title": [ -0.03439331, 0.005619049, -0.056793213, -0.041290283, 0.00048351288, -0.0064353943, -0.04144287, -0.006549835, 0.00047826767, 0.03515625, -0.027236938, 0.01966858, -0.0032577515, 0.0030403137, -0.040740967, 0.0000192523, -0.020339966, ...
http://arxiv.org/abs/2311.06025v1
ChiMed-GPT: A Chinese Medical Large Language Model with Full Training Regime and Better Alignment to Human Preferences
Yuanhe Tian, Ruyi Gan, Yan Song, Jiaxing Zhang, Yongdong Zhang
cs.CL
Recently, the increasing demand for superior medical services has highlighted the discrepancies in the medical infrastructure. With big data, especially texts, forming the foundation of medical services, there is an exigent need for effective natural language processing (NLP) solutions tailored to the healthcare domain...
http://arxiv.org/pdf/2311.06025v1
2023-11-10T12:25:32
2023-11-10T12:25:32
{ "title": [ -0.033569336, -0.010688782, -0.02986145, -0.04724121, -0.008399963, 0.002357483, -0.051330566, -0.012245178, -0.015220642, 0.031555176, 0.016906738, -0.014801025, 0.01512146, -0.053833008, -0.022979736, 0.00907135, -0.009979248, 0.02...
http://arxiv.org/abs/2311.05965v1
Large Language Models are Zero Shot Hypothesis Proposers
Biqing Qi, Kaiyan Zhang, Haoxiang Li, Kai Tian, Sihang Zeng, Zhang-Ren Chen, Bowen Zhou
cs.CL
Significant scientific discoveries have driven the progress of human civilisation. The explosion of scientific literature and data has created information barriers across disciplines that have slowed the pace of scientific discovery. Large Language Models (LLMs) hold a wealth of global and interdisciplinary knowledge t...
http://arxiv.org/pdf/2311.05965v1
2023-11-10T10:03:49
2023-11-10T10:03:49
{ "title": [ 0.012664795, -0.013374329, -0.020431519, -0.04510498, -0.06842041, -0.020370483, -0.04748535, 0.014564514, -0.022277832, 0.035186768, -0.025238037, -0.017028809, 0.012054443, -0.043823242, 0.007949829, -0.007019043, 0.006088257, 0.05...
http://arxiv.org/abs/2311.07594v1
How to Bridge the Gap between Modalities: A Comprehensive Survey on Multimodal Large Language Model
Shezheng Song, Xiaopeng Li, Shasha Li
cs.CL, cs.AI, cs.CV, cs.MM
This review paper explores Multimodal Large Language Models (MLLMs), which integrate Large Language Models (LLMs) like GPT-4 to handle multimodal data such as text and vision. MLLMs demonstrate capabilities like generating image narratives and answering image-based questions, bridging the gap towards real-world human-c...
http://arxiv.org/pdf/2311.07594v1
2023-11-10T09:51:24
2023-11-10T09:51:24
{ "title": [ -0.034698486, 0.009864807, -0.056243896, -0.070007324, 0.007648468, -0.002527237, -0.07537842, -0.0039787292, -0.0040893555, 0.03543091, 0.0026245117, -0.0007138252, 0.022781372, -0.032714844, -0.022705078, 0.011482239, -0.01928711, ...
http://arxiv.org/abs/2311.05928v1
The Shape of Learning: Anisotropy and Intrinsic Dimensions in Transformer-Based Models
Anton Razzhigaev, Matvey Mikhalchuk, Elizaveta Goncharova, Ivan Oseledets, Denis Dimitrov, Andrey Kuznetsov
cs.CL, cs.AI, cs.IT, cs.LG, math.GN, math.IT
In this study, we present an investigation into the anisotropy dynamics and intrinsic dimension of embeddings in transformer architectures, focusing on the dichotomy between encoders and decoders. Our findings reveal that the anisotropy profile in transformer decoders exhibits a distinct bell-shaped curve, with the hig...
http://arxiv.org/pdf/2311.05928v1
2023-11-10T08:25:02
2023-11-10T08:25:02
{ "title": [ -0.01789856, 0.010231018, -0.030227661, 0.0038661957, 0.008354187, -0.017929077, 0.0040283203, -0.04119873, -0.034332275, 0.0132751465, -0.021392822, 0.019744873, -0.011665344, -0.028396606, -0.0027179718, -0.009132385, 0.0008120537, ...
http://arxiv.org/abs/2311.05922v2
Chain of Thought with Explicit Evidence Reasoning for Few-shot Relation Extraction
Xilai Ma, Jing Li, Min Zhang
cs.CL
Few-shot relation extraction involves identifying the type of relationship between two specific entities within a text, using a limited number of annotated samples. A variety of solutions to this problem have emerged by applying meta-learning and neural graph techniques which typically necessitate a training process fo...
http://arxiv.org/pdf/2311.05922v2
2023-11-15T05:23:04
2023-11-10T08:12:00
{ "title": [ 0.019165039, -0.020217896, -0.040863037, -0.06713867, -0.0059318542, -0.0016527176, -0.026367188, 0.014953613, 0.031280518, -0.0009727478, -0.02558899, -0.0003080368, -0.021636963, -0.025039673, -0.026260376, -0.0082473755, 0.0027236938,...
http://arxiv.org/abs/2311.05915v2
Fake Alignment: Are LLMs Really Aligned Well?
Yixu Wang, Yan Teng, Kexin Huang, Chengqi Lyu, Songyang Zhang, Wenwei Zhang, Xingjun Ma, Yu-Gang Jiang, Yu Qiao, Yingchun Wang
cs.CL, cs.AI
The growing awareness of safety concerns in large language models (LLMs) has sparked considerable interest in the evaluation of safety within current research endeavors. This study investigates an interesting issue pertaining to the evaluation of LLMs, namely the substantial discrepancy in performance between multiple-...
http://arxiv.org/pdf/2311.05915v2
2023-11-14T07:49:56
2023-11-10T08:01:23
{ "title": [ -0.040893555, 0.008544922, -0.031555176, 0.0146484375, -0.052490234, -0.0051574707, -0.035583496, -0.034423828, -0.058929443, 0.023452759, -0.013442993, 0.007282257, 0.01361084, -0.049591064, 0.04348755, -0.037139893, -0.012268066, 0...
http://arxiv.org/abs/2311.05902v1
Citation Recommendation on Scholarly Legal Articles
Doğukan Arslan, Saadet Sena Erdoğan, Gülşen Eryiğit
cs.IR, cs.CL, H.3.3; I.2.7
Citation recommendation is the task of finding appropriate citations based on a given piece of text. The proposed datasets for this task consist mainly of several scientific fields, lacking some core ones, such as law. Furthermore, citation recommendation is used within the legal domain to identify supporting arguments...
http://arxiv.org/pdf/2311.05902v1
2023-11-10T07:11:55
2023-11-10T07:11:55
{ "title": [ -0.00033426285, 0.017120361, -0.037963867, -0.025756836, -0.0076141357, -0.026687622, -0.017623901, -0.022964478, -0.027832031, 0.046783447, -0.050476074, 0.023651123, 0.01776123, -0.005718231, -0.019104004, 0.0027923584, -0.006088257, ...
http://arxiv.org/abs/2311.07593v1
Follow-Up Differential Descriptions: Language Models Resolve Ambiguities for Image Classification
Reza Esfandiarpoor, Stephen H. Bach
cs.CL, cs.CV, cs.LG
A promising approach for improving the performance of vision-language models like CLIP for image classification is to extend the class descriptions (i.e., prompts) with related attributes, e.g., using brown sparrow instead of sparrow. However, current zero-shot methods select a subset of attributes regardless of common...
http://arxiv.org/pdf/2311.07593v1
2023-11-10T05:24:07
2023-11-10T05:24:07
{ "title": [ -0.031982422, 0.009246826, -0.025421143, -0.034118652, -0.008796692, -0.007881165, -0.04586792, -0.031799316, 0.016967773, 0.014801025, -0.03564453, 0.0013132095, -0.0079422, -0.015701294, -0.029403687, -0.012390137, 0.000787735, 0.0...
http://arxiv.org/abs/2311.05876v1
Trends in Integration of Knowledge and Large Language Models: A Survey and Taxonomy of Methods, Benchmarks, and Applications
Zhangyin Feng, Weitao Ma, Weijiang Yu, Lei Huang, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, Ting liu
cs.CL
Large language models (LLMs) exhibit superior performance on various natural language tasks, but they are susceptible to issues stemming from outdated data and domain-specific limitations. In order to address these challenges, researchers have pursued two primary strategies, knowledge editing and retrieval augmentation...
http://arxiv.org/pdf/2311.05876v1
2023-11-10T05:24:04
2023-11-10T05:24:04
{ "title": [ -0.024902344, -0.017044067, -0.064697266, -0.08947754, 0.017669678, 0.006374359, -0.06652832, 0.018493652, 0.012763977, 0.024490356, -0.017105103, 0.009552002, 0.006793976, -0.016159058, -0.034942627, 0.019760132, -0.003921509, 0.019...
http://arxiv.org/abs/2311.05845v1
Tamil-Llama: A New Tamil Language Model Based on Llama 2
Abhinand Balachandran
cs.CL, cs.AI, cs.LG
Language modeling has witnessed remarkable advancements in recent years, with Large Language Models (LLMs) like ChatGPT setting unparalleled benchmarks in human-like text generation. However, a prevailing limitation is the underrepresentation of languages like Tamil in these cutting-edge models, leading to suboptimal p...
http://arxiv.org/pdf/2311.05845v1
2023-11-10T03:02:39
2023-11-10T03:02:39
{ "title": [ -0.056030273, 0.0052833557, -0.060913086, -0.0256958, -0.00730896, -0.0075187683, -0.05407715, 0.0057907104, -0.032104492, 0.036071777, -0.014694214, 0.012687683, 0.018478394, -0.03543091, -0.0154418945, 0.0011081696, -0.01537323, 0....
http://arxiv.org/abs/2311.05821v1
Let's Reinforce Step by Step
Sarah Pan, Vladislav Lialin, Sherin Muckatira, Anna Rumshisky
cs.CL
While recent advances have boosted LM proficiency in linguistic benchmarks, LMs consistently struggle to reason correctly on complex tasks like mathematics. We turn to Reinforcement Learning from Human Feedback (RLHF) as a method with which to shape model reasoning processes. In particular, we explore two reward scheme...
http://arxiv.org/pdf/2311.05821v1
2023-11-10T01:35:51
2023-11-10T01:35:51
{ "title": [ 0.060516357, 0.01071167, -0.026412964, -0.09869385, 0.014526367, -0.000395298, -0.053253174, 0.002199173, -0.0501709, 0.037750244, 0.005252838, 0.04373169, -0.02557373, -0.01058197, 0.047088623, -0.02279663, 0.04019165, -0.0045394897...
http://arxiv.org/abs/2311.06318v1
Knowledge-Augmented Large Language Models for Personalized Contextual Query Suggestion
Jinheon Baek, Nirupama Chandrasekaran, Silviu Cucerzan, Allen herring, Sujay Kumar Jauhar
cs.IR, cs.AI, cs.CL, cs.LG
Large Language Models (LLMs) excel at tackling various natural language tasks. However, due to the significant costs involved in re-training or fine-tuning them, they remain largely static and difficult to personalize. Nevertheless, a variety of applications could benefit from generations that are tailored to users' pr...
http://arxiv.org/pdf/2311.06318v1
2023-11-10T01:18:47
2023-11-10T01:18:47
{ "title": [ 0.017440796, 0.0040512085, -0.035491943, -0.075805664, 0.017227173, -0.0009965897, -0.03881836, 0.0035686493, 0.00793457, 0.059387207, -0.021087646, 0.006122589, 0.029846191, 0.00036120415, -0.05303955, 0.010665894, -0.005077362, 0.0...
http://arxiv.org/abs/2311.05812v1
CFBenchmark: Chinese Financial Assistant Benchmark for Large Language Model
Yang Lei, Jiangtong Li, Ming Jiang, Junjie Hu, Dawei Cheng, Zhijun Ding, Changjun Jiang
cs.CL
Large language models (LLMs) have demonstrated great potential in the financial domain. Thus, it becomes important to assess the performance of LLMs in the financial tasks. In this work, we introduce CFBenchmark, to evaluate the performance of LLMs for Chinese financial assistant. The basic version of CFBenchmark is de...
http://arxiv.org/pdf/2311.05812v1
2023-11-10T01:12:03
2023-11-10T01:12:03
{ "title": [ -0.03451538, -0.00856781, -0.038513184, -0.035705566, -0.0368042, -0.010314941, -0.06890869, 0.0015363693, -0.0048217773, 0.061553955, -0.0184021, -0.022064209, 0.056762695, -0.03475952, -0.035888672, -0.006500244, -0.015602112, 0.03...
http://arxiv.org/abs/2311.05800v1
Leveraging LLMs for Synthesizing Training Data Across Many Languages in Multilingual Dense Retrieval
Nandan Thakur, Jianmo Ni, Gustavo Hernández Ábrego, John Wieting, Jimmy Lin, Daniel Cer
cs.IR, cs.AI, cs.CL
Dense retrieval models have predominantly been studied for English, where models have shown great success, due to the availability of human-labeled training pairs. However, there has been limited success for multilingual retrieval so far, as training data is uneven or scarcely available across multiple languages. Synth...
http://arxiv.org/pdf/2311.05800v1
2023-11-10T00:17:10
2023-11-10T00:17:10
{ "title": [ -0.025650024, -0.01083374, -0.04650879, -0.06274414, -0.007637024, 0.00025129318, -0.06109619, 0.0065078735, 0.0011377335, 0.04660034, -0.02319336, 0.026748657, 0.0018043518, -0.0033416748, -0.02859497, 0.0033874512, 0.0026950836, 0....
http://arxiv.org/abs/2311.07592v1
Hallucination-minimized Data-to-answer Framework for Financial Decision-makers
Sohini Roychowdhury, Andres Alvarez, Brian Moore, Marko Krema, Maria Paz Gelpi, Federico Martin Rodriguez, Angel Rodriguez, Jose Ramon Cabrejas, Pablo Martinez Serrano, Punit Agrawal, Arijit Mukherjee
cs.CL, cs.AI, cs.IR
Large Language Models (LLMs) have been applied to build several automation and personalized question-answering prototypes so far. However, scaling such prototypes to robust products with minimized hallucinations or fake responses still remains an open challenge, especially in niche data-table heavy domains such as fina...
http://arxiv.org/pdf/2311.07592v1
2023-11-09T22:53:52
2023-11-09T22:53:52
{ "title": [ 0.007270813, -0.00831604, -0.0496521, -0.0046310425, -0.023406982, -0.048583984, 0.024017334, 0.0043640137, 0.0029239655, 0.012268066, -0.05581665, 0.01979065, -0.0118637085, -0.021911621, -0.0057792664, -0.019470215, 0.027633667, 0....
http://arxiv.org/abs/2311.05769v1
Chatbots Are Not Reliable Text Annotators
Ross Deans Kristensen-McLachlan, Miceal Canavan, Márton Kardos, Mia Jacobsen, Lene Aarøe
cs.CL, cs.AI
Recent research highlights the significant potential of ChatGPT for text annotation in social science research. However, ChatGPT is a closed-source product which has major drawbacks with regards to transparency, reproducibility, cost, and data protection. Recent advances in open-source (OS) large language models (LLMs)...
http://arxiv.org/pdf/2311.05769v1
2023-11-09T22:28:14
2023-11-09T22:28:14
{ "title": [ -0.014213562, -0.011665344, 0.028961182, -0.0036354065, -0.05731201, 0.032196045, -0.013702393, -0.012634277, -0.02029419, 0.093566895, -0.019439697, -0.026733398, 0.028762817, -0.0020923615, -0.04989624, -0.011260986, -0.00667572, 0...
http://arxiv.org/abs/2311.07591v1
Identification of Books That are Suitable for Middle School Students Using Artificial Neural Networks
Alp Niksarli, Sadik Ozan Gorgu, Ege Gencer
cs.CL, cs.LG
Reading right books contributes to children's imagination and brain development, enhances their language and emotional comprehension abilities, and strengthens their relationships with others. Building upon the critical role of reading books in individual development, this paper aims to develop an algorithm that determ...
http://arxiv.org/pdf/2311.07591v1
2023-11-09T22:10:52
2023-11-09T22:10:52
{ "title": [ 0.0006766319, 0.019592285, -0.023498535, 0.021957397, -0.0046691895, -0.029785156, 0.03353882, -0.04611206, -0.041259766, 0.015060425, -0.016708374, 0.017974854, -0.030731201, -0.031433105, 0.01033783, -0.039642334, -0.0060691833, 0....
http://arxiv.org/abs/2311.05754v1
Deep Natural Language Feature Learning for Interpretable Prediction
Felipe Urrutia, Cristian Buc, Valentin Barriere
cs.CL, cs.AI
We propose a general method to break down a main complex task into a set of intermediary easier sub-tasks, which are formulated in natural language as binary questions related to the final target task. Our method allows for representing each example by a vector consisting of the answers to these questions. We call this...
http://arxiv.org/pdf/2311.05754v1
2023-11-09T21:43:27
2023-11-09T21:43:27
{ "title": [ -0.025665283, -0.004714966, -0.038726807, -0.0836792, 0.012542725, 0.0065231323, -0.06549072, -0.007762909, 0.008155823, 0.013412476, -0.008781433, -0.012771606, -0.007369995, -0.016448975, -0.045562744, -0.007598877, -0.00038266182, ...
http://arxiv.org/abs/2311.05746v1
Bridging the Digital Divide: Performance Variation across Socio-Economic Factors in Vision-Language Models
Joan Nwatu, Oana Ignat, Rada Mihalcea
cs.CY, cs.AI, cs.CL, cs.CV
Despite the impressive performance of current AI models reported across various tasks, performance reports often do not include evaluations of how these models perform on the specific groups that will be impacted by these technologies. Among the minority groups under-represented in AI, data from low-income households a...
http://arxiv.org/pdf/2311.05746v1
2023-11-09T21:10:52
2023-11-09T21:10:52
{ "title": [ -0.03665161, -0.0044136047, -0.03918457, -0.025985718, -0.009048462, -0.025238037, -0.043884277, -0.023742676, 0.011642456, 0.026535034, -0.018478394, -0.0045318604, 0.008705139, -0.024124146, -0.044952393, -0.00062036514, -0.0042419434,...
http://arxiv.org/abs/2311.05741v1
Efficiently Adapting Pretrained Language Models To New Languages
Zoltan Csaki, Pian Pawakapan, Urmish Thakker, Qiantong Xu
cs.CL, cs.AI, cs.LG
Recent large language models (LLM) exhibit sub-optimal performance on low-resource languages, as the training data of these models is usually dominated by English and other high-resource languages. Furthermore, it is challenging to train models for low-resource languages, especially from scratch, due to a lack of high ...
http://arxiv.org/pdf/2311.05741v1
2023-11-09T20:59:08
2023-11-09T20:59:08
{ "title": [ -0.01777649, -0.020355225, -0.008087158, -0.07293701, 0.0010643005, 0.014549255, -0.06945801, 0.0037078857, -0.004623413, 0.035583496, -0.009376526, -0.0007429123, 0.0037059784, -0.016143799, -0.04119873, 0.011161804, -0.00018763542, ...
http://arxiv.org/abs/2311.05720v1
Long-Horizon Dialogue Understanding for Role Identification in the Game of Avalon with Large Language Models
Simon Stepputtis, Joseph Campbell, Yaqi Xie, Zhengyang Qi, Wenxin Sharon Zhang, Ruiyi Wang, Sanketh Rangreji, Michael Lewis, Katia Sycara
cs.CL, cs.AI, cs.LG
Deception and persuasion play a critical role in long-horizon dialogues between multiple parties, especially when the interests, goals, and motivations of the participants are not aligned. Such complex tasks pose challenges for current Large Language Models (LLM) as deception and persuasion can easily mislead them, esp...
http://arxiv.org/pdf/2311.05720v1
2023-11-09T20:04:08
2023-11-09T20:04:08
{ "title": [ -0.0060577393, 0.00843811, -0.04006958, -0.050445557, -0.0028266907, -0.0076065063, -0.0519104, -0.0061035156, -0.0021591187, 0.0012655258, 0.0023994446, -0.0061912537, -0.025390625, -0.047698975, -0.024032593, -0.012924194, -0.020874023...
http://arxiv.org/abs/2311.05608v1
FigStep: Jailbreaking Large Vision-language Models via Typographic Visual Prompts
Yichen Gong, Delong Ran, Jinyuan Liu, Conglei Wang, Tianshuo Cong, Anyu Wang, Sisi Duan, Xiaoyun Wang
cs.CR, cs.AI, cs.CL
Large vision-language models (VLMs) like GPT-4V represent an unprecedented revolution in the field of artificial intelligence (AI). Compared to single-modal large language models (LLMs), VLMs possess more versatile capabilities by incorporating additional modalities (e.g., images). Meanwhile, there's a rising enthusias...
http://arxiv.org/pdf/2311.05608v1
2023-11-09T18:59:11
2023-11-09T18:59:11
{ "title": [ -0.039520264, -0.033966064, -0.047607422, -0.035461426, -0.014595032, -0.0024433136, -0.04849243, -0.006149292, 0.02166748, 0.000054180622, -0.015640259, -0.028533936, 0.01586914, -0.027450562, -0.033172607, -0.0074310303, -0.00032496452...
http://arxiv.org/abs/2311.05601v1
FAMuS: Frames Across Multiple Sources
Siddharth Vashishtha, Alexander Martin, William Gantt, Benjamin Van Durme, Aaron Steven White
cs.CL
Understanding event descriptions is a central aspect of language processing, but current approaches focus overwhelmingly on single sentences or documents. Aggregating information about an event \emph{across documents} can offer a much richer understanding. To this end, we present FAMuS, a new corpus of Wikipedia passag...
http://arxiv.org/pdf/2311.05601v1
2023-11-09T18:57:39
2023-11-09T18:57:39
{ "title": [ 0.0049324036, -0.0041885376, -0.0625, -0.05368042, -0.024993896, -0.015510559, -0.047912598, -0.009613037, -0.013923645, 0.031402588, -0.021148682, -0.009803772, 0.04345703, -0.04748535, 0.01789856, -0.0027370453, 0.00046634674, 0.04...
http://arxiv.org/abs/2311.05591v1
Accuracy of a Vision-Language Model on Challenging Medical Cases
Thomas Buckley, James A. Diao, Adam Rodman, Arjun K. Manrai
cs.CV, cs.AI, cs.CL
Background: General-purpose large language models that utilize both text and images have not been evaluated on a diverse array of challenging medical cases. Methods: Using 934 cases from the NEJM Image Challenge published between 2005 and 2023, we evaluated the accuracy of the recently released Generative Pre-trained...
http://arxiv.org/pdf/2311.05591v1
2023-11-09T18:48:02
2023-11-09T18:48:02
{ "title": [ 0.012229919, -0.005794525, -0.029205322, -0.03543091, -0.014175415, -0.03152466, -0.03277588, -0.03778076, 0.0024642944, 0.029006958, -0.023101807, -0.017059326, 0.025634766, -0.027511597, -0.033691406, 0.0011005402, 0.017105103, 0.0...
http://arxiv.org/abs/2311.05584v1
Zero-Shot Goal-Directed Dialogue via RL on Imagined Conversations
Joey Hong, Sergey Levine, Anca Dragan
cs.LG, cs.AI, cs.CL
Large language models (LLMs) have emerged as powerful and general solutions to many natural language tasks. However, many of the most important applications of language generation are interactive, where an agent has to talk to a person to reach a desired outcome. For example, a teacher might try to understand their stu...
http://arxiv.org/pdf/2311.05584v1
2023-11-09T18:45:16
2023-11-09T18:45:16
{ "title": [ 0.028137207, -0.033599854, 0.0056381226, -0.019210815, -0.05987549, -0.0058784485, 0.002292633, -0.0104599, -0.016967773, 0.024169922, 0.044189453, -0.019119263, 0.01689148, -0.07373047, 0.0070152283, 0.009735107, -0.009262085, 0.028...
http://arxiv.org/abs/2311.05553v2
Removing RLHF Protections in GPT-4 via Fine-Tuning
Qiusi Zhan, Richard Fang, Rohan Bindu, Akul Gupta, Tatsunori Hashimoto, Daniel Kang
cs.CL, cs.AI
As large language models (LLMs) have increased in their capabilities, so does their potential for dual use. To reduce harmful outputs, produces and vendors of LLMs have used reinforcement learning with human feedback (RLHF). In tandem, LLM vendors have been increasingly enabling fine-tuning of their most powerful model...
http://arxiv.org/pdf/2311.05553v2
2023-11-10T21:17:17
2023-11-09T17:54:59
{ "title": [ 0.03668213, 0.009346008, -0.008125305, -0.042266846, -0.014373779, -0.013183594, -0.00093746185, -0.0029315948, -0.06555176, 0.037384033, 0.040527344, -0.010566711, -0.005809784, -0.053466797, 0.008598328, -0.006225586, -0.0107040405, ...
http://arxiv.org/abs/2311.05552v1
The Iron(ic) Melting Pot: Reviewing Human Evaluation in Humour, Irony and Sarcasm Generation
Tyler Loakman, Aaron Maladry, Chenghua Lin
cs.CL
Human evaluation is often considered to be the gold standard method of evaluating a Natural Language Generation system. However, whilst its importance is accepted by the community at large, the quality of its execution is often brought into question. In this position paper, we argue that the generation of more esoteric...
http://arxiv.org/pdf/2311.05552v1
2023-11-09T17:50:23
2023-11-09T17:50:23
{ "title": [ -0.00390625, -0.0020618439, -0.03869629, 0.008842468, -0.005592346, -0.042785645, 0.020584106, -0.023742676, -0.019821167, 0.02859497, -0.024169922, -0.025222778, 0.017745972, -0.028244019, -0.052978516, -0.017044067, -0.04006958, 0....
http://arxiv.org/abs/2311.05550v1
Towards End-to-End Spoken Grammatical Error Correction
Stefano Bannò, Rao Ma, Mengjie Qian, Kate M. Knill, Mark J. F. Gales
cs.CL, cs.LG, eess.AS
Grammatical feedback is crucial for L2 learners, teachers, and testers. Spoken grammatical error correction (GEC) aims to supply feedback to L2 learners on their use of grammar when speaking. This process usually relies on a cascaded pipeline comprising an ASR system, disfluency removal, and GEC, with the associated co...
http://arxiv.org/pdf/2311.05550v1
2023-11-09T17:49:02
2023-11-09T17:49:02
{ "title": [ 0.02507019, 0.0013484955, -0.025863647, -0.08068848, -0.0024356842, -0.006790161, -0.05050659, 0.016220093, -0.0069999695, 0.04434204, -0.0115737915, -0.011749268, 0.026107788, -0.026321411, -0.043151855, 0.014251709, -0.03062439, 0....
http://arxiv.org/abs/2311.07590v1
Technical Report: Large Language Models can Strategically Deceive their Users when Put Under Pressure
Jérémy Scheurer, Mikita Balesni, Marius Hobbhahn
cs.CL, cs.AI, cs.LG
We demonstrate a situation in which Large Language Models, trained to be helpful, harmless, and honest, can display misaligned behavior and strategically deceive their users about this behavior without being instructed to do so. Concretely, we deploy GPT-4 as an agent in a realistic, simulated environment, where it ass...
http://arxiv.org/pdf/2311.07590v1
2023-11-09T17:12:44
2023-11-09T17:12:44
{ "title": [ 0.002292633, -0.012550354, -0.0017547607, 0.022537231, -0.04257202, -0.025009155, 0.01737976, -0.045166016, -0.038208008, 0.0138549805, -0.024795532, 0.017074585, -0.025543213, -0.025939941, 0.006729126, -0.053710938, -0.015304565, 0...
http://arxiv.org/abs/2311.05472v1
Text Representation Distillation via Information Bottleneck Principle
Yanzhao Zhang, Dingkun Long, Zehan Li, Pengjun Xie
cs.CL, cs.AI
Pre-trained language models (PLMs) have recently shown great success in text representation field. However, the high computational cost and high-dimensional representation of PLMs pose significant challenges for practical applications. To make models more accessible, an effective method is to distill large models into ...
http://arxiv.org/pdf/2311.05472v1
2023-11-09T16:04:17
2023-11-09T16:04:17
{ "title": [ -0.00843811, 0.016693115, -0.047668457, -0.054382324, 0.003129959, -0.014732361, -0.05215454, -0.019439697, 0.02911377, 0.06463623, -0.027511597, -0.0038700104, 0.029312134, -0.009124756, -0.058166504, 0.0074005127, 0.0005221367, 0.0...
http://arxiv.org/abs/2311.05451v1
All Should Be Equal in the Eyes of Language Models: Counterfactually Aware Fair Text Generation
Pragyan Banerjee, Abhinav Java, Surgan Jandial, Simra Shahid, Shaz Furniturewala, Balaji Krishnamurthy, Sumit Bhatia
cs.CL, cs.CY, cs.LG
Fairness in Language Models (LMs) remains a longstanding challenge, given the inherent biases in training data that can be perpetuated by models and affect the downstream tasks. Recent methods employ expensive retraining or attempt debiasing during inference by constraining model outputs to contrast from a reference se...
http://arxiv.org/pdf/2311.05451v1
2023-11-09T15:39:40
2023-11-09T15:39:40
{ "title": [ -0.015319824, -0.007888794, -0.035888672, -0.042297363, -0.010314941, -0.0062026978, -0.05090332, -0.008575439, 0.02029419, 0.03982544, -0.032592773, 0.005771637, 0.031036377, -0.012817383, -0.061645508, 0.009498596, -0.034057617, 0....
http://arxiv.org/abs/2311.05450v1
Cognitively Inspired Components for Social Conversational Agents
Alex Clay, Eduardo Alonso, Esther Mondragón
cs.CL, cs.AI, I.2.0
Current conversational agents (CA) have seen improvement in conversational quality in recent years due to the influence of large language models (LLMs) like GPT3. However, two key categories of problem remain. Firstly there are the unique technical problems resulting from the approach taken in creating the CA, such as ...
http://arxiv.org/pdf/2311.05450v1
2023-11-09T15:38:58
2023-11-09T15:38:58
{ "title": [ -0.004425049, -0.0030574799, -0.020996094, -0.049194336, -0.00198555, -0.0121154785, -0.033111572, -0.0019760132, 0.029205322, 0.03289795, -0.03074646, -0.012016296, 0.015182495, -0.006828308, -0.05593872, 0.0012636185, -0.010696411, ...
http://arxiv.org/abs/2311.05437v1
LLaVA-Plus: Learning to Use Tools for Creating Multimodal Agents
Shilong Liu, Hao Cheng, Haotian Liu, Hao Zhang, Feng Li, Tianhe Ren, Xueyan Zou, Jianwei Yang, Hang Su, Jun Zhu, Lei Zhang, Jianfeng Gao, Chunyuan Li
cs.CV, cs.AI, cs.CL, cs.LG, cs.MM
LLaVA-Plus is a general-purpose multimodal assistant that expands the capabilities of large multimodal models. It maintains a skill repository of pre-trained vision and vision-language models and can activate relevant tools based on users' inputs to fulfill real-world tasks. LLaVA-Plus is trained on multimodal instruct...
http://arxiv.org/pdf/2311.05437v1
2023-11-09T15:22:26
2023-11-09T15:22:26
{ "title": [ -0.032989502, 0.00039219856, -0.0423584, -0.025344849, 0.0075149536, -0.0024356842, -0.021057129, -0.016998291, -0.0030555725, 0.023513794, -0.021774292, 0.019104004, 0.0038928986, -0.010070801, -0.023269653, -0.0009422302, -0.020690918,...
http://arxiv.org/abs/2311.05419v1
Mirror: A Universal Framework for Various Information Extraction Tasks
Tong Zhu, Junfei Ren, Zijian Yu, Mengsong Wu, Guoliang Zhang, Xiaoye Qu, Wenliang Chen, Zhefeng Wang, Baoxing Huai, Min Zhang
cs.CL, cs.AI
Sharing knowledge between information extraction tasks has always been a challenge due to the diverse data formats and task variations. Meanwhile, this divergence leads to information waste and increases difficulties in building complex applications in real scenarios. Recent studies often formulate IE tasks as a triple...
http://arxiv.org/pdf/2311.05419v1
2023-11-09T14:58:46
2023-11-09T14:58:46
{ "title": [ -0.04006958, -0.01361084, -0.050567627, -0.06530762, 0.015945435, 0.010650635, -0.043029785, -0.009864807, 0.010421753, 0.023590088, -0.0126571655, 0.012535095, -0.0035648346, -0.013053894, -0.04058838, 0.00920105, -0.009277344, 0.04...
http://arxiv.org/abs/2311.05379v1
Memorisation Cartography: Mapping out the Memorisation-Generalisation Continuum in Neural Machine Translation
Verna Dankers, Ivan Titov, Dieuwke Hupkes
cs.CL
When training a neural network, it will quickly memorise some source-target mappings from your dataset but never learn some others. Yet, memorisation is not easily expressed as a binary feature that is good or bad: individual datapoints lie on a memorisation-generalisation continuum. What determines a datapoint's posit...
http://arxiv.org/pdf/2311.05379v1
2023-11-09T14:03:51
2023-11-09T14:03:51
{ "title": [ -0.01939392, -0.008171082, -0.0027923584, -0.037963867, -0.023040771, 0.010719299, -0.011894226, -0.00074481964, -0.0021095276, 0.06695557, -0.007160187, 0.010055542, 0.026351929, -0.00039315224, -0.06774902, 0.016494751, -0.0015134811, ...
http://arxiv.org/abs/2311.05374v1
TencentLLMEval: A Hierarchical Evaluation of Real-World Capabilities for Human-Aligned LLMs
Shuyi Xie, Wenlin Yao, Yong Dai, Shaobo Wang, Donlin Zhou, Lifeng Jin, Xinhua Feng, Pengzhi Wei, Yujie Lin, Zhichao Hu, Dong Yu, Zhengyou Zhang, Jing Nie, Yuhong Liu
cs.CL, cs.AI
Large language models (LLMs) have shown impressive capabilities across various natural language tasks. However, evaluating their alignment with human preferences remains a challenge. To this end, we propose a comprehensive human evaluation framework to assess LLMs' proficiency in following instructions on diverse real-...
http://arxiv.org/pdf/2311.05374v1
2023-11-09T13:58:59
2023-11-09T13:58:59
{ "title": [ -0.04055786, -0.015625, -0.053619385, -0.03189087, -0.026824951, -0.017745972, -0.053619385, -0.0035037994, -0.007247925, 0.03062439, -0.024673462, -0.002544403, -0.0070533752, -0.028503418, -0.04815674, -0.013572693, -0.015159607, 0...
http://arxiv.org/abs/2311.05350v1
There's no Data Like Better Data: Using QE Metrics for MT Data Filtering
Jan-Thorsten Peter, David Vilar, Daniel Deutsch, Mara Finkelstein, Juraj Juraska, Markus Freitag
cs.CL
Quality Estimation (QE), the evaluation of machine translation output without the need of explicit references, has seen big improvements in the last years with the use of neural metrics. In this paper we analyze the viability of using QE metrics for filtering out bad quality sentence pairs in the training data of neura...
http://arxiv.org/pdf/2311.05350v1
2023-11-09T13:21:34
2023-11-09T13:21:34
{ "title": [ 0.03829956, 0.014823914, -0.008583069, -0.025421143, 0.011245728, -0.042114258, 0.03353882, -0.029891968, -0.01965332, 0.02633667, -0.043945312, -0.01411438, -0.012756348, -0.041168213, -0.012588501, -0.041137695, 0.008842468, 0.0081...
http://arxiv.org/abs/2311.05332v1
On the Road with GPT-4V(ision): Early Explorations of Visual-Language Model on Autonomous Driving
Licheng Wen, Xuemeng Yang, Daocheng Fu, Xiaofeng Wang, Pinlong Cai, Xin Li, Tao Ma, Yingxuan Li, Linran Xu, Dengke Shang, Zheng Zhu, Shaoyan Sun, Yeqi Bai, Xinyu Cai, Min Dou, Shuanglu Hu, Botian Shi
cs.CV, cs.AI, cs.CL, cs.RO
The pursuit of autonomous driving technology hinges on the sophisticated integration of perception, decision-making, and control systems. Traditional approaches, both data-driven and rule-based, have been hindered by their inability to grasp the nuance of complex driving environments and the intentions of other road us...
http://arxiv.org/pdf/2311.05332v1
2023-11-09T12:58:37
2023-11-09T12:58:37
{ "title": [ -0.0073623657, 0.00021457672, -0.016723633, -0.00021755695, -0.00006210804, -0.022506714, 0.011428833, -0.03652954, -0.022323608, 0.023330688, -0.054992676, 0.02670288, -0.02619934, -0.009635925, -0.026687622, -0.023544312, -0.0034828186...
http://arxiv.org/abs/2311.05297v1
Do personality tests generalize to Large Language Models?
Florian E. Dorner, Tom Sühr, Samira Samadi, Augustin Kelava
cs.CL, cs.AI, cs.LG
With large language models (LLMs) appearing to behave increasingly human-like in text-based interactions, it has become popular to attempt to evaluate various properties of these models using tests originally designed for humans. While re-using existing tests is a resource-efficient way to evaluate LLMs, careful adjust...
http://arxiv.org/pdf/2311.05297v1
2023-11-09T11:54:01
2023-11-09T11:54:01
{ "title": [ -0.025680542, -0.004085541, -0.023345947, -0.028045654, -0.03479004, -0.0056037903, -0.039215088, 0.0018758774, -0.010101318, 0.032806396, -0.019134521, -0.002784729, 0.023529053, -0.027175903, -0.015975952, 0.011131287, -0.0022583008, ...
http://arxiv.org/abs/2311.05296v1
DeeLM: Dependency-enhanced Large Language Model for Sentence Embeddings
Xianming Li, Jing Li
cs.CL
Recent studies have proposed using large language models (LLMs) for sentence embeddings. However, most existing LLMs are built with an autoregressive architecture that primarily captures forward dependencies while neglecting backward dependencies. Previous work has highlighted the importance of backward dependencies in...
http://arxiv.org/pdf/2311.05296v1
2023-11-09T11:53:52
2023-11-09T11:53:52
{ "title": [ -0.0057029724, -0.006439209, -0.039642334, -0.07897949, 0.014755249, 0.010139465, -0.04748535, 0.009376526, 0.02659607, 0.010971069, -0.01184082, 0.010253906, 0.00642395, -0.0140686035, -0.030517578, 0.019515991, -0.019470215, 0.0540...
http://arxiv.org/abs/2311.05286v1
Causal Inference from Text: Unveiling Interactions between Variables
Yuxiang Zhou, Yulan He
cs.CL
Adjusting for latent covariates is crucial for estimating causal effects from observational textual data. Most existing methods only account for confounding covariates that affect both treatment and outcome, potentially leading to biased causal effects. This bias arises from insufficient consideration of non-confoundin...
http://arxiv.org/pdf/2311.05286v1
2023-11-09T11:29:44
2023-11-09T11:29:44
{ "title": [ -0.024353027, -0.010871887, -0.038757324, -0.06707764, 0.0054397583, -0.002664566, -0.044891357, -0.01083374, -0.009788513, 0.0026359558, -0.015304565, 0.03152466, -0.02935791, -0.019699097, -0.01626587, -0.0020256042, -0.010375977, ...
http://arxiv.org/abs/2311.05268v1
Modelling prospective memory and resilient situated communications via Wizard of Oz
Yanzhe Li, Frank Broz, Mark Neerincx
cs.CL
This abstract presents a scenario for human-robot action in a home setting involving an older adult and a robot. The scenario is designed to explore the envisioned modelling of memory for communication with a socially assistive robots (SAR). The scenario will enable the gathering of data on failures of speech technolog...
http://arxiv.org/pdf/2311.05268v1
2023-11-09T10:50:36
2023-11-09T10:50:36
{ "title": [ 0.043701172, -0.009307861, 0.003774643, -0.06555176, -0.0038967133, -0.005634308, 0.0036296844, -0.0019283295, -0.016357422, 0.029754639, 0.020080566, 0.00031924248, 0.0036849976, -0.04055786, -0.02796936, -0.004924774, -0.010765076, ...
http://arxiv.org/abs/2311.05265v1
Don't Waste a Single Annotation: Improving Single-Label Classifiers Through Soft Labels
Ben Wu, Yue Li, Yida Mu, Carolina Scarton, Kalina Bontcheva, Xingyi Song
cs.CL, cs.AI, cs.LG
In this paper, we address the limitations of the common data annotation and training methods for objective single-label classification tasks. Typically, when annotating such tasks annotators are only asked to provide a single label for each sample and annotator disagreement is discarded when a final hard label is decid...
http://arxiv.org/pdf/2311.05265v1
2023-11-09T10:47:39
2023-11-09T10:47:39
{ "title": [ -0.02104187, -0.021408081, -0.016052246, -0.04156494, -0.006828308, 0.0029010773, -0.010383606, -0.0050735474, -0.0054130554, 0.022125244, -0.017547607, 0.020584106, -0.012825012, -0.017822266, -0.024475098, 0.00052785873, 0.0082473755, ...
http://arxiv.org/abs/2311.05263v1
Model-Based Minimum Bayes Risk Decoding
Yuu Jinnai, Tetsuro Morimura, Ukyo Honda, Kaito Ariu, Kenshi Abe
cs.AI, cs.CL
Minimum Bayes Risk (MBR) decoding has been shown to be a powerful alternative to beam search decoding in a variety of text generation tasks. MBR decoding selects a hypothesis from a pool of hypotheses that has the least expected risk under a probability model according to a given utility function. Since it is impractic...
http://arxiv.org/pdf/2311.05263v1
2023-11-09T10:46:09
2023-11-09T10:46:09
{ "title": [ -0.059173584, -0.015541077, -0.05895996, -0.035095215, -0.012763977, 0.0012178421, -0.041015625, -0.00623703, -0.008361816, 0.049957275, -0.012413025, 0.006126404, 0.0121536255, -0.019897461, -0.03942871, 0.009811401, -0.00052547455, ...
http://arxiv.org/abs/2311.05232v1
A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, Ting Liu
cs.CL
The emergence of large language models (LLMs) has marked a significant breakthrough in natural language processing (NLP), leading to remarkable advancements in text understanding and generation. Nevertheless, alongside these strides, LLMs exhibit a critical tendency to produce hallucinations, resulting in content that ...
http://arxiv.org/pdf/2311.05232v1
2023-11-09T09:25:37
2023-11-09T09:25:37
{ "title": [ -0.010223389, -0.0027809143, -0.015014648, -0.053710938, -0.007633209, 0.0031738281, -0.03314209, -0.010368347, 0.006603241, 0.044433594, -0.0031776428, 0.012496948, 0.030670166, -0.016738892, -0.047332764, 0.028076172, -0.03161621, ...
http://arxiv.org/abs/2311.05195v1
PRODIGy: a PROfile-based DIalogue Generation dataset
Daniela Occhipinti, Serra Sinem Tekiroglu, Marco Guerini
cs.CL
Providing dialogue agents with a profile representation can improve their consistency and coherence, leading to better conversations. However, current profile-based dialogue datasets for training such agents contain either explicit profile representations that are simple and dialogue-specific, or implicit representatio...
http://arxiv.org/pdf/2311.05195v1
2023-11-09T08:19:34
2023-11-09T08:19:34
{ "title": [ -0.0039100647, -0.011550903, -0.04309082, -0.079711914, 0.010154724, 0.00076818466, -0.052703857, 0.017974854, 0.008117676, 0.016983032, 0.004650116, 0.0027332306, -0.026794434, -0.021835327, -0.043884277, 0.0070648193, -0.020019531, ...
http://arxiv.org/abs/2311.05661v1
Prompt Engineering a Prompt Engineer
Qinyuan Ye, Maxamed Axmed, Reid Pryzant, Fereshte Khani
cs.CL, cs.AI, cs.LG
Prompt engineering is a challenging yet crucial task for optimizing the performance of large language models (LLMs). It requires complex reasoning to examine the model's errors, hypothesize what is missing or misleading in the current prompt, and communicate the task with clarity. While recent works indicate that LLMs ...
http://arxiv.org/pdf/2311.05661v1
2023-11-09T08:00:32
2023-11-09T08:00:32
{ "title": [ -0.0050697327, 0.011642456, -0.04345703, 0.008422852, -0.03314209, -0.033081055, -0.014694214, -0.032836914, -0.012786865, 0.04373169, -0.006801605, -0.034576416, 0.011291504, -0.04827881, 0.009399414, 0.003332138, 0.026794434, 0.018...
http://arxiv.org/abs/2311.05169v1
Large Language Models and Prompt Engineering for Biomedical Query Focused Multi-Document Summarisation
Diego Mollá
cs.CL
This paper reports on the use of prompt engineering and GPT-3.5 for biomedical query-focused multi-document summarisation. Using GPT-3.5 and appropriate prompts, our system achieves top ROUGE-F1 results in the task of obtaining short-paragraph-sized answers to biomedical questions in the 2023 BioASQ Challenge (BioASQ 1...
http://arxiv.org/pdf/2311.05169v1
2023-11-09T06:45:04
2023-11-09T06:45:04
{ "title": [ -0.013626099, -0.014137268, -0.03555298, -0.07928467, -0.017608643, 0.012145996, -0.081970215, 0.013542175, 0.009590149, 0.046539307, 0.00945282, -0.016998291, 0.016677856, -0.025497437, -0.06970215, 0.014595032, 0.0062789917, 0.0072...
http://arxiv.org/abs/2311.05161v1
Enhancing Computation Efficiency in Large Language Models through Weight and Activation Quantization
Jangwhan Lee, Minsoo Kim, Seungcheol Baek, Seok Joong Hwang, Wonyong Sung, Jungwook Choi
cs.CL
Large Language Models (LLMs) are proficient in natural language processing tasks, but their deployment is often restricted by extensive parameter sizes and computational demands. This paper focuses on post-training quantization (PTQ) in LLMs, specifically 4-bit weight and 8-bit activation (W4A8) quantization, to enhanc...
http://arxiv.org/pdf/2311.05161v1
2023-11-09T06:19:51
2023-11-09T06:19:51
{ "title": [ -0.027313232, -0.005683899, -0.060577393, -0.080322266, 0.012336731, -0.011077881, -0.06982422, 0.017425537, 0.010787964, 0.05508423, 0.005558014, -0.0073165894, 0.017990112, -0.01751709, -0.05783081, 0.019744873, -0.007972717, 0.036...
http://arxiv.org/abs/2311.07589v1
Dialogizer: Context-aware Conversational-QA Dataset Generation from Textual Sources
Yerin Hwang, Yongil Kim, Hyunkyung Bae, Jeesoo Bang, Hwanhee Lee, Kyomin Jung
cs.CL, cs.AI
To address the data scarcity issue in Conversational question answering (ConvQA), a dialog inpainting method, which utilizes documents to generate ConvQA datasets, has been proposed. However, the original dialog inpainting model is trained solely on the dialog reconstruction task, resulting in the generation of questio...
http://arxiv.org/pdf/2311.07589v1
2023-11-09T06:03:11
2023-11-09T06:03:11
{ "title": [ -0.020553589, -0.0024909973, -0.061401367, -0.06213379, 0.010360718, -0.008293152, -0.042419434, 0.016036987, 0.025131226, 0.044891357, -0.03475952, 0.01399231, -0.007659912, 0.0043029785, -0.07763672, -0.009338379, -0.020996094, 0.0...
http://arxiv.org/abs/2311.05155v1
Weakly-supervised Deep Cognate Detection Framework for Low-Resourced Languages Using Morphological Knowledge of Closely-Related Languages
Koustava Goswami, Priya Rani, Theodorus Fransen, John P. McCrae
cs.CL, cs.AI
Exploiting cognates for transfer learning in under-resourced languages is an exciting opportunity for language understanding tasks, including unsupervised machine translation, named entity recognition and information retrieval. Previous approaches mainly focused on supervised cognate detection tasks based on orthograph...
http://arxiv.org/pdf/2311.05155v1
2023-11-09T05:46:41
2023-11-09T05:46:41
{ "title": [ 0.002872467, 0.013282776, -0.013450623, -0.055999756, -0.004421234, 0.0016889572, -0.060394287, -0.01576233, 0.0018815994, 0.023910522, -0.02519226, 0.029174805, 0.022979736, -0.026229858, -0.0042686462, 0.013389587, -0.00674057, 0.0...
http://arxiv.org/abs/2311.05120v1
Quranic Conversations: Developing a Semantic Search tool for the Quran using Arabic NLP Techniques
Yasser Shohoud, Maged Shoman, Sarah Abdelazim
cs.CL
The Holy Book of Quran is believed to be the literal word of God (Allah) as revealed to the Prophet Muhammad (PBUH) over a period of approximately 23 years. It is the book where God provides guidance on how to live a righteous and just life, emphasizing principles like honesty, compassion, charity and justice, as well ...
http://arxiv.org/pdf/2311.05120v1
2023-11-09T03:14:54
2023-11-09T03:14:54
{ "title": [ -0.028579712, -0.014259338, -0.044189453, 0.000012040138, -0.015235901, -0.008834839, -0.00055217743, 0.0013780594, 0.0024375916, 0.041992188, -0.012367249, -0.0040245056, 0.03591919, -0.028808594, -0.030685425, -0.0010690689, 0.01296234...
http://arxiv.org/abs/2311.05117v1
Unsupervised Translation Quality Estimation Exploiting Synthetic Data and Pre-trained Multilingual Encoder
Yuto Kuroda, Atsushi Fujita, Tomoyuki Kajiwara, Takashi Ninomiya
cs.CL
Translation quality estimation (TQE) is the task of predicting translation quality without reference translations. Due to the enormous cost of creating training data for TQE, only a few translation directions can benefit from supervised training. To address this issue, unsupervised TQE methods have been studied. In thi...
http://arxiv.org/pdf/2311.05117v1
2023-11-09T03:10:42
2023-11-09T03:10:42
{ "title": [ -0.051818848, 0.0073776245, -0.01399231, -0.026809692, -0.021759033, 0.011871338, -0.04824829, -0.024383545, -0.001578331, 0.036956787, -0.016677856, 0.0049438477, 0.0041275024, -0.023712158, -0.032318115, 0.006214142, -0.00756073, 0...
http://arxiv.org/abs/2311.05113v1
Conic10K: A Challenging Math Problem Understanding and Reasoning Dataset
Haoyi Wu, Wenyang Hui, Yezeng Chen, Weiqi Wu, Kewei Tu, Yi Zhou
cs.CL
Mathematical understanding and reasoning are crucial tasks for assessing the capabilities of artificial intelligence (AI). However, existing benchmarks either require just a few steps of reasoning, or only contain a small amount of data in one specific topic, making it hard to analyse AI's behaviour with reference to d...
http://arxiv.org/pdf/2311.05113v1
2023-11-09T02:58:17
2023-11-09T02:58:17
{ "title": [ -0.010551453, -0.014595032, -0.037353516, -0.034118652, -0.004470825, -0.025405884, -0.026062012, -0.017211914, 0.012329102, 0.0061416626, -0.02961731, -0.008071899, -0.007888794, -0.030181885, -0.021011353, -0.011436462, 0.0075950623, ...
http://arxiv.org/abs/2311.05112v1
A Survey of Large Language Models in Medicine: Progress, Application, and Challenge
Hongjian Zhou, Boyang Gu, Xinyu Zou, Yiru Li, Sam S. Chen, Peilin Zhou, Junling Liu, Yining Hua, Chengfeng Mao, Xian Wu, Zheng Li, Fenglin Liu
cs.CL, cs.AI
Large language models (LLMs), such as ChatGPT, have achieved substantial attention due to their impressive human language understanding and generation capabilities. Therefore, the application of LLMs in medicine to assist physicians and patient care emerges as a promising research direction in both artificial intellige...
http://arxiv.org/pdf/2311.05112v1
2023-11-09T02:55:58
2023-11-09T02:55:58
{ "title": [ -0.03366089, -0.013076782, -0.040252686, -0.073913574, 0.004360199, 0.01335907, -0.07299805, -0.003124237, -0.003578186, 0.034484863, 0.016555786, -0.0076446533, 0.012573242, -0.04171753, -0.027328491, 0.026535034, -0.00617218, 0.028...
http://arxiv.org/abs/2311.05089v1
Legal-HNet: Mixing Legal Long-Context Tokens with Hartley Transform
Daniele Giofré, Sneha Ghantasala
cs.CL, cs.AI
Since its introduction, the transformers architecture has seen great adoption in NLP applications, but it also has limitations. Although the self-attention mechanism allows for generating very rich representations of the input text, its effectiveness may be limited in specialized domains such as legal, where, for examp...
http://arxiv.org/pdf/2311.05089v1
2023-11-09T01:27:54
2023-11-09T01:27:54
{ "title": [ -0.016784668, -0.0021629333, -0.057861328, -0.0680542, 0.016540527, -0.004283905, -0.020751953, 0.013771057, -0.0064201355, 0.04498291, -0.039733887, 0.014976501, 0.046203613, -0.021133423, -0.013931274, 0.006526947, -0.015060425, 0....