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/2310.12150v1
Understanding Retrieval Augmentation for Long-Form Question Answering
Hung-Ting Chen, Fangyuan Xu, Shane Arora, Eunsol Choi
cs.CL
We present a study of retrieval-augmented language models (LMs) on long-form question answering. We analyze how retrieval augmentation impacts different LMs, by comparing answers generated from models while using the same evidence documents, and how differing quality of retrieval document set impacts the answers genera...
http://arxiv.org/pdf/2310.12150v1
2023-10-18T17:59:10
2023-10-18T17:59:10
{ "title": [ 0.023284912, 0.002521515, -0.02734375, -0.051483154, 0.008842468, -0.011009216, 0.009941101, 0.0025100708, 0.018325806, 0.031585693, -0.029327393, 0.009284973, 0.00217247, -0.0024662018, -0.05508423, 0.0061950684, -0.0015182495, 0.01...
http://arxiv.org/abs/2310.12143v1
Simple Mechanisms for Representing, Indexing and Manipulating Concepts
Yuanzhi Li, Raghu Meka, Rina Panigrahy, Kulin Shah
cs.LG, cs.CL, stat.ML
Deep networks typically learn concepts via classifiers, which involves setting up a model and training it via gradient descent to fit the concept-labeled data. We will argue instead that learning a concept could be done by looking at its moment statistics matrix to generate a concrete representation or signature of tha...
http://arxiv.org/pdf/2310.12143v1
2023-10-18T17:54:29
2023-10-18T17:54:29
{ "title": [ 0.020263672, -0.036865234, -0.0053215027, -0.04284668, -0.027282715, -0.0124053955, -0.0030994415, -0.0020637512, 0.024459839, 0.028793335, -0.025604248, -0.019622803, 0.025863647, -0.027862549, -0.040130615, -0.012924194, 0.02229309, ...
http://arxiv.org/abs/2310.12135v1
Pseudointelligence: A Unifying Framework for Language Model Evaluation
Shikhar Murty, Orr Paradise, Pratyusha Sharma
cs.CL
With large language models surpassing human performance on an increasing number of benchmarks, we must take a principled approach for targeted evaluation of model capabilities. Inspired by pseudorandomness, we propose pseudointelligence, which captures the maxim that "(perceived) intelligence lies in the eye of the beh...
http://arxiv.org/pdf/2310.12135v1
2023-10-18T17:48:05
2023-10-18T17:48:05
{ "title": [ -0.021224976, -0.00774765, -0.025299072, -0.051727295, 0.007965088, 0.000998497, -0.045715332, -0.011116028, -0.00026988983, 0.02998352, 0.003786087, -0.005836487, -0.00093078613, -0.02696228, -0.06488037, 0.009750366, -0.019866943, ...
http://arxiv.org/abs/2310.12128v1
DiagrammerGPT: Generating Open-Domain, Open-Platform Diagrams via LLM Planning
Abhay Zala, Han Lin, Jaemin Cho, Mohit Bansal
cs.CV, cs.AI, cs.CL, cs.LG
Text-to-image (T2I) generation has seen significant growth over the past few years. Despite this, there has been little work on generating diagrams with T2I models. A diagram is a symbolic/schematic representation that explains information using structurally rich and spatially complex visualizations (e.g., a dense comb...
http://arxiv.org/pdf/2310.12128v1
2023-10-18T17:37:10
2023-10-18T17:37:10
{ "title": [ -0.017837524, -0.0069732666, -0.03933716, -0.043670654, -0.019241333, 0.006340027, -0.06951904, -0.031951904, -0.032958984, 0.011756897, -0.004131317, 0.03225708, -0.00024914742, -0.046295166, 0.03753662, 0.005886078, 0.0016183853, 0...
http://arxiv.org/abs/2310.12127v2
A Tale of Pronouns: Interpretability Informs Gender Bias Mitigation for Fairer Instruction-Tuned Machine Translation
Giuseppe Attanasio, Flor Miriam Plaza-del-Arco, Debora Nozza, Anne Lauscher
cs.CL, cs.LG
Recent instruction fine-tuned models can solve multiple NLP tasks when prompted to do so, with machine translation (MT) being a prominent use case. However, current research often focuses on standard performance benchmarks, leaving compelling fairness and ethical considerations behind. In MT, this might lead to misgend...
http://arxiv.org/pdf/2310.12127v2
2023-10-25T13:43:49
2023-10-18T17:36:55
{ "title": [ -0.049713135, -0.009994507, -0.02130127, -0.037139893, -0.023239136, 0.00274086, -0.04336548, -0.0014572144, -0.004634857, 0.026885986, -0.01576233, -0.006969452, 0.03955078, -0.018875122, -0.036102295, -0.008430481, -0.027282715, 0....
http://arxiv.org/abs/2310.12126v1
SHARCS: Efficient Transformers through Routing with Dynamic Width Sub-networks
Mohammadreza Salehi, Sachin Mehta, Aditya Kusupati, Ali Farhadi, Hannaneh Hajishirzi
cs.LG, cs.AI, cs.CL
We introduce SHARCS for adaptive inference that takes into account the hardness of input samples. SHARCS can train a router on any transformer network, enabling the model to direct different samples to sub-networks with varying widths. Our experiments demonstrate that: (1) SHARCS outperforms or complements existing per...
http://arxiv.org/pdf/2310.12126v1
2023-10-18T17:35:15
2023-10-18T17:35:15
{ "title": [ -0.05795288, -0.013763428, -0.044067383, -0.045318604, 0.00007390976, 0.027023315, -0.042236328, -0.004852295, -0.027954102, 0.03152466, -0.02911377, 0.021118164, -0.0079193115, -0.02268982, -0.015083313, -0.0027389526, 0.009635925, ...
http://arxiv.org/abs/2310.12118v1
Harnessing Dataset Cartography for Improved Compositional Generalization in Transformers
Osman Batur İnce, Tanin Zeraati, Semih Yagcioglu, Yadollah Yaghoobzadeh, Erkut Erdem, Aykut Erdem
cs.CL
Neural networks have revolutionized language modeling and excelled in various downstream tasks. However, the extent to which these models achieve compositional generalization comparable to human cognitive abilities remains a topic of debate. While existing approaches in the field have mainly focused on novel architectu...
http://arxiv.org/pdf/2310.12118v1
2023-10-18T17:14:41
2023-10-18T17:14:41
{ "title": [ 0.0068130493, 0.0041885376, -0.0178833, -0.07104492, 0.0034503937, 0.0073432922, -0.04815674, -0.0057640076, 0.0021762848, 0.037322998, -0.0541687, 0.016464233, -0.023345947, -0.0033245087, -0.04574585, -0.0209198, 0.0003771782, 0.04...
http://arxiv.org/abs/2310.12100v1
Non-Intrusive Adaptation: Input-Centric Parameter-efficient Fine-Tuning for Versatile Multimodal Modeling
Yaqing Wang, Jialin Wu, Tanmaya Dabral, Jiageng Zhang, Geoff Brown, Chun-Ta Lu, Frederick Liu, Yi Liang, Bo Pang, Michael Bendersky, Radu Soricut
cs.CL, cs.AI, cs.CV, cs.LG, cs.MM
Large language models (LLMs) and vision language models (VLMs) demonstrate excellent performance on a wide range of tasks by scaling up parameter counts from O(10^9) to O(10^{12}) levels and further beyond. These large scales make it impossible to adapt and deploy fully specialized models given a task of interest. Para...
http://arxiv.org/pdf/2310.12100v1
2023-10-18T16:43:08
2023-10-18T16:43:08
{ "title": [ -0.021514893, -0.012252808, -0.032043457, -0.065979004, -0.017440796, -0.00017297268, -0.058563232, -0.001083374, -0.0021018982, 0.045532227, 0.022857666, -0.00092840195, 0.017150879, -0.026947021, -0.043273926, 0.014556885, -0.009658813...
http://arxiv.org/abs/2310.13017v1
Position Interpolation Improves ALiBi Extrapolation
Faisal Al-Khateeb, Nolan Dey, Daria Soboleva, Joel Hestness
cs.CL, cs.AI, cs.LG
Linear position interpolation helps pre-trained models using rotary position embeddings (RoPE) to extrapolate to longer sequence lengths. We propose using linear position interpolation to extend the extrapolation range of models using Attention with Linear Biases (ALiBi). We find position interpolation significantly im...
http://arxiv.org/pdf/2310.13017v1
2023-10-18T16:41:47
2023-10-18T16:41:47
{ "title": [ 0.036956787, 0.009590149, -0.012771606, 0.007858276, -0.035186768, -0.041534424, -0.011787415, -0.06201172, -0.042053223, 0.05206299, -0.02848816, 0.007446289, 0.049072266, -0.010559082, 0.023986816, -0.006111145, -0.0032043457, 0.00...
http://arxiv.org/abs/2310.12086v1
Unveiling the Siren's Song: Towards Reliable Fact-Conflicting Hallucination Detection
Xiang Chen, Duanzheng Song, Honghao Gui, Chengxi Wang, Ningyu Zhang, Fei Huang, Chengfei Lv, Dan Zhang, Huajun Chen
cs.CL, cs.AI, cs.CV, cs.IR, cs.LG
Large Language Models (LLMs), such as ChatGPT/GPT-4, have garnered widespread attention owing to their myriad of practical applications, yet their adoption has been constrained by issues of fact-conflicting hallucinations across web platforms. The assessment of factuality in text, produced by LLMs, remains inadequately...
http://arxiv.org/pdf/2310.12086v1
2023-10-18T16:27:49
2023-10-18T16:27:49
{ "title": [ 0.024612427, 0.013420105, -0.017578125, -0.019088745, 0.017410278, -0.023361206, 0.018203735, -0.051696777, -0.028274536, 0.029830933, -0.0061912537, 0.01525116, -0.011985779, -0.021697998, -0.039855957, -0.008758545, -0.019897461, 0...
http://arxiv.org/abs/2310.12085v1
On the Benefit of Generative Foundation Models for Human Activity Recognition
Zikang Leng, Hyeokhyen Kwon, Thomas Plötz
cs.CV, cs.CL
In human activity recognition (HAR), the limited availability of annotated data presents a significant challenge. Drawing inspiration from the latest advancements in generative AI, including Large Language Models (LLMs) and motion synthesis models, we believe that generative AI can address this data scarcity by autonom...
http://arxiv.org/pdf/2310.12085v1
2023-10-18T16:27:06
2023-10-18T16:27:06
{ "title": [ 0.005622864, 0.024017334, -0.053497314, -0.062927246, 0.017456055, -0.024627686, -0.053588867, -0.021118164, 0.0006623268, 0.026290894, -0.029708862, 0.029953003, -0.02670288, -0.0095825195, -0.04006958, -0.00064468384, -0.014030457, ...
http://arxiv.org/abs/2310.12074v2
Towards Safer Operations: An Expert-involved Dataset of High-Pressure Gas Incidents for Preventing Future Failures
Shumpei Inoue, Minh-Tien Nguyen, Hiroki Mizokuchi, Tuan-Anh D. Nguyen, Huu-Hiep Nguyen, Dung Tien Le
cs.CL
This paper introduces a new IncidentAI dataset for safety prevention. Different from prior corpora that usually contain a single task, our dataset comprises three tasks: named entity recognition, cause-effect extraction, and information retrieval. The dataset is annotated by domain experts who have at least six years o...
http://arxiv.org/pdf/2310.12074v2
2023-10-23T08:01:46
2023-10-18T16:07:13
{ "title": [ 0.03994751, 0.0056610107, -0.059295654, -0.09716797, 0.028778076, -0.03817749, -0.02508545, 0.023330688, -0.048095703, 0.0037612915, -0.010856628, 0.034423828, -0.044067383, -0.03540039, 0.018585205, -0.027862549, 0.018173218, 0.0127...
http://arxiv.org/abs/2310.12072v1
SPEED: Speculative Pipelined Execution for Efficient Decoding
Coleman Hooper, Sehoon Kim, Hiva Mohammadzadeh, Hasan Genc, Kurt Keutzer, Amir Gholami, Sophia Shao
cs.CL
Generative Large Language Models (LLMs) based on the Transformer architecture have recently emerged as a dominant foundation model for a wide range of Natural Language Processing tasks. Nevertheless, their application in real-time scenarios has been highly restricted due to the significant inference latency associated ...
http://arxiv.org/pdf/2310.12072v1
2023-10-18T16:07:01
2023-10-18T16:07:01
{ "title": [ -0.044189453, -0.027877808, -0.04623413, -0.057739258, -0.04272461, -0.005680084, -0.045440674, 0.024993896, -0.015365601, 0.057769775, -0.0037956238, 0.0016698837, 0.019165039, -0.037200928, -0.033721924, 0.0050315857, 0.018844604, ...
http://arxiv.org/abs/2310.12064v1
Code Book for the Annotation of Diverse Cross-Document Coreference of Entities in News Articles
Jakob Vogel
cs.CL
This paper presents a scheme for annotating coreference across news articles, extending beyond traditional identity relations by also considering near-identity and bridging relations. It includes a precise description of how to set up Inception, a respective annotation tool, how to annotate entities in news articles, c...
http://arxiv.org/pdf/2310.12064v1
2023-10-18T15:53:45
2023-10-18T15:53:45
{ "title": [ 0.0052223206, 0.000770092, -0.02571106, -0.021835327, -0.015281677, -0.025344849, -0.018554688, -0.020599365, -0.005794525, 0.0524292, -0.05895996, 0.02696228, 0.066223145, -0.0025959015, 0.021194458, 0.010314941, -0.0020599365, 0.04...
http://arxiv.org/abs/2310.12059v2
Evaluating the Symbol Binding Ability of Large Language Models for Multiple-Choice Questions in Vietnamese General Education
Duc-Vu Nguyen, Quoc-Nam Nguyen
cs.CL
In this paper, we evaluate the ability of large language models (LLMs) to perform multiple choice symbol binding (MCSB) for multiple choice question answering (MCQA) tasks in zero-shot, one-shot, and few-shot settings. We focus on Vietnamese, with fewer challenging MCQA datasets than in English. The two existing datase...
http://arxiv.org/pdf/2310.12059v2
2023-10-23T03:33:18
2023-10-18T15:48:07
{ "title": [ -0.00957489, -0.011199951, 0.005569458, 0.019088745, -0.02696228, -0.005138397, 0.017471313, -0.025222778, 0.008491516, 0.028503418, -0.010116577, -0.018234253, -0.018707275, -0.03741455, -0.05987549, -0.026184082, -0.015579224, 0.00...
http://arxiv.org/abs/2310.12049v1
Concept-Guided Chain-of-Thought Prompting for Pairwise Comparison Scaling of Texts with Large Language Models
Patrick Y. Wu, Jonathan Nagler, Joshua A. Tucker, Solomon Messing
cs.CL, cs.CY
Existing text scaling methods often require a large corpus, struggle with short texts, or require labeled data. We develop a text scaling method that leverages the pattern recognition capabilities of generative large language models (LLMs). Specifically, we propose concept-guided chain-of-thought (CGCoT), which uses pr...
http://arxiv.org/pdf/2310.12049v1
2023-10-18T15:34:37
2023-10-18T15:34:37
{ "title": [ -0.012893677, -0.017623901, -0.017745972, -0.021499634, -0.03277588, -0.011512756, -0.027832031, -0.015113831, 0.031799316, 0.012817383, -0.04434204, 0.025848389, 0.003063202, -0.007801056, -0.04473877, -0.012771606, -0.0076522827, 0...
http://arxiv.org/abs/2310.12024v1
CORE: A Few-Shot Company Relation Classification Dataset for Robust Domain Adaptation
Philipp Borchert, Jochen De Weerdt, Kristof Coussement, Arno De Caigny, Marie-Francine Moens
cs.CL
We introduce CORE, a dataset for few-shot relation classification (RC) focused on company relations and business entities. CORE includes 4,708 instances of 12 relation types with corresponding textual evidence extracted from company Wikipedia pages. Company names and business entities pose a challenge for few-shot RC m...
http://arxiv.org/pdf/2310.12024v1
2023-10-18T14:58:13
2023-10-18T14:58:13
{ "title": [ -0.009017944, -0.012298584, -0.064941406, -0.047698975, 0.013557434, -0.024627686, -0.040222168, -0.0005803108, 0.021697998, 0.002412796, -0.021743774, -0.0034446716, -0.013954163, -0.01335907, -0.031158447, -0.0113220215, -0.023422241, ...
http://arxiv.org/abs/2310.12020v2
LoHoRavens: A Long-Horizon Language-Conditioned Benchmark for Robotic Tabletop Manipulation
Shengqiang Zhang, Philipp Wicke, Lütfi Kerem Şenel, Luis Figueredo, Abdeldjallil Naceri, Sami Haddadin, Barbara Plank, Hinrich Schütze
cs.RO, cs.CL, cs.CV
The convergence of embodied agents and large language models (LLMs) has brought significant advancements to embodied instruction following. Particularly, the strong reasoning capabilities of LLMs make it possible for robots to perform long-horizon tasks without expensive annotated demonstrations. However, public benchm...
http://arxiv.org/pdf/2310.12020v2
2023-10-23T12:29:33
2023-10-18T14:53:14
{ "title": [ -0.015731812, 0.008964539, -0.0082092285, -0.028152466, -0.0039978027, -0.0067214966, -0.026275635, -0.029220581, -0.0011014938, 0.042663574, -0.012031555, 0.05038452, 0.00026607513, -0.0040779114, -0.015853882, 0.0061950684, 0.018951416...
http://arxiv.org/abs/2310.12011v1
Gold: A Global and Local-aware Denoising Framework for Commonsense Knowledge Graph Noise Detection
Zheye Deng, Weiqi Wang, Zhaowei Wang, Xin Liu, Yangqiu Song
cs.CL
Commonsense Knowledge Graphs (CSKGs) are crucial for commonsense reasoning, yet constructing them through human annotations can be costly. As a result, various automatic methods have been proposed to construct CSKG with larger semantic coverage. However, these unsupervised approaches introduce spurious noise that can l...
http://arxiv.org/pdf/2310.12011v1
2023-10-18T14:43:07
2023-10-18T14:43:07
{ "title": [ -0.014564514, -0.0044174194, -0.06439209, -0.07092285, -0.0046844482, -0.009918213, -0.050048828, 0.007499695, 0.042236328, 0.022003174, -0.023605347, 0.015037537, 0.0055885315, -0.020751953, -0.032684326, 0.008888245, 0.0053367615, ...
http://arxiv.org/abs/2310.12008v1
Multi-view Contrastive Learning for Entity Typing over Knowledge Graphs
Zhiwei Hu, Víctor Gutiérrez-Basulto, Zhiliang Xiang, Ru Li, Jeff Z. Pan
cs.CL, cs.AI
Knowledge graph entity typing (KGET) aims at inferring plausible types of entities in knowledge graphs. Existing approaches to KGET focus on how to better encode the knowledge provided by the neighbors and types of an entity into its representation. However, they ignore the semantic knowledge provided by the way in whi...
http://arxiv.org/pdf/2310.12008v1
2023-10-18T14:41:09
2023-10-18T14:41:09
{ "title": [ -0.024536133, -0.000076532364, -0.05230713, -0.062561035, -0.00030589104, -0.0062828064, -0.048034668, 0.0057525635, 0.03933716, 0.022903442, -0.03451538, 0.0078125, 0.008796692, -0.013442993, -0.035369873, -0.0037403107, -0.026123047, ...
http://arxiv.org/abs/2310.11986v2
Sociotechnical Safety Evaluation of Generative AI Systems
Laura Weidinger, Maribeth Rauh, Nahema Marchal, Arianna Manzini, Lisa Anne Hendricks, Juan Mateos-Garcia, Stevie Bergman, Jackie Kay, Conor Griffin, Ben Bariach, Iason Gabriel, Verena Rieser, William Isaac
cs.AI, cs.CL, cs.CY
Generative AI systems produce a range of risks. To ensure the safety of generative AI systems, these risks must be evaluated. In this paper, we make two main contributions toward establishing such evaluations. First, we propose a three-layered framework that takes a structured, sociotechnical approach to evaluating the...
http://arxiv.org/pdf/2310.11986v2
2023-10-31T18:23:32
2023-10-18T14:13:58
{ "title": [ -0.0036773682, 0.025741577, -0.03213501, -0.06286621, 0.027816772, -0.014915466, -0.015289307, -0.021606445, -0.023345947, 0.02645874, -0.028808594, 0.039764404, -0.009231567, -0.010147095, -0.009223938, 0.0057373047, -0.016296387, 0...
http://arxiv.org/abs/2310.11984v1
From Interpolation to Extrapolation: Complete Length Generalization for Arithmetic Transformers
Shaoxiong Duan, Yining Shi
cs.LG, cs.CL
Since its introduction, the transformer model has demonstrated outstanding performance across various tasks. However, there are still unresolved issues regarding length generalization, particularly in algorithmic tasks. In this paper, we investigate the inherent capabilities of transformer models in learning arithmetic...
http://arxiv.org/pdf/2310.11984v1
2023-10-18T14:10:47
2023-10-18T14:10:47
{ "title": [ -0.010978699, 0.0048446655, -0.015777588, -0.06665039, 0.00655365, 0.002506256, -0.023162842, -0.018112183, -0.030761719, 0.042907715, -0.023101807, 0.02633667, -0.022216797, -0.008834839, -0.0335083, -0.007881165, 0.00043463707, 0.0...
http://arxiv.org/abs/2310.11976v1
InfoDiffusion: Information Entropy Aware Diffusion Process for Non-Autoregressive Text Generation
Renzhi Wang, Jing Li, Piji Li
cs.CL, cs.AI
Diffusion models have garnered considerable interest in the field of text generation. Several studies have explored text diffusion models with different structures and applied them to various tasks, including named entity recognition and summarization. However, there exists a notable disparity between the "easy-first" ...
http://arxiv.org/pdf/2310.11976v1
2023-10-18T14:01:39
2023-10-18T14:01:39
{ "title": [ -0.028686523, -0.003768921, -0.055786133, -0.03729248, -0.0029716492, -0.0030403137, -0.030960083, 0.001285553, 0.02268982, 0.04296875, -0.02558899, 0.035308838, 0.029022217, 0.003376007, -0.053649902, 0.01979065, -0.018600464, 0.045...
http://arxiv.org/abs/2310.11965v1
Filling in the Gaps: Efficient Event Coreference Resolution using Graph Autoencoder Networks
Loic De Langhe, Orphée De Clercq, Veronique Hoste
cs.CL
We introduce a novel and efficient method for Event Coreference Resolution (ECR) applied to a lower-resourced language domain. By framing ECR as a graph reconstruction task, we are able to combine deep semantic embeddings with structural coreference chain knowledge to create a parameter-efficient family of Graph Autoen...
http://arxiv.org/pdf/2310.11965v1
2023-10-18T13:44:58
2023-10-18T13:44:58
{ "title": [ 0.027374268, 0.0033817291, -0.023666382, -0.1048584, 0.02003479, 0.0008983612, -0.056121826, -0.0016794205, 0.005821228, 0.048553467, 0.022964478, 0.007926941, 0.04989624, -0.030776978, -0.022659302, 0.044952393, -0.0037975311, 0.027...
http://arxiv.org/abs/2310.11964v1
AMR Parsing with Causal Hierarchical Attention and Pointers
Chao Lou, Kewei Tu
cs.CL
Translation-based AMR parsers have recently gained popularity due to their simplicity and effectiveness. They predict linearized graphs as free texts, avoiding explicit structure modeling. However, this simplicity neglects structural locality in AMR graphs and introduces unnecessary tokens to represent coreferences. In...
http://arxiv.org/pdf/2310.11964v1
2023-10-18T13:44:26
2023-10-18T13:44:26
{ "title": [ -0.02166748, -0.005367279, -0.027053833, -0.1015625, -0.012329102, 0.028579712, -0.080078125, 0.020401001, -0.032989502, 0.042877197, -0.004688263, 0.00705719, 0.01612854, -0.019042969, -0.031707764, 0.0050468445, -0.021453857, 0.040...
http://arxiv.org/abs/2310.11960v2
Fast Multipole Attention: A Divide-and-Conquer Attention Mechanism for Long Sequences
Yanming Kang, Giang Tran, Hans De Sterck
cs.CL, cs.LG
Transformer-based models have achieved state-of-the-art performance in many areas. However, the quadratic complexity of self-attention with respect to the input length hinders the applicability of Transformer-based models to long sequences. To address this, we present Fast Multipole Attention, a new attention mechanism...
http://arxiv.org/pdf/2310.11960v2
2023-10-21T01:56:32
2023-10-18T13:40:41
{ "title": [ 0.028839111, 0.0012922287, -0.0127334595, -0.06286621, 0.009147644, -0.020065308, -0.0012302399, -0.010040283, 0.042541504, 0.055480957, -0.039794922, 0.0052108765, 0.021728516, 0.0051498413, -0.05404663, 0.016464233, -0.010429382, 0...
http://arxiv.org/abs/2310.11958v1
Emptying the Ocean with a Spoon: Should We Edit Models?
Yuval Pinter, Michael Elhadad
cs.CL, cs.LG
We call into question the recently popularized method of direct model editing as a means of correcting factual errors in LLM generations. We contrast model editing with three similar but distinct approaches that pursue better defined objectives: (1) retrieval-based architectures, which decouple factual memory from infe...
http://arxiv.org/pdf/2310.11958v1
2023-10-18T13:38:03
2023-10-18T13:38:03
{ "title": [ -0.029281616, 0.00026512146, -0.04071045, 0.031829834, -0.0211792, -0.019104004, 0.0062026978, -0.03253174, -0.020355225, 0.0051574707, -0.018157959, -0.026504517, -0.006061554, -0.049865723, -0.016677856, -0.049194336, 0.0061073303, ...
http://arxiv.org/abs/2310.11954v2
MusicAgent: An AI Agent for Music Understanding and Generation with Large Language Models
Dingyao Yu, Kaitao Song, Peiling Lu, Tianyu He, Xu Tan, Wei Ye, Shikun Zhang, Jiang Bian
cs.CL, cs.MM, eess.AS
AI-empowered music processing is a diverse field that encompasses dozens of tasks, ranging from generation tasks (e.g., timbre synthesis) to comprehension tasks (e.g., music classification). For developers and amateurs, it is very difficult to grasp all of these task to satisfy their requirements in music processing, e...
http://arxiv.org/pdf/2310.11954v2
2023-10-25T13:34:13
2023-10-18T13:31:10
{ "title": [ -0.031036377, -0.029388428, -0.039855957, -0.06567383, -0.012924194, 0.00083589554, -0.049682617, 0.020385742, 0.0044517517, 0.035491943, -0.034454346, -0.004875183, 0.03955078, -0.018875122, -0.03314209, 0.0036640167, -0.012168884, ...
http://arxiv.org/abs/2310.11938v1
Grounded and Well-rounded: A Methodological Approach to the Study of Cross-modal and Cross-lingual Grounding
Timothee Mickus, Elaine Zosa, Denis Paperno
cs.CL
Grounding has been argued to be a crucial component towards the development of more complete and truly semantically competent artificial intelligence systems. Literature has divided into two camps: While some argue that grounding allows for qualitatively different generalizations, others believe it can be compensated b...
http://arxiv.org/pdf/2310.11938v1
2023-10-18T13:05:50
2023-10-18T13:05:50
{ "title": [ -6.556511e-7, 0.009925842, -0.037994385, -0.04522705, 0.028182983, -0.020355225, -0.035339355, -0.036834717, 0.010757446, 0.03778076, -0.008293152, 0.01486969, 0.022720337, -0.03427124, -0.03439331, 0.020614624, -0.02128601, 0.086486...
http://arxiv.org/abs/2310.11923v1
Investigating semantic subspaces of Transformer sentence embeddings through linear structural probing
Dmitry Nikolaev, Sebastian Padó
cs.CL
The question of what kinds of linguistic information are encoded in different layers of Transformer-based language models is of considerable interest for the NLP community. Existing work, however, has overwhelmingly focused on word-level representations and encoder-only language models with the masked-token training ob...
http://arxiv.org/pdf/2310.11923v1
2023-10-18T12:32:07
2023-10-18T12:32:07
{ "title": [ -0.010093689, -0.004398346, -0.02130127, -0.028823853, -0.008399963, -0.012825012, -0.014045715, -0.012794495, -0.0055618286, 0.012512207, -0.03643799, 0.027877808, 0.0033779144, -0.01259613, -0.02128601, -0.0011796951, -0.017745972, ...
http://arxiv.org/abs/2310.11917v1
A Benchmark for Semi-Inductive Link Prediction in Knowledge Graphs
Adrian Kochsiek, Rainer Gemulla
cs.CL, cs.AI, cs.LG
Semi-inductive link prediction (LP) in knowledge graphs (KG) is the task of predicting facts for new, previously unseen entities based on context information. Although new entities can be integrated by retraining the model from scratch in principle, such an approach is infeasible for large-scale KGs, where retraining i...
http://arxiv.org/pdf/2310.11917v1
2023-10-18T12:13:13
2023-10-18T12:13:13
{ "title": [ -0.017227173, -0.0066223145, -0.047027588, -0.08227539, 0.0017557144, 0.0036125183, -0.06573486, 0.014320374, -0.0022621155, 0.016235352, -0.0061302185, 0.021011353, -0.010643005, -0.02494812, -0.018463135, -0.0009160042, -0.00017631054,...
http://arxiv.org/abs/2310.11906v1
Rather a Nurse than a Physician -- Contrastive Explanations under Investigation
Oliver Eberle, Ilias Chalkidis, Laura Cabello, Stephanie Brandl
cs.CL
Contrastive explanations, where one decision is explained in contrast to another, are supposed to be closer to how humans explain a decision than non-contrastive explanations, where the decision is not necessarily referenced to an alternative. This claim has never been empirically validated. We analyze four English tex...
http://arxiv.org/pdf/2310.11906v1
2023-10-18T11:54:38
2023-10-18T11:54:38
{ "title": [ 0.027374268, -0.04156494, -0.00042772293, 0.01713562, -0.043701172, -0.02696228, 0.052764893, -0.015914917, -0.08642578, -0.026870728, -0.022338867, 0.059570312, -0.07788086, -0.048706055, 0.09100342, -0.062927246, -0.029968262, 0.04...
http://arxiv.org/abs/2310.11884v1
From Neural Activations to Concepts: A Survey on Explaining Concepts in Neural Networks
Jae Hee Lee, Sergio Lanza, Stefan Wermter
cs.AI, cs.CL, cs.CV, cs.LG, cs.NE
In this paper, we review recent approaches for explaining concepts in neural networks. Concepts can act as a natural link between learning and reasoning: once the concepts are identified that a neural learning system uses, one can integrate those concepts with a reasoning system for inference or use a reasoning system ...
http://arxiv.org/pdf/2310.11884v1
2023-10-18T11:08:02
2023-10-18T11:08:02
{ "title": [ -0.0077552795, 0.00008189678, -0.00026130676, -0.029067993, 0.006187439, -0.0012741089, 0.007888794, -0.027893066, 0.01739502, 0.017623901, -0.0076408386, -0.008163452, -0.018844604, -0.028366089, -0.04144287, 0.005218506, -0.015701294, ...
http://arxiv.org/abs/2310.11878v4
From Dissonance to Insights: Dissecting Disagreements in Rationale Construction for Case Outcome Classification
Shanshan Xu, T. Y. S. S Santosh, Oana Ichim, Isabella Risini, Barbara Plank, Matthias Grabmair
cs.CL
In legal NLP, Case Outcome Classification (COC) must not only be accurate but also trustworthy and explainable. Existing work in explainable COC has been limited to annotations by a single expert. However, it is well-known that lawyers may disagree in their assessment of case facts. We hence collect a novel dataset RAV...
http://arxiv.org/pdf/2310.11878v4
2023-10-24T12:28:13
2023-10-18T11:04:31
{ "title": [ 0.00008279085, -0.016159058, -0.033721924, -0.06585693, -0.0020942688, -0.018173218, -0.024780273, -0.005001068, 0.0017681122, -0.011787415, -0.020553589, 0.010650635, -0.027328491, -0.036987305, -0.0054016113, -0.027832031, -0.020599365...
http://arxiv.org/abs/2310.11877v2
The Curious Case of Hallucinatory (Un)answerability: Finding Truths in the Hidden States of Over-Confident Large Language Models
Aviv Slobodkin, Omer Goldman, Avi Caciularu, Ido Dagan, Shauli Ravfogel
cs.CL
Large language models (LLMs) have been shown to possess impressive capabilities, while also raising crucial concerns about the faithfulness of their responses. A primary issue arising in this context is the management of (un)answerable queries by LLMs, which often results in hallucinatory behavior due to overconfidence...
http://arxiv.org/pdf/2310.11877v2
2023-11-12T13:14:36
2023-10-18T11:01:09
{ "title": [ -0.034301758, -0.0049552917, -0.033813477, -0.04296875, -0.005622864, -0.009780884, -0.033721924, -0.016403198, 0.0045280457, 0.0009832382, -0.025634766, 0.010101318, -0.01927185, -0.020446777, -0.04168701, -0.016113281, -0.0154418945, ...
http://arxiv.org/abs/2310.11870v1
AI Nushu: An Exploration of Language Emergence in Sisterhood -Through the Lens of Computational Linguistics
Yuqian Sun, Yuying Tang, Ze Gao, Zhijun Pan, Chuyan Xu, Yurou Chen, Kejiang Qian, Zhigang Wang, Tristan Braud, Chang Hee Lee, Ali Asadipour
cs.CL, cs.AI, 14J60 (Primary) 14F05, 14J26 (Secondary), F.2.2; I.2.7
This paper presents "AI Nushu," an emerging language system inspired by Nushu (women's scripts), the unique language created and used exclusively by ancient Chinese women who were thought to be illiterate under a patriarchal society. In this interactive installation, two artificial intelligence (AI) agents are trained ...
http://arxiv.org/pdf/2310.11870v1
2023-10-18T10:36:52
2023-10-18T10:36:52
{ "title": [ -0.06921387, 0.0029258728, -0.055114746, 0.0033435822, -0.020065308, -0.029067993, -0.02696228, -0.0027332306, 0.016204834, 0.031311035, -0.054595947, 0.007286072, 0.036102295, -0.020233154, -0.023727417, -0.01953125, -0.0005040169, ...
http://arxiv.org/abs/2310.11780v1
Text Annotation Handbook: A Practical Guide for Machine Learning Projects
Felix Stollenwerk, Joey Öhman, Danila Petrelli, Emma Wallerö, Fredrik Olsson, Camilla Bengtsson, Andreas Horndahl, Gabriela Zarzar Gandler
cs.CL
This handbook is a hands-on guide on how to approach text annotation tasks. It provides a gentle introduction to the topic, an overview of theoretical concepts as well as practical advice. The topics covered are mostly technical, but business, ethical and regulatory issues are also touched upon. The focus lies on reada...
http://arxiv.org/pdf/2310.11780v1
2023-10-18T08:19:53
2023-10-18T08:19:53
{ "title": [ -0.03237915, -0.0118637085, -0.01979065, -0.047332764, 0.010513306, 0.013420105, -0.024856567, -0.015457153, 0.0017995834, 0.036315918, -0.011291504, 0.022232056, -0.0050621033, -0.010124207, -0.053863525, 0.011436462, -0.010482788, ...
http://arxiv.org/abs/2310.11778v3
Language Agents for Detecting Implicit Stereotypes in Text-to-image Models at Scale
Qichao Wang, Tian Bian, Yian Yin, Tingyang Xu, Hong Cheng, Helen M. Meng, Zibin Zheng, Liang Chen, Bingzhe Wu
cs.CY, cs.CL
The recent surge in the research of diffusion models has accelerated the adoption of text-to-image models in various Artificial Intelligence Generated Content (AIGC) commercial products. While these exceptional AIGC products are gaining increasing recognition and sparking enthusiasm among consumers, the questions regar...
http://arxiv.org/pdf/2310.11778v3
2023-11-02T10:46:41
2023-10-18T08:16:29
{ "title": [ 0.0022888184, 0.002626419, -0.0045814514, -0.028030396, -0.008621216, 0.003932953, -0.016952515, -0.026901245, 0.041748047, 0.03994751, -0.05404663, -0.00034594536, 0.017318726, 0.0025539398, -0.04437256, -0.0012760162, -0.02230835, ...
http://arxiv.org/abs/2310.11772v3
Improving Long Document Topic Segmentation Models With Enhanced Coherence Modeling
Hai Yu, Chong Deng, Qinglin Zhang, Jiaqing Liu, Qian Chen, Wen Wang
cs.CL
Topic segmentation is critical for obtaining structured documents and improving downstream tasks such as information retrieval. Due to its ability of automatically exploring clues of topic shift from abundant labeled data, recent supervised neural models have greatly promoted the development of long document topic segm...
http://arxiv.org/pdf/2310.11772v3
2023-10-23T06:05:59
2023-10-18T07:58:47
{ "title": [ -0.023971558, 0.022872925, -0.04046631, -0.02859497, 0.004749298, -0.016693115, -0.0473938, -0.036315918, -0.003921509, 0.036376953, -0.010032654, 0.01689148, 0.0158844, -0.0129852295, -0.013282776, 0.0119018555, -0.020706177, 0.0174...
http://arxiv.org/abs/2310.11769v1
Annotated Job Ads with Named Entity Recognition
Felix Stollenwerk, Niklas Fastlund, Anna Nyqvist, Joey Öhman
cs.CL
We have trained a named entity recognition (NER) model that screens Swedish job ads for different kinds of useful information (e.g. skills required from a job seeker). It was obtained by fine-tuning KB-BERT. The biggest challenge we faced was the creation of a labelled dataset, which required manual annotation. This pa...
http://arxiv.org/pdf/2310.11769v1
2023-10-18T07:55:53
2023-10-18T07:55:53
{ "title": [ 0.03704834, 0.0006351471, -0.02607727, -0.05078125, 0.005214691, -0.03475952, 0.0011882782, -0.0028591156, -0.026977539, 0.047027588, -0.02444458, 0.020080566, 0.011459351, -0.019226074, -0.02268982, -0.004501343, 0.012145996, 0.0233...
http://arxiv.org/abs/2310.11761v1
A Comprehensive Evaluation of Large Language Models on Legal Judgment Prediction
Ruihao Shui, Yixin Cao, Xiang Wang, Tat-Seng Chua
cs.CL
Large language models (LLMs) have demonstrated great potential for domain-specific applications, such as the law domain. However, recent disputes over GPT-4's law evaluation raise questions concerning their performance in real-world legal tasks. To systematically investigate their competency in the law, we design pract...
http://arxiv.org/pdf/2310.11761v1
2023-10-18T07:38:04
2023-10-18T07:38:04
{ "title": [ -0.032714844, -0.019866943, -0.04498291, -0.06512451, -0.011558533, 0.0037021637, -0.05859375, 0.018234253, 0.02835083, 0.031799316, -0.03302002, -0.014373779, 0.024017334, -0.0049362183, -0.061309814, 0.01096344, -0.02357483, 0.0551...
http://arxiv.org/abs/2310.11753v1
Bias in Emotion Recognition with ChatGPT
Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi, Jun Takamatsu, Katsushi Ikeuchi
cs.RO, cs.CL
This technical report explores the ability of ChatGPT in recognizing emotions from text, which can be the basis of various applications like interactive chatbots, data annotation, and mental health analysis. While prior research has shown ChatGPT's basic ability in sentiment analysis, its performance in more nuanced em...
http://arxiv.org/pdf/2310.11753v1
2023-10-18T07:28:12
2023-10-18T07:28:12
{ "title": [ -0.011047363, 0.0018777847, -0.03555298, -0.009284973, -0.004508972, -0.020202637, 0.0011920929, -0.016220093, 0.00497818, 0.028808594, -0.038848877, -0.0035514832, 0.019699097, -0.030441284, -0.022979736, 0.00078725815, -0.011444092, ...
http://arxiv.org/abs/2310.11732v1
Investigating Uncertainty Calibration of Aligned Language Models under the Multiple-Choice Setting
Guande He, Peng Cui, Jianfei Chen, Wenbo Hu, Jun Zhu
cs.LG, cs.CL
Despite the significant progress made in practical applications of aligned language models (LMs), they tend to be overconfident in output answers compared to the corresponding pre-trained LMs. In this work, we systematically evaluate the impact of the alignment process on logit-based uncertainty calibration of LMs unde...
http://arxiv.org/pdf/2310.11732v1
2023-10-18T06:07:28
2023-10-18T06:07:28
{ "title": [ -0.080444336, -0.0027599335, -0.056610107, 0.005531311, -0.030059814, -0.006843567, -0.058654785, -0.011299133, 0.0103302, 0.020599365, -0.016357422, 0.010688782, -0.008117676, -0.026550293, -0.048431396, 0.0025310516, -0.017333984, ...
http://arxiv.org/abs/2310.11722v1
Quantify Health-Related Atomic Knowledge in Chinese Medical Large Language Models: A Computational Analysis
Yaxin Fan, Feng Jiang, Peifeng Li, Haizhou Li
cs.CL, cs.AI
Large Language Models (LLMs) have the potential to revolutionize the way users self-diagnose through search engines by offering direct and efficient suggestions. Recent studies primarily focused on the quality of LLMs evaluated by GPT-4 or their ability to pass medical exams, no studies have quantified the extent of he...
http://arxiv.org/pdf/2310.11722v1
2023-10-18T05:42:22
2023-10-18T05:42:22
{ "title": [ -0.042541504, -0.008712769, -0.05255127, -0.072021484, -0.0017433167, -0.0019817352, -0.06262207, 0.011749268, -0.009239197, 0.046051025, 0.015686035, -0.0007491112, 0.018325806, -0.046173096, -0.025222778, 0.004348755, 0.015914917, ...
http://arxiv.org/abs/2310.11721v1
Chain-of-Thought Tuning: Masked Language Models can also Think Step By Step in Natural Language Understanding
Caoyun Fan, Jidong Tian, Yitian Li, Wenqing Chen, Hao He, Yaohui Jin
cs.CL, cs.LG
Chain-of-Thought (CoT) is a technique that guides Large Language Models (LLMs) to decompose complex tasks into multi-step reasoning through intermediate steps in natural language form. Briefly, CoT enables LLMs to think step by step. However, although many Natural Language Understanding (NLU) tasks also require thinkin...
http://arxiv.org/pdf/2310.11721v1
2023-10-18T05:39:20
2023-10-18T05:39:20
{ "title": [ -0.040893555, -0.012397766, -0.027252197, -0.021896362, -0.056274414, 0.007247925, -0.07183838, 0.0024585724, -0.025634766, 0.0075302124, -0.020767212, -0.0016050339, 0.016113281, -0.054382324, 0.028259277, -0.024124146, 0.0098724365, ...
http://arxiv.org/abs/2310.11716v1
Reflection-Tuning: Data Recycling Improves LLM Instruction-Tuning
Ming Li, Lichang Chen, Jiuhai Chen, Shwai He, Heng Huang, Jiuxiang Gu, Tianyi Zhou
cs.CL
Recent advancements in Large Language Models (LLMs) have expanded the horizons of natural language understanding and generation. Notably, the output control and alignment with the input of LLMs can be refined through instruction tuning. However, as highlighted in several studies, low-quality data in the training set ar...
http://arxiv.org/pdf/2310.11716v1
2023-10-18T05:13:47
2023-10-18T05:13:47
{ "title": [ -0.06768799, -0.01033783, -0.04269409, -0.046447754, -0.043548584, 0.008140564, -0.08428955, 0.014083862, -0.016418457, 0.032318115, -0.04638672, 0.02607727, 0.021362305, -0.016555786, -0.0016956329, -0.00868988, 0.011428833, 0.00609...
http://arxiv.org/abs/2310.11715v2
Enhancing Low-resource Fine-grained Named Entity Recognition by Leveraging Coarse-grained Datasets
Su Ah Lee, Seokjin Oh, Woohwan Jung
cs.CL, cs.AI
Named Entity Recognition (NER) frequently suffers from the problem of insufficient labeled data, particularly in fine-grained NER scenarios. Although $K$-shot learning techniques can be applied, their performance tends to saturate when the number of annotations exceeds several tens of labels. To overcome this problem, ...
http://arxiv.org/pdf/2310.11715v2
2023-11-13T13:18:58
2023-10-18T05:13:34
{ "title": [ 0.0034275055, 0.0007367134, -0.059692383, -0.08105469, 0.03668213, -0.0035991669, -0.020996094, 0.02279663, 0.001036644, 0.035339355, -0.045715332, 0.012023926, 0.0032348633, 0.007095337, -0.05029297, -0.016342163, -0.004020691, 0.04...
http://arxiv.org/abs/2310.11710v2
Learning Co-Speech Gesture for Multimodal Aphasia Type Detection
Daeun Lee, Sejung Son, Hyolim Jeon, Seungbae Kim, Jinyoung Han
cs.CL, cs.AI
Aphasia, a language disorder resulting from brain damage, requires accurate identification of specific aphasia types, such as Broca's and Wernicke's aphasia, for effective treatment. However, little attention has been paid to developing methods to detect different types of aphasia. Recognizing the importance of analyzi...
http://arxiv.org/pdf/2310.11710v2
2023-10-20T05:43:28
2023-10-18T04:54:32
{ "title": [ -0.007713318, 0.012672424, -0.03753662, -0.037017822, 0.00818634, -0.008232117, -0.0256958, -0.016204834, -0.036346436, 0.038330078, -0.018356323, 0.0073890686, -0.0013723373, -0.030700684, -0.03579712, -0.022140503, -0.015625, 0.043...
http://arxiv.org/abs/2310.11699v1
MISAR: A Multimodal Instructional System with Augmented Reality
Jing Bi, Nguyen Manh Nguyen, Ali Vosoughi, Chenliang Xu
cs.CL, cs.CV
Augmented reality (AR) requires the seamless integration of visual, auditory, and linguistic channels for optimized human-computer interaction. While auditory and visual inputs facilitate real-time and contextual user guidance, the potential of large language models (LLMs) in this landscape remains largely untapped. Ou...
http://arxiv.org/pdf/2310.11699v1
2023-10-18T04:15:12
2023-10-18T04:15:12
{ "title": [ -0.03668213, -0.002494812, -0.056121826, -0.011047363, -0.0042915344, -0.018707275, -0.016601562, -0.026657104, -0.0236969, 0.019561768, -0.006965637, 0.0047187805, -0.017242432, -0.029586792, -0.024291992, -0.017181396, -0.01914978, ...
http://arxiv.org/abs/2310.11689v2
Adaptation with Self-Evaluation to Improve Selective Prediction in LLMs
Jiefeng Chen, Jinsung Yoon, Sayna Ebrahimi, Sercan O Arik, Tomas Pfister, Somesh Jha
cs.CL, cs.LG
Large language models (LLMs) have recently shown great advances in a variety of tasks, including natural language understanding and generation. However, their use in high-stakes decision-making scenarios is still limited due to the potential for errors. Selective prediction is a technique that can be used to improve th...
http://arxiv.org/pdf/2310.11689v2
2023-11-11T19:29:42
2023-10-18T03:34:59
{ "title": [ -0.034088135, -0.00045347214, -0.032165527, -0.047973633, -0.0256958, 0.0060577393, -0.04550171, -0.0075149536, 0.00087690353, 0.017288208, -0.021850586, 0.006412506, -0.033966064, -0.025787354, -0.043029785, -0.009422302, -0.007255554, ...
http://arxiv.org/abs/2310.11685v1
Superiority of Softmax: Unveiling the Performance Edge Over Linear Attention
Yichuan Deng, Zhao Song, Tianyi Zhou
cs.CL, cs.LG
Large transformer models have achieved state-of-the-art results in numerous natural language processing tasks. Among the pivotal components of the transformer architecture, the attention mechanism plays a crucial role in capturing token interactions within sequences through the utilization of softmax function. Conver...
http://arxiv.org/pdf/2310.11685v1
2023-10-18T03:17:57
2023-10-18T03:17:57
{ "title": [ -0.0044898987, -0.0006785393, -0.03353882, -0.054534912, 0.024658203, -0.013801575, -0.009719849, -0.016601562, -0.018951416, 0.030822754, -0.008911133, 0.009132385, -0.014038086, -0.023986816, -0.0418396, 0.0059890747, -0.004825592, ...
http://arxiv.org/abs/2310.11681v1
Descriptive Knowledge Graph in Biomedical Domain
Kerui Zhu, Jie Huang, Kevin Chen-Chuan Chang
cs.CL, cs.AI
We present a novel system that automatically extracts and generates informative and descriptive sentences from the biomedical corpus and facilitates the efficient search for relational knowledge. Unlike previous search engines or exploration systems that retrieve unconnected passages, our system organizes descriptive s...
http://arxiv.org/pdf/2310.11681v1
2023-10-18T03:10:25
2023-10-18T03:10:25
{ "title": [ -0.030349731, -0.010604858, -0.069885254, -0.085510254, -0.0067367554, -0.0004658699, -0.06427002, 0.017593384, -0.012145996, 0.027267456, 0.0077438354, -0.024017334, -0.010421753, -0.04147339, -0.045959473, -0.00042676926, 0.01626587, ...
http://arxiv.org/abs/2310.11672v2
Open-ended Commonsense Reasoning with Unrestricted Answer Scope
Chen Ling, Xuchao Zhang, Xujiang Zhao, Yanchi Liu, Wei Cheng, Mika Oishi, Takao Osaki, Katsushi Matsuda, Haifeng Chen, Liang Zhao
cs.CL
Open-ended Commonsense Reasoning is defined as solving a commonsense question without providing 1) a short list of answer candidates and 2) a pre-defined answer scope. Conventional ways of formulating the commonsense question into a question-answering form or utilizing external knowledge to learn retrieval-based method...
http://arxiv.org/pdf/2310.11672v2
2023-10-27T13:50:09
2023-10-18T02:45:54
{ "title": [ -0.0017690659, -0.013694763, -0.04562378, -0.0847168, 0.014183044, -0.00466156, -0.027679443, 0.0049819946, 0.013618469, 0.012748718, -0.018341064, 0.0038318634, -0.031799316, -0.023880005, -0.022903442, 0.006111145, 0.007434845, 0.0...
http://arxiv.org/abs/2310.11671v1
MixEdit: Revisiting Data Augmentation and Beyond for Grammatical Error Correction
Jingheng Ye, Yinghui Li, Yangning Li, Hai-Tao Zheng
cs.CL
Data Augmentation through generating pseudo data has been proven effective in mitigating the challenge of data scarcity in the field of Grammatical Error Correction (GEC). Various augmentation strategies have been widely explored, most of which are motivated by two heuristics, i.e., increasing the distribution similari...
http://arxiv.org/pdf/2310.11671v1
2023-10-18T02:45:51
2023-10-18T02:45:51
{ "title": [ 0.0017690659, -0.010673523, -0.03363037, -0.06201172, 0.0054359436, 0.000117242336, -0.02432251, 0.0013513565, -0.005748749, 0.028961182, -0.02809143, 0.026626587, 0.0062713623, -0.0052337646, -0.041870117, 0.0048294067, -0.035064697, ...
http://arxiv.org/abs/2310.11670v3
Prototype-based HyperAdapter for Sample-Efficient Multi-task Tuning
Hao Zhao, Jie Fu, Zhaofeng He
cs.CL, cs.AI
Parameter-efficient fine-tuning (PEFT) has shown its effectiveness in adapting the pre-trained language models to downstream tasks while only updating a small number of parameters. Despite the success, most existing methods independently adapt to each task without considering knowledge transfer between tasks and are li...
http://arxiv.org/pdf/2310.11670v3
2023-11-11T15:30:30
2023-10-18T02:42:17
{ "title": [ -0.0052452087, -0.0070152283, -0.02708435, -0.09741211, 0.006893158, 0.009880066, -0.06640625, 0.0084991455, -0.0050086975, 0.04244995, -0.024520874, 0.012489319, -0.017333984, -0.0025081635, -0.049224854, -0.0126953125, 0.010643005, ...
http://arxiv.org/abs/2310.11667v1
SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents
Xuhui Zhou, Hao Zhu, Leena Mathur, Ruohong Zhang, Haofei Yu, Zhengyang Qi, Louis-Philippe Morency, Yonatan Bisk, Daniel Fried, Graham Neubig, Maarten Sap
cs.AI, cs.CL, cs.LG
Humans are social beings; we pursue social goals in our daily interactions, which is a crucial aspect of social intelligence. Yet, AI systems' abilities in this realm remain elusive. We present SOTOPIA, an open-ended environment to simulate complex social interactions between artificial agents and evaluate their social...
http://arxiv.org/pdf/2310.11667v1
2023-10-18T02:27:01
2023-10-18T02:27:01
{ "title": [ -0.0259552, -0.01360321, -0.011795044, -0.035186768, -0.00029158592, 0.0012340546, -0.015541077, -0.003932953, 0.018692017, 0.022323608, -0.0519104, 0.011222839, -0.008224487, 0.00036120415, -0.06274414, -0.009941101, -0.034820557, 0...
http://arxiv.org/abs/2310.11655v1
Field-testing items using artificial intelligence: Natural language processing with transformers
Hotaka Maeda
cs.CL
Five thousand variations of the RoBERTa model, an artificially intelligent "transformer" that can understand text language, completed an English literacy exam with 29 multiple-choice questions. Data were used to calculate the psychometric properties of the items, which showed some degree of agreement to those obtained ...
http://arxiv.org/pdf/2310.11655v1
2023-10-18T01:56:16
2023-10-18T01:56:16
{ "title": [ -0.020126343, -0.0008354187, -0.027191162, -0.041229248, -0.008026123, -0.000647068, -0.036621094, -0.021316528, -0.007896423, 0.021972656, -0.07324219, -0.003665924, -0.0065193176, -0.014381409, -0.034088135, -0.045074463, 0.0029525757,...
http://arxiv.org/abs/2310.11648v1
Zero-shot Faithfulness Evaluation for Text Summarization with Foundation Language Model
Qi Jia, Siyu Ren, Yizhu Liu, Kenny Q. Zhu
cs.CL
Despite tremendous improvements in natural language generation, summarization models still suffer from the unfaithfulness issue. Previous work evaluates faithfulness either using models trained on the other tasks or in-domain synthetic data, or prompting a large model such as ChatGPT. This paper proposes to do zero-sho...
http://arxiv.org/pdf/2310.11648v1
2023-10-18T01:20:16
2023-10-18T01:20:16
{ "title": [ -0.015792847, -0.0008831024, -0.021835327, -0.06994629, -0.036315918, 0.012527466, -0.076049805, -0.0035114288, 0.014839172, 0.062683105, 0.0018062592, 0.026794434, 0.01625061, -0.0037994385, -0.06726074, 0.018508911, -0.012741089, 0...
http://arxiv.org/abs/2310.11638v3
Systematic Assessment of Factual Knowledge in Large Language Models
Linhao Luo, Thuy-Trang Vu, Dinh Phung, Gholamreza Haffari
cs.CL
Previous studies have relied on existing question-answering benchmarks to evaluate the knowledge stored in large language models (LLMs). However, this approach has limitations regarding factual knowledge coverage, as it mostly focuses on generic domains which may overlap with the pretraining data. This paper proposes a...
http://arxiv.org/pdf/2310.11638v3
2023-10-30T12:05:31
2023-10-18T00:20:50
{ "title": [ -0.032073975, -0.011146545, -0.050994873, -0.07501221, -0.0048599243, -0.011528015, -0.05609131, 0.0042304993, 0.023361206, 0.031707764, -0.017333984, -0.0075531006, 0.011833191, -0.016571045, -0.054138184, 0.017807007, -0.011169434, ...
http://arxiv.org/abs/2310.11634v1
MAGNIFICo: Evaluating the In-Context Learning Ability of Large Language Models to Generalize to Novel Interpretations
Arkil Patel, Satwik Bhattamishra, Siva Reddy, Dzmitry Bahdanau
cs.CL
Humans possess a remarkable ability to assign novel interpretations to linguistic expressions, enabling them to learn new words and understand community-specific connotations. However, Large Language Models (LLMs) have a knowledge cutoff and are costly to finetune repeatedly. Therefore, it is crucial for LLMs to learn ...
http://arxiv.org/pdf/2310.11634v1
2023-10-18T00:02:38
2023-10-18T00:02:38
{ "title": [ -0.022155762, -0.011108398, -0.04083252, -0.05621338, -0.0075798035, -0.0067825317, -0.051452637, -0.001376152, -0.0043754578, 0.041168213, -0.019866943, 0.009796143, 0.005718231, -0.010749817, -0.047973633, 0.0055351257, 0.006752014, ...
http://arxiv.org/abs/2310.11628v1
Learn Your Tokens: Word-Pooled Tokenization for Language Modeling
Avijit Thawani, Saurabh Ghanekar, Xiaoyuan Zhu, Jay Pujara
cs.CL, cs.AI
Language models typically tokenize text into subwords, using a deterministic, hand-engineered heuristic of combining characters into longer surface-level strings such as 'ing' or whole words. Recent literature has repeatedly shown the limitations of such a tokenization strategy, particularly for documents not written i...
http://arxiv.org/pdf/2310.11628v1
2023-10-17T23:34:39
2023-10-17T23:34:39
{ "title": [ -0.03652954, -0.011100769, -0.05380249, -0.0625, 0.011047363, 0.010887146, -0.048736572, 0.0020427704, -0.0033016205, 0.030838013, -0.02720642, 0.024841309, 0.011627197, -0.0074157715, -0.044799805, 0.013519287, -0.01826477, 0.040313...
http://arxiv.org/abs/2310.11616v1
Unveiling the General Intelligence Factor in Language Models: A Psychometric Approach
David Ilić
cs.CL, cs.AI
This study uncovers the factor of general intelligence, or g, in language models, extending the psychometric theory traditionally applied to humans and certain animal species. Utilizing factor analysis on two extensive datasets - Open LLM Leaderboard with 1,232 models and General Language Understanding Evaluation (GLUE...
http://arxiv.org/pdf/2310.11616v1
2023-10-17T22:42:12
2023-10-17T22:42:12
{ "title": [ -0.041534424, -0.003736496, -0.037872314, -0.030349731, -0.008865356, -0.00806427, -0.03970337, -0.010925293, -0.0056991577, 0.036743164, -0.0047912598, -0.017852783, 0.0020694733, -0.0236969, -0.076171875, 0.0025844574, -0.032714844, ...
http://arxiv.org/abs/2310.11604v1
Language Models as Zero-Shot Trajectory Generators
Teyun Kwon, Norman Di Palo, Edward Johns
cs.RO, cs.AI, cs.CL, cs.HC, cs.LG
Large Language Models (LLMs) have recently shown promise as high-level planners for robots when given access to a selection of low-level skills. However, it is often assumed that LLMs do not possess sufficient knowledge to be used for the low-level trajectories themselves. In this work, we address this assumption thoro...
http://arxiv.org/pdf/2310.11604v1
2023-10-17T21:57:36
2023-10-17T21:57:36
{ "title": [ -0.0146484375, -0.021392822, -0.01789856, -0.06707764, -0.022155762, 0.0064201355, -0.04776001, 0.012840271, 0.002029419, 0.0362854, 0.008270264, 0.0050315857, -0.0046653748, -0.017547607, -0.04373169, 0.02619934, 0.002500534, 0.0314...
http://arxiv.org/abs/2310.11593v1
Automated Evaluation of Personalized Text Generation using Large Language Models
Yaqing Wang, Jiepu Jiang, Mingyang Zhang, Cheng Li, Yi Liang, Qiaozhu Mei, Michael Bendersky
cs.CL, cs.AI, cs.LG
Personalized text generation presents a specialized mechanism for delivering content that is specific to a user's personal context. While the research progress in this area has been rapid, evaluation still presents a challenge. Traditional automated metrics such as BLEU and ROUGE primarily measure lexical similarity to...
http://arxiv.org/pdf/2310.11593v1
2023-10-17T21:35:06
2023-10-17T21:35:06
{ "title": [ -0.036071777, -0.0066871643, -0.0362854, -0.052001953, -0.011512756, 0.017410278, -0.06237793, -0.002117157, -0.01109314, 0.04788208, -0.0017547607, 0.015945435, 0.013191223, -0.00856781, -0.045776367, 0.007408142, -0.023773193, 0.05...
http://arxiv.org/abs/2310.13015v1
Audio-AdapterFusion: A Task-ID-free Approach for Efficient and Non-Destructive Multi-task Speech Recognition
Hillary Ngai, Rohan Agrawal, Neeraj Gaur, Ronny Huang, Parisa Haghani, Pedro Moreno Mengibar
cs.CL, cs.AI, eess.AS
Adapters are an efficient, composable alternative to full fine-tuning of pre-trained models and help scale the deployment of large ASR models to many tasks. In practice, a task ID is commonly prepended to the input during inference to route to single-task adapters for the specified task. However, one major limitation o...
http://arxiv.org/pdf/2310.13015v1
2023-10-17T21:21:40
2023-10-17T21:21:40
{ "title": [ -0.000091671944, 0.02130127, -0.08288574, -0.093933105, 0.030914307, -0.018417358, -0.05871582, 0.027297974, -0.020385742, 0.053375244, -0.021881104, -0.0013942719, 0.0039520264, -0.007633209, -0.06262207, -0.00969696, -0.012168884, ...
http://arxiv.org/abs/2310.11589v1
Eliciting Human Preferences with Language Models
Belinda Z. Li, Alex Tamkin, Noah Goodman, Jacob Andreas
cs.CL, cs.AI, cs.LG
Language models (LMs) can be directed to perform target tasks by using labeled examples or natural language prompts. But selecting examples or writing prompts for can be challenging--especially in tasks that involve unusual edge cases, demand precise articulation of nebulous preferences, or require an accurate mental m...
http://arxiv.org/pdf/2310.11589v1
2023-10-17T21:11:21
2023-10-17T21:11:21
{ "title": [ -0.01687622, -0.000688076, -0.022277832, -0.031463623, 0.004585266, -0.0046806335, -0.032043457, -0.02217102, 0.0033626556, 0.019439697, -0.031555176, -0.011474609, -0.012069702, -0.011413574, -0.046020508, -0.006706238, -0.022567749, ...
http://arxiv.org/abs/2310.11584v1
BasahaCorpus: An Expanded Linguistic Resource for Readability Assessment in Central Philippine Languages
Joseph Marvin Imperial, Ekaterina Kochmar
cs.CL
Current research on automatic readability assessment (ARA) has focused on improving the performance of models in high-resource languages such as English. In this work, we introduce and release BasahaCorpus as part of an initiative aimed at expanding available corpora and baseline models for readability assessment in lo...
http://arxiv.org/pdf/2310.11584v1
2023-10-17T21:05:20
2023-10-17T21:05:20
{ "title": [ -0.030227661, -0.011108398, -0.0036449432, -0.019088745, -0.025344849, 0.020477295, -0.057250977, -0.02357483, 0.013298035, 0.045043945, -0.061279297, 0.011634827, -0.009056091, 0.014404297, -0.085754395, -0.021575928, -0.026260376, ...
http://arxiv.org/abs/2310.11571v1
What is a good question? Task-oriented asking with fact-level masking
Matthew Toles, Yukun Huang, Zhou Yu, Luis Gravano
cs.CL, cs.LG
Asking questions is an important element of real-life collaboration on reasoning tasks like question answering. For example, a legal assistant chatbot may be unable to make accurate recommendations without specific information on the user's circumstances. However, large language models are usually deployed to solve rea...
http://arxiv.org/pdf/2310.11571v1
2023-10-17T20:40:59
2023-10-17T20:40:59
{ "title": [ -0.005344391, 0.0007123947, -0.05947876, -0.04864502, 0.00724411, -0.025268555, -0.0046844482, -0.0010051727, -0.011260986, 0.01687622, -0.011856079, 0.009010315, -0.021224976, -0.02407837, -0.032836914, -0.005466461, 0.005683899, 0....
http://arxiv.org/abs/2310.11564v1
Personalized Soups: Personalized Large Language Model Alignment via Post-hoc Parameter Merging
Joel Jang, Seungone Kim, Bill Yuchen Lin, Yizhong Wang, Jack Hessel, Luke Zettlemoyer, Hannaneh Hajishirzi, Yejin Choi, Prithviraj Ammanabrolu
cs.CL
While Reinforcement Learning from Human Feedback (RLHF) aligns Large Language Models (LLMs) with general, aggregate human preferences, it is suboptimal for learning diverse, individual perspectives. In this work, we study Reinforcement Learning from Personalized Human Feedback (RLPHF) problem, wherein LLMs are aligned ...
http://arxiv.org/pdf/2310.11564v1
2023-10-17T20:22:13
2023-10-17T20:22:13
{ "title": [ -0.012489319, -0.0043792725, -0.04675293, -0.0947876, 0.004940033, 0.021408081, -0.083984375, 0.0099487305, -0.013053894, 0.040863037, 0.0076446533, 0.005634308, 0.02178955, -0.02720642, -0.025924683, 0.011833191, -0.013893127, 0.033...
http://arxiv.org/abs/2310.11541v1
MUST&P-SRL: Multi-lingual and Unified Syllabification in Text and Phonetic Domains for Speech Representation Learning
Noé Tits
cs.CL, cs.AI, cs.LG, eess.AS
In this paper, we present a methodology for linguistic feature extraction, focusing particularly on automatically syllabifying words in multiple languages, with a design to be compatible with a forced-alignment tool, the Montreal Forced Aligner (MFA). In both the textual and phonetic domains, our method focuses on the ...
http://arxiv.org/pdf/2310.11541v1
2023-10-17T19:27:23
2023-10-17T19:27:23
{ "title": [ -0.012802124, 0.007637024, -0.02973938, -0.05581665, 0.001206398, 0.01335907, -0.07006836, 0.0041503906, -0.045562744, 0.04058838, -0.0047912598, 0.018798828, 0.011550903, -0.029052734, -0.010261536, -0.0146484375, -0.011436462, 0.04...
http://arxiv.org/abs/2310.11532v1
Multi-stage Large Language Model Correction for Speech Recognition
Jie Pu, Thai-Son Nguyen, Sebastian Stüker
cs.CL, eess.AS
In this paper, we investigate the usage of large language models (LLMs) to improve the performance of competitive speech recognition systems. Different from traditional language models that focus on one single data domain, the rise of LLMs brings us the opportunity to push the limit of state-of-the-art ASR performance,...
http://arxiv.org/pdf/2310.11532v1
2023-10-17T19:02:40
2023-10-17T19:02:40
{ "title": [ -0.021057129, -0.007751465, -0.04034424, -0.060668945, 0.008613586, 0.0040512085, -0.057373047, 0.017425537, -0.0061569214, 0.036865234, -0.022979736, -0.017318726, 0.002538681, -0.018493652, -0.04711914, -0.0016889572, -0.018447876, ...
http://arxiv.org/abs/2310.11523v1
Group Preference Optimization: Few-Shot Alignment of Large Language Models
Siyan Zhao, John Dang, Aditya Grover
cs.LG, cs.AI, cs.CL
Many applications of large language models (LLMs), ranging from chatbots to creative writing, require nuanced subjective judgments that can differ significantly across different groups. Existing alignment algorithms can be expensive to align for each group, requiring prohibitive amounts of group-specific preference dat...
http://arxiv.org/pdf/2310.11523v1
2023-10-17T18:41:57
2023-10-17T18:41:57
{ "title": [ -0.038208008, -0.016677856, -0.056915283, -0.05908203, -0.0036907196, -0.0029354095, -0.06500244, 0.029312134, -0.0066871643, 0.041809082, -0.008804321, 0.0001872778, 0.01576233, -0.01499939, -0.040405273, 0.002796173, -0.0069465637, ...
http://arxiv.org/abs/2310.11520v1
Automatic News Summerization
Kavach Dheer, Arpit Dhankhar
cs.CL, cs.LG
Natural Language Processing is booming with its applications in the real world, one of which is Text Summarization for large texts including news articles. This research paper provides an extensive comparative evaluation of extractive and abstractive approaches for news text summarization, with an emphasis on the ROUGE...
http://arxiv.org/pdf/2310.11520v1
2023-10-17T18:38:03
2023-10-17T18:38:03
{ "title": [ 0.005519867, 0.0079574585, -0.011940002, -0.038391113, -0.049072266, -0.0064468384, -0.02909851, -0.015029907, -0.008491516, 0.093322754, -0.02029419, 0.0045661926, 0.037719727, -0.01247406, -0.023529053, 0.021865845, 0.022994995, -0...
http://arxiv.org/abs/2310.11511v1
Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection
Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, Hannaneh Hajishirzi
cs.CL, cs.AI, cs.LG
Despite their remarkable capabilities, large language models (LLMs) often produce responses containing factual inaccuracies due to their sole reliance on the parametric knowledge they encapsulate. Retrieval-Augmented Generation (RAG), an ad hoc approach that augments LMs with retrieval of relevant knowledge, decreases ...
http://arxiv.org/pdf/2310.11511v1
2023-10-17T18:18:32
2023-10-17T18:18:32
{ "title": [ -0.01789856, -0.02835083, -0.03390503, -0.031677246, -0.013946533, -0.0056114197, -0.013923645, -0.008232117, -0.0317688, 0.01473999, 0.025131226, -0.0029468536, -0.028961182, -0.041931152, -0.03878784, -0.016098022, 0.007949829, 0.0...
http://arxiv.org/abs/2310.11501v1
CoMPosT: Characterizing and Evaluating Caricature in LLM Simulations
Myra Cheng, Tiziano Piccardi, Diyi Yang
cs.CL, cs.AI, cs.CY
Recent work has aimed to capture nuances of human behavior by using LLMs to simulate responses from particular demographics in settings like social science experiments and public opinion surveys. However, there are currently no established ways to discuss or evaluate the quality of such LLM simulations. Moreover, there...
http://arxiv.org/pdf/2310.11501v1
2023-10-17T18:00:25
2023-10-17T18:00:25
{ "title": [ -0.009811401, -0.012664795, 0.011894226, -0.028030396, -0.045043945, 0.013084412, -0.05041504, -0.027954102, -0.008781433, 0.03012085, -0.0010986328, 0.003921509, 0.0079193115, -0.044891357, -0.0036792755, -0.0017662048, -0.03363037, ...
http://arxiv.org/abs/2310.11454v1
VeRA: Vector-based Random Matrix Adaptation
Dawid Jan Kopiczko, Tijmen Blankevoort, Yuki Markus Asano
cs.CL
Low-rank adapation (LoRA) is a popular method that reduces the number of trainable parameters when finetuning large language models, but still faces acute storage challenges when scaling to even larger models or deploying numerous per-user or per-task adapted models. In this work, we present Vector-based Random Matrix ...
http://arxiv.org/pdf/2310.11454v1
2023-10-17T17:59:46
2023-10-17T17:59:46
{ "title": [ -0.046661377, -0.017974854, -0.022918701, -0.01902771, -0.010261536, 0.009056091, -0.011199951, -0.031341553, 0.009857178, 0.026290894, 0.0053482056, -0.000095427036, -0.0037956238, -0.03164673, -0.03894043, 0.02128601, 0.021881104, ...
http://arxiv.org/abs/2310.11453v1
BitNet: Scaling 1-bit Transformers for Large Language Models
Hongyu Wang, Shuming Ma, Li Dong, Shaohan Huang, Huaijie Wang, Lingxiao Ma, Fan Yang, Ruiping Wang, Yi Wu, Furu Wei
cs.CL
The increasing size of large language models has posed challenges for deployment and raised concerns about environmental impact due to high energy consumption. In this work, we introduce BitNet, a scalable and stable 1-bit Transformer architecture designed for large language models. Specifically, we introduce BitLinear...
http://arxiv.org/pdf/2310.11453v1
2023-10-17T17:59:15
2023-10-17T17:59:15
{ "title": [ -0.02468872, -0.029769897, -0.046020508, -0.087646484, -0.0024909973, 0.013702393, -0.064086914, 0.036254883, 0.014961243, 0.03414917, -0.030792236, -0.007686615, 0.028640747, -0.0112838745, -0.05215454, 0.014778137, -0.00907135, 0.0...
http://arxiv.org/abs/2310.11451v1
Seeking Neural Nuggets: Knowledge Transfer in Large Language Models from a Parametric Perspective
Ming Zhong, Chenxin An, Weizhu Chen, Jiawei Han, Pengcheng He
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) inherently encode a wealth of knowledge within their parameters through pre-training on extensive corpora. While prior research has delved into operations on these parameters to manipulate the underlying implicit knowledge (encompassing detection, editing, and merging), there remains an amb...
http://arxiv.org/pdf/2310.11451v1
2023-10-17T17:58:34
2023-10-17T17:58:34
{ "title": [ -0.02722168, -0.017837524, -0.0362854, -0.053833008, -0.015945435, -0.010398865, -0.04812622, 0.0008068085, -0.010528564, 0.019805908, -0.023025513, -0.003282547, 0.008598328, -0.016555786, -0.03463745, -0.008834839, 0.006263733, 0.0...
http://arxiv.org/abs/2310.13014v1
Large Language Model Prediction Capabilities: Evidence from a Real-World Forecasting Tournament
Philipp Schoenegger, Peter S. Park
cs.CY, cs.AI, cs.CL, cs.LG
Accurately predicting the future would be an important milestone in the capabilities of artificial intelligence. However, research on the ability of large language models to provide probabilistic predictions about future events remains nascent. To empirically test this ability, we enrolled OpenAI's state-of-the-art lar...
http://arxiv.org/pdf/2310.13014v1
2023-10-17T17:58:17
2023-10-17T17:58:17
{ "title": [ -0.003452301, -0.009559631, -0.033294678, -0.08123779, -0.009841919, -0.013656616, -0.06384277, 0.0121154785, 0.004688263, 0.035308838, 0.000746727, -0.014389038, 0.005794525, -0.017654419, -0.060821533, 0.003080368, -0.0035133362, 0...
http://arxiv.org/abs/2310.11446v1
Functional Invariants to Watermark Large Transformers
Fernandez Pierre, Couairon Guillaume, Furon Teddy, Douze Matthijs
cs.CR, cs.AI, cs.CL
The rapid growth of transformer-based models increases the concerns about their integrity and ownership insurance. Watermarking addresses this issue by embedding a unique identifier into the model, while preserving its performance. However, most existing approaches require to optimize the weights to imprint the waterma...
http://arxiv.org/pdf/2310.11446v1
2023-10-17T17:56:18
2023-10-17T17:56:18
{ "title": [ 0.020339966, 0.048980713, -0.03591919, -0.03781128, 0.038391113, -0.013832092, -0.013084412, -0.045928955, -0.032958984, 0.04473877, -0.0006508827, 0.013870239, 0.0033740997, -0.031433105, -0.017700195, 0.0073242188, 0.0119018555, 0....
http://arxiv.org/abs/2310.11441v2
Set-of-Mark Prompting Unleashes Extraordinary Visual Grounding in GPT-4V
Jianwei Yang, Hao Zhang, Feng Li, Xueyan Zou, Chunyuan Li, Jianfeng Gao
cs.CV, cs.AI, cs.CL, cs.HC
We present Set-of-Mark (SoM), a new visual prompting method, to unleash the visual grounding abilities of large multimodal models (LMMs), such as GPT-4V. As illustrated in Fig. 1 (right), we employ off-the-shelf interactive segmentation models, such as SEEM/SAM, to partition an image into regions at different levels of...
http://arxiv.org/pdf/2310.11441v2
2023-11-06T07:39:49
2023-10-17T17:51:31
{ "title": [ 0.05130005, 0.003534317, 0.017196655, -0.023132324, -0.0034809113, -0.024032593, 0.015914917, -0.03427124, -0.015220642, 0.010307312, -0.005088806, -0.0037250519, 0.011199951, -0.047668457, -0.012367249, -0.020065308, -0.011054993, 0...
http://arxiv.org/abs/2310.11430v1
An Empirical Study of Translation Hypothesis Ensembling with Large Language Models
António Farinhas, José G. C. de Souza, André F. T. Martins
cs.CL
Large language models (LLMs) are becoming a one-fits-many solution, but they sometimes hallucinate or produce unreliable output. In this paper, we investigate how hypothesis ensembling can improve the quality of the generated text for the specific problem of LLM-based machine translation. We experiment with several tec...
http://arxiv.org/pdf/2310.11430v1
2023-10-17T17:40:21
2023-10-17T17:40:21
{ "title": [ -0.054473877, -0.012779236, -0.032440186, -0.04119873, -0.01399231, 0.005924225, -0.049682617, 0.0025138855, 0.0030460358, 0.034210205, -0.024887085, 0.016983032, 0.012359619, -0.011482239, -0.038604736, 0.016830444, 0.0011444092, 0....
http://arxiv.org/abs/2310.11398v2
Neural Attention: Enhancing QKV Calculation in Self-Attention Mechanism with Neural Networks
Muhan Zhang
cs.CL, cs.AI
In the realm of deep learning, the self-attention mechanism has substantiated its pivotal role across a myriad of tasks, encompassing natural language processing and computer vision. Despite achieving success across diverse applications, the traditional self-attention mechanism primarily leverages linear transformation...
http://arxiv.org/pdf/2310.11398v2
2023-10-24T17:12:49
2023-10-17T17:06:26
{ "title": [ -0.00579834, 0.005542755, -0.01939392, -0.0016241074, 0.011367798, -0.029037476, 0.024963379, -0.03390503, 0.0034999847, 0.04800415, -0.0022563934, -0.019989014, 0.018615723, -0.020568848, -0.07702637, 0.0023174286, 0.003921509, 0.02...
http://arxiv.org/abs/2310.11379v1
Robust Wake-Up Word Detection by Two-stage Multi-resolution Ensembles
Fernando López, Jordi Luque, Carlos Segura, Pablo Gómez
cs.SD, cs.CL, eess.AS
Voice-based interfaces rely on a wake-up word mechanism to initiate communication with devices. However, achieving a robust, energy-efficient, and fast detection remains a challenge. This paper addresses these real production needs by enhancing data with temporal alignments and using detection based on two phases with ...
http://arxiv.org/pdf/2310.11379v1
2023-10-17T16:22:18
2023-10-17T16:22:18
{ "title": [ -0.037078857, 0.02168274, -0.05960083, 0.008804321, 0.010940552, -0.03112793, -0.00090408325, -0.03277588, -0.0005264282, 0.04031372, -0.040405273, 0.00308609, -0.0032844543, -0.018081665, -0.064575195, -0.019180298, -0.004337311, 0....
http://arxiv.org/abs/2310.11374v1
DialogueLLM: Context and Emotion Knowledge-Tuned LLaMA Models for Emotion Recognition in Conversations
Yazhou Zhang, Mengyao Wang, Prayag Tiwari, Qiuchi Li, Benyou Wang, Jing Qin
cs.CL
Large language models (LLMs) and their variants have shown extraordinary efficacy across numerous downstream natural language processing (NLP) tasks, which has presented a new vision for the development of NLP. Despite their remarkable performance in natural language generating (NLG), LLMs lack a distinct focus on the ...
http://arxiv.org/pdf/2310.11374v1
2023-10-17T16:15:34
2023-10-17T16:15:34
{ "title": [ -0.022262573, 0.0038757324, -0.039855957, -0.048461914, 0.00021541119, -0.001698494, -0.056915283, -0.006591797, 0.022583008, 0.027450562, -0.01689148, -0.00592041, 0.0059928894, -0.02973938, -0.04449463, 0.0022659302, -0.010826111, ...
http://arxiv.org/abs/2310.11368v4
VECHR: A Dataset for Explainable and Robust Classification of Vulnerability Type in the European Court of Human Rights
Shanshan Xu, Leon Staufer, T. Y. S. S Santosh, Oana Ichim, Corina Heri, Matthias Grabmair
cs.CL
Recognizing vulnerability is crucial for understanding and implementing targeted support to empower individuals in need. This is especially important at the European Court of Human Rights (ECtHR), where the court adapts Convention standards to meet actual individual needs and thus ensures effective human rights protect...
http://arxiv.org/pdf/2310.11368v4
2023-10-24T12:30:25
2023-10-17T16:05:52
{ "title": [ 0.03643799, 0.016738892, -0.043304443, -0.056549072, 0.018310547, -0.022262573, -0.02355957, -0.015380859, 0.00333786, 0.025939941, -0.023803711, 0.00066137314, -0.0004069805, -0.028656006, -0.03768921, -0.021560669, -0.02557373, 0.0...
http://arxiv.org/abs/2310.11363v1
Disentangling the Linguistic Competence of Privacy-Preserving BERT
Stefan Arnold, Nils Kemmerzell, Annika Schreiner
cs.CL
Differential Privacy (DP) has been tailored to address the unique challenges of text-to-text privatization. However, text-to-text privatization is known for degrading the performance of language models when trained on perturbed text. Employing a series of interpretation techniques on the internal representations extrac...
http://arxiv.org/pdf/2310.11363v1
2023-10-17T16:00:26
2023-10-17T16:00:26
{ "title": [ -0.018218994, -0.036621094, -0.019363403, -0.031311035, -0.04373169, -0.0051651, -0.011604309, 0.017623901, -0.02822876, 0.04663086, 0.008979797, -0.00969696, 0.025115967, -0.032714844, -0.008888245, 0.009086609, -0.0152282715, 0.024...
http://arxiv.org/abs/2310.11360v1
Enhancing Neural Machine Translation with Semantic Units
Langlin Huang, Shuhao Gu, Zhuocheng Zhang, Yang Feng
cs.CL, I.2.7
Conventional neural machine translation (NMT) models typically use subwords and words as the basic units for model input and comprehension. However, complete words and phrases composed of several tokens are often the fundamental units for expressing semantics, referred to as semantic units. To address this issue, we pr...
http://arxiv.org/pdf/2310.11360v1
2023-10-17T15:55:31
2023-10-17T15:55:31
{ "title": [ -0.019851685, -0.0073928833, -0.021499634, -0.05307007, -0.0048713684, 0.003627777, -0.034698486, 0.00756073, -0.009941101, 0.05645752, -0.009597778, 0.0135650635, 0.028900146, -0.007133484, -0.04248047, 0.02468872, -0.00774765, 0.04...
http://arxiv.org/abs/2310.11344v1
The effect of stemming and lemmatization on Portuguese fake news text classification
Lucca de Freitas Santos, Murilo Varges da Silva
cs.CL, cs.AI
With the popularization of the internet, smartphones and social media, information is being spread quickly and easily way, which implies bigger traffic of information in the world, but there is a problem that is harming society with the dissemination of fake news. With a bigger flow of information, some people are tryi...
http://arxiv.org/pdf/2310.11344v1
2023-10-17T15:26:40
2023-10-17T15:26:40
{ "title": [ -0.014083862, 0.010559082, -0.031219482, -0.029342651, 0.005382538, 0.009170532, -0.015045166, -0.013465881, -0.028396606, 0.04348755, -0.03265381, 0.01133728, 0.02482605, -0.010192871, -0.032989502, -0.015716553, -0.03643799, 0.0575...
http://arxiv.org/abs/2310.11324v1
Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting
Melanie Sclar, Yejin Choi, Yulia Tsvetkov, Alane Suhr
cs.CL, cs.AI, cs.LG
As large language models (LLMs) are adopted as a fundamental component of language technologies, it is crucial to accurately characterize their performance. Because choices in prompt design can strongly influence model behavior, this design process is critical in effectively using any modern pre-trained generative lang...
http://arxiv.org/pdf/2310.11324v1
2023-10-17T15:03:30
2023-10-17T15:03:30
{ "title": [ -0.004184723, -0.016860962, -0.013343811, 0.008262634, -0.05078125, -0.026855469, -0.03265381, -0.043670654, -0.016235352, 0.04168701, -0.039367676, -0.03427124, 0.018844604, -0.02558899, -0.02406311, -0.03125, -0.0107421875, 0.04309...
http://arxiv.org/abs/2310.11318v1
Utilising a Large Language Model to Annotate Subject Metadata: A Case Study in an Australian National Research Data Catalogue
Shiwei Zhang, Mingfang Wu, Xiuzhen Zhang
cs.CL, cs.AI
In support of open and reproducible research, there has been a rapidly increasing number of datasets made available for research. As the availability of datasets increases, it becomes more important to have quality metadata for discovering and reusing them. Yet, it is a common issue that datasets often lack quality met...
http://arxiv.org/pdf/2310.11318v1
2023-10-17T14:52:33
2023-10-17T14:52:33
{ "title": [ -0.030548096, -0.0135269165, -0.04727173, -0.05722046, -0.014221191, -0.010231018, -0.049804688, 0.00869751, 0.010719299, 0.050994873, 0.0040512085, -0.016540527, -0.007637024, -0.017959595, -0.08087158, -0.012649536, -0.008392334, 0...
http://arxiv.org/abs/2310.13013v1
Generative error correction for code-switching speech recognition using large language models
Chen Chen, Yuchen Hu, Chao-Han Huck Yang, Hexin Liu, Sabato Marco Siniscalchi, Eng Siong Chng
cs.CL, cs.AI, cs.SD, eess.AS
Code-switching (CS) speech refers to the phenomenon of mixing two or more languages within the same sentence. Despite the recent advances in automatic speech recognition (ASR), CS-ASR is still a challenging task ought to the grammatical structure complexity of the phenomenon and the data scarcity of specific training c...
http://arxiv.org/pdf/2310.13013v1
2023-10-17T14:49:48
2023-10-17T14:49:48
{ "title": [ -0.017654419, 0.000541687, -0.035064697, -0.031433105, -0.0019454956, -0.0029907227, -0.044433594, -0.0038394928, -0.007423401, 0.045135498, -0.034484863, -0.015853882, 0.015609741, -0.020507812, -0.043670654, -0.016143799, -0.013557434,...
http://arxiv.org/abs/2310.11303v1
QADYNAMICS: Training Dynamics-Driven Synthetic QA Diagnostic for Zero-Shot Commonsense Question Answering
Haochen Shi, Weiqi Wang, Tianqing Fang, Baixuan Xu, Wenxuan Ding, Xin Liu, Yangqiu Song
cs.CL
Zero-shot commonsense Question-Answering (QA) requires models to reason about general situations beyond specific benchmarks. State-of-the-art approaches fine-tune language models on QA pairs constructed from CommonSense Knowledge Bases (CSKBs) to equip the models with more commonsense knowledge in a QA context. However...
http://arxiv.org/pdf/2310.11303v1
2023-10-17T14:27:34
2023-10-17T14:27:34
{ "title": [ 0.023635864, -0.026428223, 0.0009250641, -0.048309326, -0.029083252, -0.017730713, 0.00762558, 0.012184143, 0.01776123, 0.034851074, -0.020050049, 0.011489868, -0.012458801, -0.007232666, -0.0473938, -0.0067825317, -0.00027251244, 0....
http://arxiv.org/abs/2310.11282v1
ChapGTP, ILLC's Attempt at Raising a BabyLM: Improving Data Efficiency by Automatic Task Formation
Jaap Jumelet, Michael Hanna, Marianne de Heer Kloots, Anna Langedijk, Charlotte Pouw, Oskar van der Wal
cs.CL
We present the submission of the ILLC at the University of Amsterdam to the BabyLM challenge (Warstadt et al., 2023), in the strict-small track. Our final model, ChapGTP, is a masked language model that was trained for 200 epochs, aided by a novel data augmentation technique called Automatic Task Formation. We discuss ...
http://arxiv.org/pdf/2310.11282v1
2023-10-17T14:06:06
2023-10-17T14:06:06
{ "title": [ -0.018600464, -0.024780273, -0.058898926, -0.021759033, -0.01007843, -0.033111572, 0.003414154, 0.021636963, -0.039520264, 0.041656494, -0.037719727, 0.03463745, -0.008399963, -0.01486969, -0.0034866333, -0.04168701, -0.006362915, -0...
http://arxiv.org/abs/2310.11275v1
xMEN: A Modular Toolkit for Cross-Lingual Medical Entity Normalization
Florian Borchert, Ignacio Llorca, Roland Roller, Bert Arnrich, Matthieu-P. Schapranow
cs.CL
Objective: To improve performance of medical entity normalization across many languages, especially when fewer language resources are available compared to English. Materials and Methods: We introduce xMEN, a modular system for cross-lingual medical entity normalization, which performs well in both low- and high-reso...
http://arxiv.org/pdf/2310.11275v1
2023-10-17T13:53:57
2023-10-17T13:53:57
{ "title": [ -0.015174866, -0.00073480606, -0.046783447, -0.08544922, 0.034088135, 0.0116119385, -0.050354004, -0.0018005371, -0.0033054352, 0.044433594, 0.025222778, 0.019760132, 0.033233643, -0.02432251, -0.017044067, 0.041137695, -0.0025367737, ...
http://arxiv.org/abs/2310.11266v1
Emulating Human Cognitive Processes for Expert-Level Medical Question-Answering with Large Language Models
Khushboo Verma, Marina Moore, Stephanie Wottrich, Karla Robles López, Nishant Aggarwal, Zeel Bhatt, Aagamjit Singh, Bradford Unroe, Salah Basheer, Nitish Sachdeva, Prinka Arora, Harmanjeet Kaur, Tanupreet Kaur, Tevon Hood, Anahi Marquez, Tushar Varshney, Nanfu Deng, Azaan Ramani, Pawanraj Ishwara, Maimoona Saeed, Tatia...
cs.CL, cs.AI, cs.NE
In response to the pressing need for advanced clinical problem-solving tools in healthcare, we introduce BooksMed, a novel framework based on a Large Language Model (LLM). BooksMed uniquely emulates human cognitive processes to deliver evidence-based and reliable responses, utilizing the GRADE (Grading of Recommendatio...
http://arxiv.org/pdf/2310.11266v1
2023-10-17T13:39:26
2023-10-17T13:39:26
{ "title": [ 0.0015716553, -0.009109497, -0.015052795, -0.05633545, -0.00756073, -0.00069856644, -0.031707764, -0.01134491, 0.009216309, 0.022720337, -0.004711151, -0.026947021, 0.019454956, -0.0211792, -0.053955078, 0.005619049, 0.015991211, 0.0...
http://arxiv.org/abs/2310.11258v2
Utilizing Weak Supervision To Generate Indonesian Conservation Dataset
Mega Fransiska, Diah Pitaloka, Saripudin, Satrio Putra, Lintang Sutawika
cs.CL
Weak supervision has emerged as a promising approach for rapid and large-scale dataset creation in response to the increasing demand for accelerated NLP development. By leveraging labeling functions, weak supervision allows practitioners to generate datasets quickly by creating learned label models that produce soft-la...
http://arxiv.org/pdf/2310.11258v2
2023-10-24T12:01:14
2023-10-17T13:23:18
{ "title": [ 0.00806427, -0.024795532, -0.018005371, -0.042175293, -0.021697998, -0.009849548, -0.026489258, -0.00894928, 0.001329422, 0.00022017956, -0.04058838, 0.00094747543, -0.080078125, -0.01574707, -0.044189453, -0.052886963, -0.014305115, ...
http://arxiv.org/abs/2310.11252v1
Revealing the Unwritten: Visual Investigation of Beam Search Trees to Address Language Model Prompting Challenges
Thilo Spinner, Rebecca Kehlbeck, Rita Sevastjanova, Tobias Stähle, Daniel A. Keim, Oliver Deussen, Andreas Spitz, Mennatallah El-Assady
cs.CL, cs.AI, cs.HC, H.5.2; I.2.7
The growing popularity of generative language models has amplified interest in interactive methods to guide model outputs. Prompt refinement is considered one of the most effective means to influence output among these methods. We identify several challenges associated with prompting large language models, categorized ...
http://arxiv.org/pdf/2310.11252v1
2023-10-17T13:20:16
2023-10-17T13:20:16
{ "title": [ -0.02999878, -0.03125, 0.01272583, -0.011299133, -0.080566406, 0.010810852, -0.0826416, -0.03225708, 0.014411926, 0.018829346, -0.029251099, -0.0026245117, -0.0002875328, -0.019424438, -0.04748535, -0.012916565, 0.0090408325, 0.02343...
http://arxiv.org/abs/2310.11248v1
CrossCodeEval: A Diverse and Multilingual Benchmark for Cross-File Code Completion
Yangruibo Ding, Zijian Wang, Wasi Uddin Ahmad, Hantian Ding, Ming Tan, Nihal Jain, Murali Krishna Ramanathan, Ramesh Nallapati, Parminder Bhatia, Dan Roth, Bing Xiang
cs.LG, cs.CL, cs.SE
Code completion models have made significant progress in recent years, yet current popular evaluation datasets, such as HumanEval and MBPP, predominantly focus on code completion tasks within a single file. This over-simplified setting falls short of representing the real-world software development scenario where repos...
http://arxiv.org/pdf/2310.11248v1
2023-10-17T13:18:01
2023-10-17T13:18:01
{ "title": [ -0.035705566, -0.025268555, -0.04547119, -0.087524414, 0.0010433197, 0.0055351257, -0.06085205, 0.018081665, 0.006000519, 0.05078125, -0.036254883, 0.04159546, 0.023223877, -0.0001963377, -0.016693115, 0.019714355, -0.02178955, 0.043...
http://arxiv.org/abs/2310.11244v1
Entity Matching using Large Language Models
Ralph Peeters, Christian Bizer
cs.CL, cs.LG
Entity Matching is the task of deciding whether two entity descriptions refer to the same real-world entity. Entity Matching is a central step in most data integration pipelines and an enabler for many e-commerce applications which require to match products offers from different vendors. State-of-the-art entity matchin...
http://arxiv.org/pdf/2310.11244v1
2023-10-17T13:12:32
2023-10-17T13:12:32
{ "title": [ -0.020217896, 0.002571106, -0.041503906, -0.07519531, 0.03137207, 0.0056762695, -0.045440674, -0.0056991577, 0.026855469, 0.02470398, -0.016311646, -0.010322571, 0.029312134, -0.014984131, -0.034576416, 0.020263672, -0.012008667, 0.0...