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/2309.05951v1
Balanced and Explainable Social Media Analysis for Public Health with Large Language Models
Yan Jiang, Ruihong Qiu, Yi Zhang, Peng-Fei Zhang
cs.CL
As social media becomes increasingly popular, more and more public health activities emerge, which is worth noting for pandemic monitoring and government decision-making. Current techniques for public health analysis involve popular models such as BERT and large language models (LLMs). Although recent progress in LLMs ...
http://arxiv.org/pdf/2309.05951v1
2023-09-12T04:15:34
2023-09-12T04:15:34
{ "title": [ 0.010612488, 0.00623703, -0.038208008, -0.08404541, 0.035827637, -0.0018844604, -0.04171753, -0.0035190582, -0.033081055, 0.0234375, -0.015617371, 0.0345459, -0.032592773, -0.011703491, -0.014083862, -0.007247925, -0.0023708344, 0.02...
http://arxiv.org/abs/2309.05950v2
Language Models as Black-Box Optimizers for Vision-Language Models
Shihong Liu, Samuel Yu, Zhiqiu Lin, Deepak Pathak, Deva Ramanan
cs.CL, cs.CV, cs.LG, cs.MM
Vision-language models (VLMs) pre-trained on web-scale datasets have demonstrated remarkable capabilities across a variety of vision and multimodal tasks. Currently, fine-tuning methods for VLMs mainly operate in a white-box setting, requiring access to model parameters for backpropagation. However, many VLMs rely on p...
http://arxiv.org/pdf/2309.05950v2
2023-09-25T04:35:32
2023-09-12T04:03:41
{ "title": [ -0.033447266, -0.0016107559, -0.022888184, -0.045776367, -0.012710571, 0.0005426407, -0.062683105, -0.017440796, 0.015594482, 0.033111572, -0.030914307, -0.005607605, 0.009025574, -0.005680084, -0.0513916, 0.0005364418, 0.0066833496, ...
http://arxiv.org/abs/2309.05938v2
Answering Subjective Induction Questions on Products by Summarizing Multi-sources Multi-viewpoints Knowledge
Yufeng Zhang, Meng-xiang Wang, Jianxing Yu
cs.CL, cs.AI
This paper proposes a new task in the field of Answering Subjective Induction Question on Products (SUBJPQA). The answer to this kind of question is non-unique, but can be interpreted from many perspectives. For example, the answer to 'whether the phone is heavy' has a variety of different viewpoints. A satisfied answe...
http://arxiv.org/pdf/2309.05938v2
2023-10-06T12:35:20
2023-09-12T03:27:08
{ "title": [ -0.01701355, -0.015716553, -0.051818848, -0.06311035, -0.016357422, -0.011810303, -0.0418396, 0.02079773, -0.004638672, 0.006465912, -0.046569824, 0.015716553, -0.044433594, -0.017623901, -0.033996582, -0.030471802, 0.0031490326, 0.0...
http://arxiv.org/abs/2309.05936v1
Do PLMs Know and Understand Ontological Knowledge?
Weiqi Wu, Chengyue Jiang, Yong Jiang, Pengjun Xie, Kewei Tu
cs.CL
Ontological knowledge, which comprises classes and properties and their relationships, is integral to world knowledge. It is significant to explore whether Pretrained Language Models (PLMs) know and understand such knowledge. However, existing PLM-probing studies focus mainly on factual knowledge, lacking a systematic ...
http://arxiv.org/pdf/2309.05936v1
2023-09-12T03:20:50
2023-09-12T03:20:50
{ "title": [ -0.014587402, 0.008110046, -0.042419434, -0.044311523, -0.025939941, 0.00063705444, -0.0713501, -0.012794495, 0.0056152344, -0.008560181, -0.030410767, -0.026489258, -0.013809204, -0.043121338, 0.005115509, -0.030685425, 0.01651001, ...
http://arxiv.org/abs/2309.05922v1
A Survey of Hallucination in Large Foundation Models
Vipula Rawte, Amit Sheth, Amitava Das
cs.AI, cs.CL, cs.IR
Hallucination in a foundation model (FM) refers to the generation of content that strays from factual reality or includes fabricated information. This survey paper provides an extensive overview of recent efforts that aim to identify, elucidate, and tackle the problem of hallucination, with a particular focus on ``Larg...
http://arxiv.org/pdf/2309.05922v1
2023-09-12T02:34:06
2023-09-12T02:34:06
{ "title": [ -0.0077552795, 0.021820068, -0.027008057, -0.0059661865, -0.0076904297, -0.011398315, -0.0037975311, -0.026794434, 0.0236969, 0.03366089, -0.007980347, -0.006954193, -0.010955811, -0.02734375, -0.049926758, 0.019714355, -0.00983429, ...
http://arxiv.org/abs/2309.05920v1
SAGE: Structured Attribute Value Generation for Billion-Scale Product Catalogs
Athanasios N. Nikolakopoulos, Swati Kaul, Siva Karthik Gade, Bella Dubrov, Umit Batur, Suleiman Ali Khan
cs.IR, cs.AI, cs.CL
We introduce SAGE; a Generative LLM for inferring attribute values for products across world-wide e-Commerce catalogs. We introduce a novel formulation of the attribute-value prediction problem as a Seq2Seq summarization task, across languages, product types and target attributes. Our novel modeling approach lifts the ...
http://arxiv.org/pdf/2309.05920v1
2023-09-12T02:24:16
2023-09-12T02:24:16
{ "title": [ -0.03338623, -0.022598267, -0.07836914, -0.08270264, -0.0041656494, -0.012321472, -0.06732178, 0.022354126, -0.010169983, 0.014205933, -0.016555786, 0.003894806, 0.0018587112, -0.024765015, -0.032409668, -0.00087976456, -0.006965637, ...
http://arxiv.org/abs/2309.05918v3
Stochastic LLMs do not Understand Language: Towards Symbolic, Explainable and Ontologically Based LLMs
Walid S. Saba
cs.CL, cs.AI
In our opinion the exuberance surrounding the relative success of data-driven large language models (LLMs) is slightly misguided and for several reasons (i) LLMs cannot be relied upon for factual information since for LLMs all ingested text (factual or non-factual) was created equal; (ii) due to their subsymbolic na-tu...
http://arxiv.org/pdf/2309.05918v3
2023-09-14T12:58:39
2023-09-12T02:14:05
{ "title": [ -0.06097412, 0.0011520386, -0.04663086, -0.04763794, -0.02217102, 0.008049011, -0.08642578, -0.0154418945, -0.005443573, -0.0018014908, -0.024475098, 0.013320923, -0.02468872, -0.030807495, -0.0005121231, -0.009986877, -0.02079773, 0...
http://arxiv.org/abs/2309.08624v1
Challenges in Annotating Datasets to Quantify Bias in Under-represented Society
Vithya Yogarajan, Gillian Dobbie, Timothy Pistotti, Joshua Bensemann, Kobe Knowles
cs.CL, cs.AI
Recent advances in artificial intelligence, including the development of highly sophisticated large language models (LLM), have proven beneficial in many real-world applications. However, evidence of inherent bias encoded in these LLMs has raised concerns about equity. In response, there has been an increase in researc...
http://arxiv.org/pdf/2309.08624v1
2023-09-11T22:24:39
2023-09-11T22:24:39
{ "title": [ 0.03692627, 0.03652954, -0.019226074, -0.03744507, 0.025268555, -0.03540039, -0.008728027, -0.03024292, -0.0010385513, 0.03050232, -0.059539795, 0.013160706, -0.01789856, -0.004207611, -0.019943237, -0.0137786865, -0.018447876, 0.054...
http://arxiv.org/abs/2309.07062v1
Large Language Models for Compiler Optimization
Chris Cummins, Volker Seeker, Dejan Grubisic, Mostafa Elhoushi, Youwei Liang, Baptiste Roziere, Jonas Gehring, Fabian Gloeckle, Kim Hazelwood, Gabriel Synnaeve, Hugh Leather
cs.PL, cs.AI, cs.CL, cs.LG
We explore the novel application of Large Language Models to code optimization. We present a 7B-parameter transformer model trained from scratch to optimize LLVM assembly for code size. The model takes as input unoptimized assembly and outputs a list of compiler options to best optimize the program. Crucially, during t...
http://arxiv.org/pdf/2309.07062v1
2023-09-11T22:11:46
2023-09-11T22:11:46
{ "title": [ -0.06427002, -0.017501831, -0.035339355, -0.06262207, -0.007987976, 0.02368164, -0.072021484, 0.02758789, 0.00005042553, 0.0287323, -0.017669678, 0.0074539185, -0.011070251, -0.020843506, -0.03225708, 0.002910614, -0.011634827, 0.020...
http://arxiv.org/abs/2309.05833v3
PACE-LM: Prompting and Augmentation for Calibrated Confidence Estimation with GPT-4 in Cloud Incident Root Cause Analysis
Dylan Zhang, Xuchao Zhang, Chetan Bansal, Pedro Las-Casas, Rodrigo Fonseca, Saravan Rajmohan
cs.CL, cs.AI, cs.LG, cs.SE
Major cloud providers have employed advanced AI-based solutions like large language models to aid humans in identifying the root causes of cloud incidents. Despite the growing prevalence of AI-driven assistants in the root cause analysis process, their effectiveness in assisting on-call engineers is constrained by low ...
http://arxiv.org/pdf/2309.05833v3
2023-09-29T16:25:16
2023-09-11T21:24:00
{ "title": [ 0.009918213, 0.009284973, -0.055236816, -0.06359863, 0.0038166046, -0.015914917, -0.06008911, -0.0031013489, -0.024597168, -0.014678955, -0.023635864, 0.03942871, -0.038726807, -0.059051514, 0.036865234, -0.018859863, 0.0017356873, -...
http://arxiv.org/abs/2309.05804v1
Hi Model, generating 'nice' instead of 'good' is not as bad as generating 'rice'! Towards Context and Semantic Infused Dialogue Generation Loss Function and Evaluation Metric
Abhisek Tiwari, Muhammed Sinan, Kaushik Roy, Amit Sheth, Sriparna Saha, Pushpak Bhattacharyya
cs.CL
Over the past two decades, dialogue modeling has made significant strides, moving from simple rule-based responses to personalized and persuasive response generation. However, despite these advancements, the objective functions and evaluation metrics for dialogue generation have remained stagnant, i.e., cross-entropy a...
http://arxiv.org/pdf/2309.05804v1
2023-09-11T20:16:38
2023-09-11T20:16:38
{ "title": [ -0.025894165, -0.021881104, -0.015853882, -0.009460449, -0.031036377, 0.013412476, -0.00881958, 0.005897522, 0.015960693, 0.024749756, -0.010223389, 0.0044059753, -0.012252808, -0.014862061, -0.023620605, -0.023040771, -0.032592773, ...
http://arxiv.org/abs/2309.13061v2
Applying BioBERT to Extract Germline Gene-Disease Associations for Building a Knowledge Graph from the Biomedical Literature
Armando D. Diaz Gonzalez, Songhui Yue, Sean T. Hayes, Kevin S. Hughes
cs.CL, cs.CY
Published biomedical information has and continues to rapidly increase. The recent advancements in Natural Language Processing (NLP), have generated considerable interest in automating the extraction, normalization, and representation of biomedical knowledge about entities such as genes and diseases. Our study analyzes...
http://arxiv.org/pdf/2309.13061v2
2023-09-30T04:31:22
2023-09-11T18:05:12
{ "title": [ -0.0259552, 0.0033073425, -0.052093506, -0.07550049, -0.017562866, 0.012161255, -0.0736084, -0.001540184, 0.008384705, 0.03845215, 0.021377563, 0.0020961761, -0.0028762817, -0.022659302, -0.052581787, 0.011138916, -0.015548706, 0.048...
http://arxiv.org/abs/2309.05660v1
Hypothesis Search: Inductive Reasoning with Language Models
Ruocheng Wang, Eric Zelikman, Gabriel Poesia, Yewen Pu, Nick Haber, Noah D. Goodman
cs.LG, cs.AI, cs.CL
Inductive reasoning is a core problem-solving capacity: humans can identify underlying principles from a few examples, which can then be robustly generalized to novel scenarios. Recent work has evaluated large language models (LLMs) on inductive reasoning tasks by directly prompting them yielding "in context learning."...
http://arxiv.org/pdf/2309.05660v1
2023-09-11T17:56:57
2023-09-11T17:56:57
{ "title": [ -0.04107666, -0.031555176, -0.025527954, -0.048858643, 0.0014781952, 0.013786316, -0.032043457, -0.0072746277, -0.0015192032, 0.010726929, -0.025680542, 0.0072669983, -0.020980835, -0.024871826, -0.034057617, 0.0008788109, -0.008369446, ...
http://arxiv.org/abs/2309.05689v1
Large Language Model for Science: A Study on P vs. NP
Qingxiu Dong, Li Dong, Ke Xu, Guangyan Zhou, Yaru Hao, Zhifang Sui, Furu Wei
cs.CL, cs.AI
In this work, we use large language models (LLMs) to augment and accelerate research on the P versus NP problem, one of the most important open problems in theoretical computer science and mathematics. Specifically, we propose Socratic reasoning, a general framework that promotes in-depth thinking with LLMs for complex...
http://arxiv.org/pdf/2309.05689v1
2023-09-11T17:49:27
2023-09-11T17:49:27
{ "title": [ -0.05267334, -0.018463135, -0.032165527, -0.032165527, -0.011756897, 0.0068511963, -0.044677734, -0.013092041, -0.0102005005, 0.02330017, -0.0026893616, -0.0020198822, -0.019866943, -0.018325806, -0.040222168, -0.00028824806, -0.02754211...
http://arxiv.org/abs/2309.05653v3
MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuning
Xiang Yue, Xingwei Qu, Ge Zhang, Yao Fu, Wenhao Huang, Huan Sun, Yu Su, Wenhu Chen
cs.CL
We introduce MAmmoTH, a series of open-source large language models (LLMs) specifically tailored for general math problem-solving. The MAmmoTH models are trained on MathInstruct, our meticulously curated instruction tuning dataset. MathInstruct is compiled from 13 math datasets with intermediate rationales, six of whic...
http://arxiv.org/pdf/2309.05653v3
2023-10-03T02:48:42
2023-09-11T17:47:22
{ "title": [ -0.04849243, -0.01777649, -0.02104187, -0.03173828, -0.0231781, 0.0033283234, -0.025558472, 0.006969452, -0.008583069, 0.0155181885, -0.03817749, 0.02696228, -0.02104187, -0.009208679, -0.019561768, -0.01436615, 0.0019016266, 0.00703...
http://arxiv.org/abs/2309.05619v1
Effective Proxy for Human Labeling: Ensemble Disagreement Scores in Large Language Models for Industrial NLP
Wei Du, Laksh Advani, Yashmeet Gambhir, Daniel J Perry, Prashant Shiralkar, Zhengzheng Xing, Aaron Colak
cs.CL
Large language models (LLMs) have demonstrated significant capability to generalize across a large number of NLP tasks. For industry applications, it is imperative to assess the performance of the LLM on unlabeled production data from time to time to validate for a real-world setting. Human labeling to assess model err...
http://arxiv.org/pdf/2309.05619v1
2023-09-11T17:07:01
2023-09-11T17:07:01
{ "title": [ -0.041412354, -0.0032997131, -0.04196167, -0.033843994, -0.017822266, -0.01537323, -0.053344727, -0.0037231445, -0.00970459, 0.0074539185, -0.056396484, 0.013702393, -0.015563965, -0.010948181, -0.033081055, -0.024002075, -0.032806396, ...
http://arxiv.org/abs/2309.05608v1
Incorporating Pre-trained Model Prompting in Multimodal Stock Volume Movement Prediction
Ruibo Chen, Zhiyuan Zhang, Yi Liu, Ruihan Bao, Keiko Harimoto, Xu Sun
cs.CL, cs.CE
Multimodal stock trading volume movement prediction with stock-related news is one of the fundamental problems in the financial area. Existing multimodal works that train models from scratch face the problem of lacking universal knowledge when modeling financial news. In addition, the models ability may be limited by t...
http://arxiv.org/pdf/2309.05608v1
2023-09-11T16:47:01
2023-09-11T16:47:01
{ "title": [ 0.015449524, 0.009284973, -0.04437256, -0.07574463, -0.0029678345, -0.013420105, -0.05404663, -0.009490967, 0.008239746, 0.040283203, -0.009033203, -0.010978699, 0.0030536652, -0.020614624, -0.055755615, -0.002445221, 0.01651001, 0.0...
http://arxiv.org/abs/2309.05605v2
Memory Injections: Correcting Multi-Hop Reasoning Failures during Inference in Transformer-Based Language Models
Mansi Sakarvadia, Aswathy Ajith, Arham Khan, Daniel Grzenda, Nathaniel Hudson, André Bauer, Kyle Chard, Ian Foster
cs.CL, cs.AI, cs.LG
Answering multi-hop reasoning questions requires retrieving and synthesizing information from diverse sources. Large Language Models (LLMs) struggle to perform such reasoning consistently. Here we propose an approach to pinpoint and rectify multi-hop reasoning failures through targeted memory injections on LLM attentio...
http://arxiv.org/pdf/2309.05605v2
2023-09-12T20:18:20
2023-09-11T16:39:30
{ "title": [ -0.022735596, -0.026168823, -0.023605347, -0.05014038, -0.02041626, 0.002878189, -0.058624268, 0.012710571, -0.010894775, 0.023330688, -0.041412354, 0.011077881, 0.03982544, -0.015571594, -0.01222229, -0.010437012, -0.025009155, 0.03...
http://arxiv.org/abs/2309.05569v1
ITI-GEN: Inclusive Text-to-Image Generation
Cheng Zhang, Xuanbai Chen, Siqi Chai, Chen Henry Wu, Dmitry Lagun, Thabo Beeler, Fernando De la Torre
cs.CV, cs.AI, cs.CL, cs.LG
Text-to-image generative models often reflect the biases of the training data, leading to unequal representations of underrepresented groups. This study investigates inclusive text-to-image generative models that generate images based on human-written prompts and ensure the resulting images are uniformly distributed ac...
http://arxiv.org/pdf/2309.05569v1
2023-09-11T15:54:30
2023-09-11T15:54:30
{ "title": [ -0.018035889, -0.0031986237, -0.035736084, -0.05178833, -0.0035171509, 0.0033302307, -0.040161133, -0.02331543, 0.02659607, 0.051757812, -0.021575928, 0.02053833, 0.0068588257, -0.008079529, -0.047912598, 0.016738892, 0.000026345253, ...
http://arxiv.org/abs/2309.05557v3
An Empirical Study of NetOps Capability of Pre-Trained Large Language Models
Yukai Miao, Yu Bai, Li Chen, Dan Li, Haifeng Sun, Xizheng Wang, Ziqiu Luo, Yanyu Ren, Dapeng Sun, Xiuting Xu, Qi Zhang, Chao Xiang, Xinchi Li
cs.CL, cs.AI, cs.NI
Nowadays, the versatile capabilities of Pre-trained Large Language Models (LLMs) have attracted much attention from the industry. However, some vertical domains are more interested in the in-domain capabilities of LLMs. For the Networks domain, we present NetEval, an evaluation set for measuring the comprehensive capab...
http://arxiv.org/pdf/2309.05557v3
2023-09-19T16:04:25
2023-09-11T15:45:40
{ "title": [ -0.012466431, -0.02923584, -0.02218628, -0.08105469, -0.015083313, -0.0015745163, -0.05630493, 0.02520752, -0.009628296, 0.033325195, -0.021957397, -0.008583069, -0.005874634, -0.016525269, -0.051696777, -0.01499176, -0.008651733, 0....
http://arxiv.org/abs/2309.05542v1
Kani: A Lightweight and Highly Hackable Framework for Building Language Model Applications
Andrew Zhu, Liam Dugan, Alyssa Hwang, Chris Callison-Burch
cs.SE, cs.AI, cs.CL, I.2.7
Language model applications are becoming increasingly popular and complex, often including features like tool usage and retrieval augmentation. However, existing frameworks for such applications are often opinionated, deciding for developers how their prompts ought to be formatted and imposing limitations on customizab...
http://arxiv.org/pdf/2309.05542v1
2023-09-11T15:27:59
2023-09-11T15:27:59
{ "title": [ -0.059539795, -0.031402588, -0.05517578, -0.064331055, -0.016052246, 0.0184021, -0.08258057, 0.011802673, 0.0017776489, 0.03062439, 0.005176544, -0.015930176, 0.023254395, -0.028076172, -0.04736328, 0.010856628, -0.007896423, 0.03851...
http://arxiv.org/abs/2309.05534v1
PAI-Diffusion: Constructing and Serving a Family of Open Chinese Diffusion Models for Text-to-image Synthesis on the Cloud
Chengyu Wang, Zhongjie Duan, Bingyan Liu, Xinyi Zou, Cen Chen, Kui Jia, Jun Huang
cs.CL, cs.AI, cs.CV
Text-to-image synthesis for the Chinese language poses unique challenges due to its large vocabulary size, and intricate character relationships. While existing diffusion models have shown promise in generating images from textual descriptions, they often neglect domain-specific contexts and lack robustness in handling...
http://arxiv.org/pdf/2309.05534v1
2023-09-11T15:18:28
2023-09-11T15:18:28
{ "title": [ -0.061462402, -0.0054512024, -0.025009155, -0.021499634, -0.0184021, 0.031402588, -0.040008545, -0.0067329407, 0.009269714, 0.04257202, -0.028930664, 0.030380249, 0.017852783, -0.024795532, -0.018585205, 0.005844116, -0.010848999, 0....
http://arxiv.org/abs/2309.05519v2
NExT-GPT: Any-to-Any Multimodal LLM
Shengqiong Wu, Hao Fei, Leigang Qu, Wei Ji, Tat-Seng Chua
cs.AI, cs.CL, cs.LG
While recently Multimodal Large Language Models (MM-LLMs) have made exciting strides, they mostly fall prey to the limitation of only input-side multimodal understanding, without the ability to produce content in multiple modalities. As we humans always perceive the world and communicate with people through various mod...
http://arxiv.org/pdf/2309.05519v2
2023-09-13T16:49:34
2023-09-11T15:02:25
{ "title": [ -0.020355225, -0.0035171509, -0.041748047, -0.053985596, -0.0473938, 0.013877869, -0.104003906, 0.009819031, -0.0155181885, 0.011985779, -0.03201294, 0.021377563, 0.010093689, -0.035003662, 0.04159546, -0.028198242, -0.013267517, 0.0...
http://arxiv.org/abs/2309.05516v2
Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs
Wenhua Cheng, Weiwei Zhang, Haihao Shen, Yiyang Cai, Xin He, Kaokao Lv
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have proven their exceptional capabilities in performing language-related tasks. However, their deployment poses significant challenges due to their considerable memory and storage requirements. In response to this issue, weight-only quantization, particularly 3 and 4-bit weight-only quanti...
http://arxiv.org/pdf/2309.05516v2
2023-09-28T09:05:57
2023-09-11T14:58:23
{ "title": [ -0.03677368, -0.011566162, -0.05545044, -0.04574585, -0.03933716, -0.01524353, -0.07574463, 0.006996155, -0.030700684, 0.050231934, -0.010734558, 0.0021076202, -0.0037765503, -0.027114868, -0.027526855, -0.02722168, 0.029525757, -0.0...
http://arxiv.org/abs/2309.05503v1
Long-Range Transformer Architectures for Document Understanding
Thibault Douzon, Stefan Duffner, Christophe Garcia, Jérémy Espinas
cs.CL, 68T01, I.2.7
Since their release, Transformers have revolutionized many fields from Natural Language Understanding to Computer Vision. Document Understanding (DU) was not left behind with first Transformer based models for DU dating from late 2019. However, the computational complexity of the self-attention operation limits their c...
http://arxiv.org/pdf/2309.05503v1
2023-09-11T14:45:24
2023-09-11T14:45:24
{ "title": [ -0.016464233, 0.0077323914, -0.012458801, -0.05154419, 0.0056762695, 0.0038032532, -0.03427124, -0.013473511, 0.009231567, 0.04058838, -0.045410156, -0.01927185, 0.024810791, -0.008483887, -0.04421997, -0.001909256, 0.0093688965, 0.0...
http://arxiv.org/abs/2309.05501v1
Black-Box Analysis: GPTs Across Time in Legal Textual Entailment Task
Ha-Thanh Nguyen, Randy Goebel, Francesca Toni, Kostas Stathis, Ken Satoh
cs.CL, cs.AI
The evolution of Generative Pre-trained Transformer (GPT) models has led to significant advancements in various natural language processing applications, particularly in legal textual entailment. We present an analysis of GPT-3.5 (ChatGPT) and GPT-4 performances on COLIEE Task 4 dataset, a prominent benchmark in this d...
http://arxiv.org/pdf/2309.05501v1
2023-09-11T14:43:54
2023-09-11T14:43:54
{ "title": [ 0.015098572, -0.011856079, -0.044677734, -0.07385254, -0.0054855347, -0.02229309, -0.025146484, 0.0132751465, 0.0345459, 0.039733887, -0.050933838, 0.031677246, 0.04888916, 0.014312744, -0.041992188, 0.01008606, -0.020187378, 0.04904...
http://arxiv.org/abs/2309.05500v1
NeCo@ALQAC 2023: Legal Domain Knowledge Acquisition for Low-Resource Languages through Data Enrichment
Hai-Long Nguyen, Dieu-Quynh Nguyen, Hoang-Trung Nguyen, Thu-Trang Pham, Huu-Dong Nguyen, Thach-Anh Nguyen, Thi-Hai-Yen Vuong, Ha-Thanh Nguyen
cs.CL, cs.AI
In recent years, natural language processing has gained significant popularity in various sectors, including the legal domain. This paper presents NeCo Team's solutions to the Vietnamese text processing tasks provided in the Automated Legal Question Answering Competition 2023 (ALQAC 2023), focusing on legal domain know...
http://arxiv.org/pdf/2309.05500v1
2023-09-11T14:43:45
2023-09-11T14:43:45
{ "title": [ -0.046722412, -0.028579712, -0.07183838, -0.07397461, -0.015007019, -0.0052337646, -0.066345215, 0.02935791, 0.005695343, 0.025726318, -0.04034424, 0.024246216, -0.00008571148, -0.0181427, -0.02720642, -0.003610611, -0.029403687, 0.0...
http://arxiv.org/abs/2309.05497v1
Personality Detection and Analysis using Twitter Data
Abhilash Datta, Souvic Chakraborty, Animesh Mukherjee
cs.CL, cs.CY
Personality types are important in various fields as they hold relevant information about the characteristics of a human being in an explainable format. They are often good predictors of a person's behaviors in a particular environment and have applications ranging from candidate selection to marketing and mental healt...
http://arxiv.org/pdf/2309.05497v1
2023-09-11T14:39:04
2023-09-11T14:39:04
{ "title": [ -0.0066337585, 0.007507324, -0.033691406, -0.0491333, 0.03277588, -0.0009379387, -0.0055618286, -0.009284973, -0.0056266785, 0.04083252, -0.020324707, 0.013626099, -0.0023078918, -0.0073394775, -0.04611206, 0.012199402, -0.0069351196, ...
http://arxiv.org/abs/2309.05494v2
CrisisTransformers: Pre-trained language models and sentence encoders for crisis-related social media texts
Rabindra Lamsal, Maria Rodriguez Read, Shanika Karunasekera
cs.CL
Social media platforms play an essential role in crisis communication, but analyzing crisis-related social media texts is challenging due to their informal nature. Transformer-based pre-trained models like BERT and RoBERTa have shown success in various NLP tasks, but they are not tailored for crisis-related texts. Furt...
http://arxiv.org/pdf/2309.05494v2
2023-10-01T05:29:11
2023-09-11T14:36:16
{ "title": [ -0.00062036514, 0.009803772, -0.017700195, -0.04714966, -0.007686615, -0.004573822, -0.04220581, -0.01739502, 0.0050621033, 0.029418945, -0.031982422, 0.015197754, 0.026657104, -0.013847351, -0.022949219, -0.0002784729, -0.0054969788, ...
http://arxiv.org/abs/2309.05475v2
Zero-shot Learning with Minimum Instruction to Extract Social Determinants and Family History from Clinical Notes using GPT Model
Neel Bhate, Ansh Mittal, Zhe He, Xiao Luo
cs.CL
Demographics, Social determinants of health, and family history documented in the unstructured text within the electronic health records are increasingly being studied to understand how this information can be utilized with the structured data to improve healthcare outcomes. After the GPT models were released, many stu...
http://arxiv.org/pdf/2309.05475v2
2023-09-13T20:30:22
2023-09-11T14:16:27
{ "title": [ 0.01902771, -0.015655518, -0.018661499, -0.080322266, -0.02507019, -0.002866745, -0.068725586, -0.0008621216, 0.0042648315, -0.00027394295, -0.011444092, 0.0030155182, -0.027526855, -0.04244995, -0.0006508827, -0.018325806, 0.01020813, ...
http://arxiv.org/abs/2309.05472v1
LeBenchmark 2.0: a Standardized, Replicable and Enhanced Framework for Self-supervised Representations of French Speech
Titouan Parcollet, Ha Nguyen, Solene Evain, Marcely Zanon Boito, Adrien Pupier, Salima Mdhaffar, Hang Le, Sina Alisamir, Natalia Tomashenko, Marco Dinarelli, Shucong Zhang, Alexandre Allauzen, Maximin Coavoux, Yannick Esteve, Mickael Rouvier, Jerome Goulian, Benjamin Lecouteux, Francois Portet, Solange Rossato, Fabien ...
cs.CL, cs.AI, cs.SD, eess.AS
Self-supervised learning (SSL) is at the origin of unprecedented improvements in many different domains including computer vision and natural language processing. Speech processing drastically benefitted from SSL as most of the current domain-related tasks are now being approached with pre-trained models. This work int...
http://arxiv.org/pdf/2309.05472v1
2023-09-11T14:13:09
2023-09-11T14:13:09
{ "title": [ 0.017364502, -0.002691269, -0.023483276, -0.07891846, -0.008651733, -0.003730774, -0.050842285, 0.012992859, -0.023864746, 0.034606934, -0.0073661804, -0.011886597, -0.0014877319, -0.039001465, -0.037506104, 0.0029850006, -0.0013418198, ...
http://arxiv.org/abs/2309.05463v1
Textbooks Are All You Need II: phi-1.5 technical report
Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar, Yin Tat Lee
cs.CL, cs.AI
We continue the investigation into the power of smaller Transformer-based language models as initiated by \textbf{TinyStories} -- a 10 million parameter model that can produce coherent English -- and the follow-up work on \textbf{phi-1}, a 1.3 billion parameter model with Python coding performance close to the state-of...
http://arxiv.org/pdf/2309.05463v1
2023-09-11T14:01:45
2023-09-11T14:01:45
{ "title": [ 0.009178162, 0.011711121, -0.019165039, -0.04043579, -0.029159546, -0.004837036, -0.07208252, -0.030303955, -0.038482666, 0.06781006, 0.0016727448, 0.020095825, 0.03451538, -0.023880005, 0.008110046, 0.0012540817, 0.0056266785, 0.000...
http://arxiv.org/abs/2309.05454v2
Flesch or Fumble? Evaluating Readability Standard Alignment of Instruction-Tuned Language Models
Joseph Marvin Imperial, Harish Tayyar Madabushi
cs.CL
Readability metrics and standards such as Flesch Kincaid Grade Level (FKGL) and the Common European Framework of Reference for Languages (CEFR) exist to guide teachers and educators to properly assess the complexity of educational materials before administering them for classroom use. In this study, we select a diverse...
http://arxiv.org/pdf/2309.05454v2
2023-11-03T21:23:06
2023-09-11T13:50:38
{ "title": [ -0.05819702, -0.016326904, -0.03543091, -0.017333984, -0.03604126, -0.011917114, -0.062072754, -0.012138367, -0.028320312, 0.044403076, -0.038757324, 0.005443573, 0.011909485, -0.0033798218, -0.038146973, -0.023803711, -0.015525818, ...
http://arxiv.org/abs/2309.05452v1
Evaluating the Deductive Competence of Large Language Models
S. M. Seals, Valerie L. Shalin
cs.CL
The development of highly fluent large language models (LLMs) has prompted increased interest in assessing their reasoning and problem-solving capabilities. We investigate whether several LLMs can solve a classic type of deductive reasoning problem from the cognitive science literature. The tested LLMs have limited abi...
http://arxiv.org/pdf/2309.05452v1
2023-09-11T13:47:07
2023-09-11T13:47:07
{ "title": [ -0.024032593, -0.0042266846, -0.039916992, -0.081848145, 0.011116028, 0.013458252, -0.06573486, -0.005783081, 0.021102905, 0.012313843, -0.013061523, -0.034057617, -0.0056610107, -0.027740479, -0.05041504, -0.006958008, -0.017669678, ...
http://arxiv.org/abs/2309.05448v1
Panoptic Vision-Language Feature Fields
Haoran Chen, Kenneth Blomqvist, Francesco Milano, Roland Siegwart
cs.CV, cs.AI, cs.CL
Recently, methods have been proposed for 3D open-vocabulary semantic segmentation. Such methods are able to segment scenes into arbitrary classes given at run-time using their text description. In this paper, we propose to our knowledge the first algorithm for open-vocabulary panoptic segmentation, simultaneously perfo...
http://arxiv.org/pdf/2309.05448v1
2023-09-11T13:41:27
2023-09-11T13:41:27
{ "title": [ -0.02368164, -0.0063209534, -0.028823853, -0.03540039, -0.018325806, -0.018814087, -0.03564453, -0.011520386, 0.020858765, 0.0141067505, -0.037231445, -0.03149414, -0.00010108948, -0.04156494, -0.023788452, -0.008811951, 0.02293396, ...
http://arxiv.org/abs/2309.05447v1
TeGit: Generating High-Quality Instruction-Tuning Data with Text-Grounded Task Design
Yongrui Chen, Haiyun Jiang, Xinting Huang, Shuming Shi, Guilin Qi
cs.CL
High-quality instruction-tuning data is critical to improving LLM capabilities. Existing data collection methods are limited by unrealistic manual labeling costs or by the hallucination of relying solely on LLM generation. To address the problems, this paper presents a scalable method to automatically collect high-qual...
http://arxiv.org/pdf/2309.05447v1
2023-09-11T13:41:18
2023-09-11T13:41:18
{ "title": [ -0.06124878, -0.01739502, -0.039215088, -0.020889282, -0.02796936, -0.0017728806, -0.052215576, -0.017852783, -0.008529663, 0.0143585205, -0.019577026, 0.030288696, -0.01890564, -0.008735657, -0.01876831, -0.0060806274, -0.01637268, ...
http://arxiv.org/abs/2309.05444v1
Pushing Mixture of Experts to the Limit: Extremely Parameter Efficient MoE for Instruction Tuning
Ted Zadouri, Ahmet Üstün, Arash Ahmadian, Beyza Ermiş, Acyr Locatelli, Sara Hooker
cs.CL, cs.LG
The Mixture of Experts (MoE) is a widely known neural architecture where an ensemble of specialized sub-models optimizes overall performance with a constant computational cost. However, conventional MoEs pose challenges at scale due to the need to store all experts in memory. In this paper, we push MoE to the limit. We...
http://arxiv.org/pdf/2309.05444v1
2023-09-11T13:31:00
2023-09-11T13:31:00
{ "title": [ -0.053619385, -0.011795044, -0.023269653, -0.07373047, -0.00440979, 0.01574707, -0.021728516, 0.03186035, -0.0053482056, 0.031204224, -0.022598267, 0.025497437, 0.0024299622, -0.011260986, -0.02973938, 0.0043907166, -0.0007557869, 0....
http://arxiv.org/abs/2309.05429v1
Improving Information Extraction on Business Documents with Specific Pre-Training Tasks
Thibault Douzon, Stefan Duffner, Christophe Garcia, Jérémy Espinas
cs.CL, cs.AI, 68T01, I.2.7
Transformer-based Language Models are widely used in Natural Language Processing related tasks. Thanks to their pre-training, they have been successfully adapted to Information Extraction in business documents. However, most pre-training tasks proposed in the literature for business documents are too generic and not su...
http://arxiv.org/pdf/2309.05429v1
2023-09-11T13:05:23
2023-09-11T13:05:23
{ "title": [ -0.017486572, -0.0004274845, -0.051574707, -0.054748535, -0.002729416, -0.022216797, -0.032043457, 0.00869751, -0.018600464, 0.02960205, -0.053741455, 0.019424438, 0.0056648254, 0.007472992, -0.006549835, -0.022537231, -0.00065374374, ...
http://arxiv.org/abs/2309.05423v1
Multi-Modal Automatic Prosody Annotation with Contrastive Pretraining of SSWP
Jinzuomu Zhong, Yang Li, Hui Huang, Jie Liu, Zhiba Su, Jing Guo, Benlai Tang, Fengjie Zhu
eess.AS, cs.AI, cs.CL, cs.SD
In the realm of expressive Text-to-Speech (TTS), explicit prosodic boundaries significantly advance the naturalness and controllability of synthesized speech. While human prosody annotation contributes a lot to the performance, it is a labor-intensive and time-consuming process, often resulting in inconsistent outcomes...
http://arxiv.org/pdf/2309.05423v1
2023-09-11T12:50:28
2023-09-11T12:50:28
{ "title": [ 0.008781433, 0.007671356, -0.019119263, -0.09692383, -0.013809204, 0.008926392, -0.07824707, 0.005443573, 0.018615723, 0.045806885, 0.02003479, 0.0035476685, 0.026977539, -0.04660034, -0.026184082, 0.019729614, -0.0067443848, 0.02378...
http://arxiv.org/abs/2309.05312v1
Experimenting with UD Adaptation of an Unsupervised Rule-based Approach for Sentiment Analysis of Mexican Tourist Texts
Olga Kellert, Mahmud Uz Zaman, Nicholas Hill Matlis, Carlos Gómez-Rodríguez
cs.CL, 68T50, I.2.7
This paper summarizes the results of experimenting with Universal Dependencies (UD) adaptation of an Unsupervised, Compositional and Recursive (UCR) rule-based approach for Sentiment Analysis (SA) submitted to the Shared Task at Rest-Mex 2023 (Team Olga/LyS-SALSA) (within the IberLEF 2023 conference). By using basic sy...
http://arxiv.org/pdf/2309.05312v1
2023-09-11T08:57:02
2023-09-11T08:57:02
{ "title": [ -0.0065612793, -0.012390137, -0.02709961, -0.01965332, -0.0075912476, -0.008728027, -0.0046920776, 0.00021970272, 0.03479004, 0.039764404, -0.056243896, -0.0046539307, -0.010383606, -0.006652832, -0.07476807, -0.029037476, -0.020950317, ...
http://arxiv.org/abs/2309.05311v1
Analysing Cross-Lingual Transfer in Low-Resourced African Named Entity Recognition
Michael Beukman, Manuel Fokam
cs.CL
Transfer learning has led to large gains in performance for nearly all NLP tasks while making downstream models easier and faster to train. This has also been extended to low-resourced languages, with some success. We investigate the properties of cross-lingual transfer learning between ten low-resourced languages, fro...
http://arxiv.org/pdf/2309.05311v1
2023-09-11T08:56:47
2023-09-11T08:56:47
{ "title": [ -0.035095215, -0.012268066, -0.0178833, -0.026855469, -0.0124053955, 0.003030777, -0.03805542, -0.010002136, -0.013679504, 0.035125732, -0.026901245, 0.028564453, 0.038330078, -0.0034866333, -0.021942139, 0.009918213, -0.013259888, 0...
http://arxiv.org/abs/2309.05272v1
Minuteman: Machine and Human Joining Forces in Meeting Summarization
František Kmječ, Ondřej Bojar
cs.CL
Many meetings require creating a meeting summary to keep everyone up to date. Creating minutes of sufficient quality is however very cognitively demanding. Although we currently possess capable models for both audio speech recognition (ASR) and summarization, their fully automatic use is still problematic. ASR models f...
http://arxiv.org/pdf/2309.05272v1
2023-09-11T07:10:47
2023-09-11T07:10:47
{ "title": [ 0.0056991577, 0.008552551, -0.0569458, -0.06222534, -0.008880615, -0.019546509, -0.046813965, -0.01828003, 0.007843018, 0.064331055, -0.01889038, 0.014923096, 0.03704834, -0.0028457642, -0.062164307, 0.006000519, -0.008850098, 0.0396...
http://arxiv.org/abs/2309.05270v2
CONFLATOR: Incorporating Switching Point based Rotatory Positional Encodings for Code-Mixed Language Modeling
Mohsin Ali, Kandukuri Sai Teja, Neeharika Gupta, Parth Patwa, Anubhab Chatterjee, Vinija Jain, Aman Chadha, Amitava Das
cs.CL, cs.LG
The mixing of two or more languages is called Code-Mixing (CM). CM is a social norm in multilingual societies. Neural Language Models (NLMs) like transformers have been effective on many NLP tasks. However, NLM for CM is an under-explored area. Though transformers are capable and powerful, they cannot always encode pos...
http://arxiv.org/pdf/2309.05270v2
2023-10-18T23:48:40
2023-09-11T07:02:13
{ "title": [ -0.04055786, -0.016464233, -0.02027893, -0.0034542084, -0.02482605, 0.0027599335, -0.044403076, -0.011062622, 0.0012693405, 0.055114746, -0.012550354, 0.007232666, 0.03970337, -0.025222778, -0.034423828, 0.012321472, -0.02279663, 0.0...
http://arxiv.org/abs/2309.06557v1
Unsupervised Bias Detection in College Student Newspapers
Adam M. Lehavi, William McCormack, Noah Kornfeld, Solomon Glazer
cs.CL, cs.AI, cs.LG, I.2.7; H.2.8; J.4
This paper presents a pipeline with minimal human influence for scraping and detecting bias on college newspaper archives. This paper introduces a framework for scraping complex archive sites that automated tools fail to grab data from, and subsequently generates a dataset of 14 student papers with 23,154 entries. This...
http://arxiv.org/pdf/2309.06557v1
2023-09-11T06:51:09
2023-09-11T06:51:09
{ "title": [ -0.012550354, 0.013366699, -0.034210205, -0.0072021484, 0.02619934, -0.016983032, 0.018447876, -0.042633057, 0.0028820038, 0.019973755, -0.054534912, 0.0071258545, -0.01374054, 0.00907135, -0.04901123, -0.023376465, -0.03390503, 0.04...
http://arxiv.org/abs/2309.05227v1
Detecting Natural Language Biases with Prompt-based Learning
Md Abdul Aowal, Maliha T Islam, Priyanka Mary Mammen, Sandesh Shetty
cs.CL, cs.AI
In this project, we want to explore the newly emerging field of prompt engineering and apply it to the downstream task of detecting LM biases. More concretely, we explore how to design prompts that can indicate 4 different types of biases: (1) gender, (2) race, (3) sexual orientation, and (4) religion-based. Within our...
http://arxiv.org/pdf/2309.05227v1
2023-09-11T04:20:36
2023-09-11T04:20:36
{ "title": [ -0.005168915, -0.0069351196, -0.042510986, -0.0357666, 0.0014791489, -0.020996094, -0.027832031, -0.02809143, -0.005607605, 0.0014219284, -0.042663574, 0.0065956116, -0.023132324, -0.021469116, -0.019195557, -0.026168823, -0.022750854, ...
http://arxiv.org/abs/2309.05221v1
Exploring the Law of Numbers: Evidence from China's Real Estate
Fuqian Zhang, Zhenhua Wang
cs.CL
The renowned proverb, Numbers do not lie, underscores the reliability and insight that lie beneath numbers, a concept of undisputed importance, especially in economics and finance etc. Despite the prosperity of Benford's Law in the first digit analysis, its scope fails to remain comprehensiveness when it comes to decip...
http://arxiv.org/pdf/2309.05221v1
2023-09-11T03:54:38
2023-09-11T03:54:38
{ "title": [ -0.0015287399, -0.020736694, -0.034301758, -0.0034751892, -0.025344849, -0.03314209, 0.01586914, -0.008651733, -0.03036499, 0.014968872, -0.06707764, 0.010269165, 0.0044441223, -0.024032593, 0.007774353, -0.028747559, -0.0067596436, ...
http://arxiv.org/abs/2309.05217v1
Quantifying and Attributing the Hallucination of Large Language Models via Association Analysis
Li Du, Yequan Wang, Xingrun Xing, Yiqun Ya, Xiang Li, Xin Jiang, Xuezhi Fang
cs.AI, cs.CL
Although demonstrating superb performance on various NLP tasks, large language models (LLMs) still suffer from the hallucination problem, which threatens the reliability of LLMs. To measure the level of hallucination of LLMs, previous works first categorize the hallucination according to the phenomenon similarity, then...
http://arxiv.org/pdf/2309.05217v1
2023-09-11T03:35:00
2023-09-11T03:35:00
{ "title": [ -0.026519775, 0.00078105927, -0.05230713, -0.04434204, -0.004299164, -0.005077362, -0.044036865, -0.011161804, -0.0035858154, 0.031402588, 0.00019693375, 0.0021648407, 0.015235901, -0.016143799, -0.036987305, 0.010536194, -0.02760315, ...
http://arxiv.org/abs/2309.05210v3
Understanding the Impact of Post-Training Quantization on Large Language Models
Somnath Roy
cs.CL
Large language models (LLMs) are rapidly increasing in size, with the number of parameters becoming a key factor in the success of many commercial models, such as ChatGPT, Claude, and Bard. Even the recently released publicly accessible models for commercial usage, such as Falcon and Llama2, come equipped with billions...
http://arxiv.org/pdf/2309.05210v3
2023-09-17T05:38:46
2023-09-11T02:58:32
{ "title": [ -0.021194458, 0.0071411133, -0.04901123, -0.05987549, 0.0115737915, -0.01878357, -0.05569458, -0.004627228, -0.015151978, 0.054718018, 0.008979797, -0.0051116943, 0.015357971, -0.019073486, -0.055023193, 0.013923645, -0.0044784546, 0...
http://arxiv.org/abs/2309.05203v1
From Artificially Real to Real: Leveraging Pseudo Data from Large Language Models for Low-Resource Molecule Discovery
Yuhan Chen, Nuwa Xi, Yanrui Du, Haochun Wang, Chen Jianyu, Sendong Zhao, Bing Qin
cs.CL
Molecule discovery serves as a cornerstone in numerous scientific domains, fueling the development of new materials and innovative drug designs. Recent developments of in-silico molecule discovery have highlighted the promising results of cross-modal techniques, which bridge molecular structures with their descriptive ...
http://arxiv.org/pdf/2309.05203v1
2023-09-11T02:35:36
2023-09-11T02:35:36
{ "title": [ -0.046295166, -0.029647827, -0.054595947, -0.044799805, -0.031707764, -0.008110046, -0.05142212, 0.027267456, -0.0046653748, 0.017623901, -0.0062561035, 0.00012230873, -0.011528015, -0.026504517, -0.028411865, -0.0032291412, 0.0020923615...
http://arxiv.org/abs/2309.05201v1
Two is Better Than One: Answering Complex Questions by Multiple Knowledge Sources with Generalized Links
Minhao Zhang, Yongliang Ma, Yanzeng Li, Ruoyu Zhang, Lei Zou, Ming Zhou
cs.CL
Incorporating multiple knowledge sources is proven to be beneficial for answering complex factoid questions. To utilize multiple knowledge bases (KB), previous works merge all KBs into a single graph via entity alignment and reduce the problem to question-answering (QA) over the fused KB. In reality, various link relat...
http://arxiv.org/pdf/2309.05201v1
2023-09-11T02:31:41
2023-09-11T02:31:41
{ "title": [ -0.002811432, -0.010627747, -0.054534912, -0.09057617, 0.017440796, -0.012931824, -0.029067993, 0.009254456, 0.006198883, 0.007835388, -0.0026435852, 0.020812988, -0.028411865, -0.022064209, -0.027191162, 0.0018386841, 0.013137817, 0...
http://arxiv.org/abs/2309.05196v1
Does Writing with Language Models Reduce Content Diversity?
Vishakh Padmakumar, He He
cs.CL, cs.CY, cs.HC, cs.LG
Large language models (LLMs) have led to a surge in collaborative writing with model assistance. As different users incorporate suggestions from the same model, there is a risk of decreased diversity in the produced content, potentially limiting diverse perspectives in public discourse. In this work, we measure the imp...
http://arxiv.org/pdf/2309.05196v1
2023-09-11T02:16:47
2023-09-11T02:16:47
{ "title": [ -0.04550171, 0.025375366, -0.0574646, 0.01473999, -0.021255493, -0.021743774, -0.04071045, -0.02407837, -0.024993896, 0.049194336, 0.0028133392, -0.0022659302, 0.020523071, -0.023162842, -0.012138367, 0.0017681122, -0.009269714, 0.03...
http://arxiv.org/abs/2309.05173v2
DePT: Decomposed Prompt Tuning for Parameter-Efficient Fine-tuning
Zhengxiang Shi, Aldo Lipani
cs.CL, cs.AI, cs.CV, cs.LG
Prompt tuning (PT), where a small amount of trainable soft (continuous) prompt vectors is affixed to the input of language models (LM), has shown promising results across various tasks and models for parameter-efficient fine-tuning (PEFT). PT stands out from other PEFT approaches because it maintains competitive perfor...
http://arxiv.org/pdf/2309.05173v2
2023-10-12T17:25:44
2023-09-11T00:02:05
{ "title": [ -0.005241394, -0.02848816, -0.016326904, -0.10546875, -0.011100769, 0.006965637, -0.061065674, 0.018295288, -0.043670654, 0.016830444, 0.01663208, 0.0026664734, -0.0042419434, -0.04815674, 0.0023555756, 0.00082731247, 0.007358551, 0....
http://arxiv.org/abs/2309.05162v1
Collecting Visually-Grounded Dialogue with A Game Of Sorts
Bram Willemsen, Dmytro Kalpakchi, Gabriel Skantze
cs.CL, cs.AI, cs.CV
An idealized, though simplistic, view of the referring expression production and grounding process in (situated) dialogue assumes that a speaker must merely appropriately specify their expression so that the target referent may be successfully identified by the addressee. However, referring in conversation is a collabo...
http://arxiv.org/pdf/2309.05162v1
2023-09-10T23:00:35
2023-09-10T23:00:35
{ "title": [ 0.030075073, -0.02357483, -0.031829834, -0.021240234, -0.045135498, -0.024932861, 0.009857178, 0.008399963, -0.0056915283, -0.012237549, -0.020767212, 0.003917694, -0.05117798, -0.046783447, 0.009178162, -0.039855957, 0.024978638, 0....
http://arxiv.org/abs/2309.05142v1
Large Language Models for Difficulty Estimation of Foreign Language Content with Application to Language Learning
Michalis Vlachos, Mircea Lungu, Yash Raj Shrestha, Johannes-Rudolf David
cs.CL, cs.AI
We use large language models to aid learners enhance proficiency in a foreign language. This is accomplished by identifying content on topics that the user is interested in, and that closely align with the learner's proficiency level in that foreign language. Our work centers on French content, but our approach is read...
http://arxiv.org/pdf/2309.05142v1
2023-09-10T21:23:09
2023-09-10T21:23:09
{ "title": [ -0.024047852, -0.0012483597, -0.015701294, -0.011047363, -0.008796692, -0.006893158, -0.019180298, -0.024978638, 0.009681702, 0.013420105, -0.01878357, -0.0011005402, -0.01914978, -0.022918701, -0.046081543, -0.0039405823, -0.032318115, ...
http://arxiv.org/abs/2309.05103v1
AGent: A Novel Pipeline for Automatically Creating Unanswerable Questions
Son Quoc Tran, Gia-Huy Do, Phong Nguyen-Thuan Do, Matt Kretchmar, Xinya Du
cs.CL, cs.AI
The development of large high-quality datasets and high-performing models have led to significant advancements in the domain of Extractive Question Answering (EQA). This progress has sparked considerable interest in exploring unanswerable questions within the EQA domain. Training EQA models with unanswerable questions ...
http://arxiv.org/pdf/2309.05103v1
2023-09-10T18:13:11
2023-09-10T18:13:11
{ "title": [ -0.0055999756, -0.012435913, -0.029953003, -0.036499023, -0.017807007, 0.0046577454, -0.015823364, -0.02355957, 0.014007568, 0.0051727295, -0.027526855, 0.0039138794, -0.031082153, -0.01979065, -0.038085938, -0.0053520203, -0.0009918213,...
http://arxiv.org/abs/2309.05086v2
Neural-Hidden-CRF: A Robust Weakly-Supervised Sequence Labeler
Zhijun Chen, Hailong Sun, Wanhao Zhang, Chunyi Xu, Qianren Mao, Pengpeng Chen
cs.CL, cs.AI
We propose a neuralized undirected graphical model called Neural-Hidden-CRF to solve the weakly-supervised sequence labeling problem. Under the umbrella of probabilistic undirected graph theory, the proposed Neural-Hidden-CRF embedded with a hidden CRF layer models the variables of word sequence, latent ground truth se...
http://arxiv.org/pdf/2309.05086v2
2023-09-28T19:44:09
2023-09-10T17:13:25
{ "title": [ -0.007949829, 0.011039734, -0.057495117, -0.070739746, 0.018417358, -0.01991272, -0.042907715, -0.012672424, 0.0042304993, 0.029067993, -0.028900146, 0.02418518, 0.0016870499, -0.010513306, -0.023239136, 0.0041923523, -0.0073509216, ...
http://arxiv.org/abs/2309.05076v1
An Appraisal-Based Chain-Of-Emotion Architecture for Affective Language Model Game Agents
Maximilian Croissant, Madeleine Frister, Guy Schofield, Cade McCall
cs.AI, cs.CL, cs.HC
The development of believable, natural, and interactive digital artificial agents is a field of growing interest. Theoretical uncertainties and technical barriers present considerable challenges to the field, particularly with regards to developing agents that effectively simulate human emotions. Large language models ...
http://arxiv.org/pdf/2309.05076v1
2023-09-10T16:55:49
2023-09-10T16:55:49
{ "title": [ -0.011451721, -0.005493164, -0.012168884, -0.019058228, -0.017120361, -0.0079574585, 0.0063323975, -0.012550354, 0.02909851, 0.05117798, -0.024993896, -0.019958496, 0.023651123, -0.019042969, -0.06829834, 0.003522873, -0.019210815, -...
http://arxiv.org/abs/2309.05044v1
The Effect of Alignment Objectives on Code-Switching Translation
Mohamed Anwar
cs.CL
One of the things that need to change when it comes to machine translation is the models' ability to translate code-switching content, especially with the rise of social media and user-generated content. In this paper, we are proposing a way of training a single machine translation model that is able to translate monol...
http://arxiv.org/pdf/2309.05044v1
2023-09-10T14:46:31
2023-09-10T14:46:31
{ "title": [ -0.05807495, 0.00071430206, -0.031158447, 0.007911682, -0.01751709, -0.009246826, -0.013702393, -0.014846802, -0.024673462, 0.06427002, 0.005897522, 0.0071487427, 0.022216797, -0.0289917, -0.04598999, 0.010826111, -0.008872986, 0.013...
http://arxiv.org/abs/2309.05021v1
Chat2Brain: A Method for Mapping Open-Ended Semantic Queries to Brain Activation Maps
Yaonai Wei, Tuo Zhang, Han Zhang, Tianyang Zhong, Lin Zhao, Zhengliang Liu, Chong Ma, Songyao Zhang, Muheng Shang, Lei Du, Xiao Li, Tianming Liu, Junwei Han
cs.CL
Over decades, neuroscience has accumulated a wealth of research results in the text modality that can be used to explore cognitive processes. Meta-analysis is a typical method that successfully establishes a link from text queries to brain activation maps using these research results, but it still relies on an ideal qu...
http://arxiv.org/pdf/2309.05021v1
2023-09-10T13:06:45
2023-09-10T13:06:45
{ "title": [ 0.0036239624, 0.011543274, -0.013191223, -0.024765015, 0.005756378, -0.011116028, -0.01638794, -0.04837036, -0.0014190674, 0.03567505, 0.015296936, 0.0010185242, -0.007019043, -0.033050537, -0.042785645, 0.016693115, 0.009346008, 0.0...
http://arxiv.org/abs/2309.05007v2
FOLLOWUPQG: Towards Information-Seeking Follow-up Question Generation
Yan Meng, Liangming Pan, Yixin Cao, Min-Yen Kan
cs.CL, cs.AI
Humans ask follow-up questions driven by curiosity, which reflects a creative human cognitive process. We introduce the task of real-world information-seeking follow-up question generation (FQG), which aims to generate follow-up questions seeking a more in-depth understanding of an initial question and answer. We const...
http://arxiv.org/pdf/2309.05007v2
2023-09-19T07:51:03
2023-09-10T11:58:29
{ "title": [ -0.0028743744, -0.0075531006, -0.023223877, -0.059417725, -0.0093307495, -0.0044937134, -0.022216797, -0.010131836, 0.018234253, 0.03652954, -0.004917145, 0.024276733, -0.02796936, -0.0149383545, -0.06744385, 0.0129776, -0.0027580261, ...
http://arxiv.org/abs/2309.06527v1
Machine Translation Models Stand Strong in the Face of Adversarial Attacks
Pavel Burnyshev, Elizaveta Kostenok, Alexey Zaytsev
cs.CL, cs.CR, cs.LG
Adversarial attacks expose vulnerabilities of deep learning models by introducing minor perturbations to the input, which lead to substantial alterations in the output. Our research focuses on the impact of such adversarial attacks on sequence-to-sequence (seq2seq) models, specifically machine translation models. We in...
http://arxiv.org/pdf/2309.06527v1
2023-09-10T11:22:59
2023-09-10T11:22:59
{ "title": [ 0.022644043, -0.0077323914, -0.00087594986, -0.05142212, -0.02671814, -0.015716553, -0.025604248, -0.012969971, -0.022750854, 0.05569458, 0.02798462, 0.002231598, 0.03378296, -0.046966553, -0.01725769, 0.015419006, -0.0076408386, 0.0...
http://arxiv.org/abs/2309.04992v1
Mitigating Word Bias in Zero-shot Prompt-based Classifiers
Adian Liusie, Potsawee Manakul, Mark J. F. Gales
cs.CL
Prompt-based classifiers are an attractive approach for zero-shot classification. However, the precise choice of the prompt template and label words can largely influence performance, with semantically equivalent settings often showing notable performance difference. This discrepancy can be partly attributed to word bi...
http://arxiv.org/pdf/2309.04992v1
2023-09-10T10:57:41
2023-09-10T10:57:41
{ "title": [ -0.0030021667, -0.0066719055, -0.033325195, -0.018432617, -0.006214142, -0.025466919, -0.014099121, -0.027664185, 0.022720337, 0.018814087, -0.032470703, 0.012107849, -0.017807007, -0.0121154785, -0.040496826, -0.007904053, -0.008903503,...
http://arxiv.org/abs/2309.04979v1
Retrieval-Augmented Meta Learning for Low-Resource Text Classification
Rongsheng Li, Yangning Li, Yinghui Li, Chaiyut Luoyiching, Hai-Tao Zheng, Nannan Zhou, Hanjing Su
cs.CL
Meta learning have achieved promising performance in low-resource text classification which aims to identify target classes with knowledge transferred from source classes with sets of small tasks named episodes. However, due to the limited training data in the meta-learning scenario and the inherent properties of param...
http://arxiv.org/pdf/2309.04979v1
2023-09-10T10:05:03
2023-09-10T10:05:03
{ "title": [ -0.011878967, -0.0016508102, -0.025054932, -0.045562744, 0.0032844543, 0.0018625259, -0.030776978, -0.027053833, 0.017089844, 0.049316406, -0.02017212, 0.01449585, 0.0003504753, -0.0033473969, -0.06774902, 0.0032424927, -0.013679504, ...
http://arxiv.org/abs/2309.04977v1
RGAT: A Deeper Look into Syntactic Dependency Information for Coreference Resolution
Yuan Meng, Xuhao Pan, Jun Chang, Yue Wang
cs.CL, cs.AI, 14J60 (Primary) 14F05, 14J26 (Secondary)
Although syntactic information is beneficial for many NLP tasks, combining it with contextual information between words to solve the coreference resolution problem needs to be further explored. In this paper, we propose an end-to-end parser that combines pre-trained BERT with a Syntactic Relation Graph Attention Networ...
http://arxiv.org/pdf/2309.04977v1
2023-09-10T09:46:38
2023-09-10T09:46:38
{ "title": [ 0.005809784, -0.014060974, -0.027694702, -0.110961914, 0.007789612, 0.011245728, -0.054260254, 0.013557434, -0.0032997131, 0.03463745, -0.010543823, 0.038330078, 0.01739502, -0.0069389343, -0.031433105, 0.0234375, -0.012542725, 0.040...
http://arxiv.org/abs/2309.04971v1
Prompt Learning With Knowledge Memorizing Prototypes For Generalized Few-Shot Intent Detection
Chaiyut Luoyiching, Yangning Li, Yinghui Li, Rongsheng Li, Hai-Tao Zheng, Nannan Zhou, Hanjing Su
cs.CL
Generalized Few-Shot Intent Detection (GFSID) is challenging and realistic because it needs to categorize both seen and novel intents simultaneously. Previous GFSID methods rely on the episodic learning paradigm, which makes it hard to extend to a generalized setup as they do not explicitly learn the classification of ...
http://arxiv.org/pdf/2309.04971v1
2023-09-10T09:16:38
2023-09-10T09:16:38
{ "title": [ 0.034210205, -0.0017871857, -0.034973145, -0.028945923, 0.017150879, -0.034301758, 0.017532349, 0.0004787445, 0.014671326, 0.0033035278, -0.046142578, 0.011184692, -0.02053833, -0.0256958, -0.028808594, -0.026046753, 0.018722534, 0.0...
http://arxiv.org/abs/2309.04965v2
Prefix-diffusion: A Lightweight Diffusion Model for Diverse Image Captioning
Guisheng Liu, Yi Li, Zhengcong Fei, Haiyan Fu, Xiangyang Luo, Yanqing Guo
cs.CV, cs.AI, cs.CL
While impressive performance has been achieved in image captioning, the limited diversity of the generated captions and the large parameter scale remain major barriers to the real-word application of these systems. In this work, we propose a lightweight image captioning network in combination with continuous diffusion,...
http://arxiv.org/pdf/2309.04965v2
2023-10-17T01:30:57
2023-09-10T08:55:24
{ "title": [ 0.00032901764, 0.0019073486, -0.03604126, -0.040924072, 0.013252258, -0.013710022, 0.006969452, -0.0036621094, 0.013427734, 0.04486084, -0.014671326, 0.0289917, -0.013725281, -0.004638672, -0.049804688, 0.001786232, -0.022201538, 0.0...
http://arxiv.org/abs/2309.04951v2
Multi-document Summarization: A Comparative Evaluation
Kushan Hewapathirana, Nisansa de Silva, C. D. Athuraliya
cs.CL, cs.AI
This paper is aimed at evaluating state-of-the-art models for Multi-document Summarization (MDS) on different types of datasets in various domains and investigating the limitations of existing models to determine future research directions. To address this gap, we conducted an extensive literature review to identify st...
http://arxiv.org/pdf/2309.04951v2
2023-09-12T04:19:49
2023-09-10T07:43:42
{ "title": [ -0.010940552, 0.0010080338, -0.026290894, -0.07965088, 0.0068626404, 0.013923645, -0.054138184, -0.0068130493, -0.00037264824, 0.056243896, -0.010475159, 0.02583313, -0.0076904297, 0.01020813, -0.059661865, 0.007904053, -0.011421204, ...
http://arxiv.org/abs/2309.04940v1
What's Hard in English RST Parsing? Predictive Models for Error Analysis
Yang Janet Liu, Tatsuya Aoyama, Amir Zeldes
cs.CL
Despite recent advances in Natural Language Processing (NLP), hierarchical discourse parsing in the framework of Rhetorical Structure Theory remains challenging, and our understanding of the reasons for this are as yet limited. In this paper, we examine and model some of the factors associated with parsing difficulties...
http://arxiv.org/pdf/2309.04940v1
2023-09-10T06:10:03
2023-09-10T06:10:03
{ "title": [ -0.023468018, -0.002954483, 0.0005245209, -0.053894043, -0.018585205, 0.025878906, -0.07098389, -0.03277588, -0.012901306, 0.02508545, -0.023147583, 0.016998291, -0.025848389, -0.025558472, -0.043548584, 0.00018334389, -0.015853882, ...
http://arxiv.org/abs/2309.04919v1
Unsupervised Chunking with Hierarchical RNN
Zijun Wu, Anup Anand Deshmukh, Yongkang Wu, Jimmy Lin, Lili Mou
cs.CL
In Natural Language Processing (NLP), predicting linguistic structures, such as parsing and chunking, has mostly relied on manual annotations of syntactic structures. This paper introduces an unsupervised approach to chunking, a syntactic task that involves grouping words in a non-hierarchical manner. We present a two-...
http://arxiv.org/pdf/2309.04919v1
2023-09-10T02:55:12
2023-09-10T02:55:12
{ "title": [ -0.033599854, 0.007675171, -0.07849121, -0.0960083, 0.013458252, 0.0044059753, -0.0736084, 0.019927979, -0.0143585205, 0.0413208, -0.025222778, 0.012786865, -0.00039863586, -0.012931824, -0.03463745, 0.012687683, 0.0073890686, 0.0300...
http://arxiv.org/abs/2309.11508v1
Towards LLM-based Autograding for Short Textual Answers
Johannes Schneider, Bernd Schenk, Christina Niklaus, Michaelis Vlachos
cs.CL, cs.AI
Grading of exams is an important, labor intensive, subjective, repetitive and frequently challenging task. The feasibility of autograding textual responses has greatly increased thanks to the availability of large language models (LLMs) such as ChatGPT and because of the substantial influx of data brought about by digi...
http://arxiv.org/pdf/2309.11508v1
2023-09-09T22:25:56
2023-09-09T22:25:56
{ "title": [ -0.046417236, -0.010505676, -0.03213501, -0.038269043, -0.038146973, 0.019729614, -0.06915283, 0.0028266907, 0.01637268, 0.045837402, -0.026290894, 0.030715942, -0.0034599304, -0.00096321106, -0.04864502, 0.012290955, -0.0072021484, ...
http://arxiv.org/abs/2309.04862v1
Distributional Data Augmentation Methods for Low Resource Language
Mosleh Mahamud, Zed Lee, Isak Samsten
cs.CL, cs.AI
Text augmentation is a technique for constructing synthetic data from an under-resourced corpus to improve predictive performance. Synthetic data generation is common in numerous domains. However, recently text augmentation has emerged in natural language processing (NLP) to improve downstream tasks. One of the current...
http://arxiv.org/pdf/2309.04862v1
2023-09-09T19:01:59
2023-09-09T19:01:59
{ "title": [ -0.02142334, -0.0064315796, -0.053100586, -0.066833496, 0.004585266, 0.0017147064, -0.06695557, 0.008171082, 0.025863647, 0.03967285, -0.017196655, -0.0074501038, 0.021224976, -0.02609253, -0.044311523, 0.027374268, -0.00894165, 0.05...
http://arxiv.org/abs/2309.04858v1
Reverse-Engineering Decoding Strategies Given Blackbox Access to a Language Generation System
Daphne Ippolito, Nicholas Carlini, Katherine Lee, Milad Nasr, Yun William Yu
cs.LG, cs.CL, cs.CR
Neural language models are increasingly deployed into APIs and websites that allow a user to pass in a prompt and receive generated text. Many of these systems do not reveal generation parameters. In this paper, we present methods to reverse-engineer the decoding method used to generate text (i.e., top-$k$ or nucleus s...
http://arxiv.org/pdf/2309.04858v1
2023-09-09T18:19:47
2023-09-09T18:19:47
{ "title": [ -0.03488159, -0.010360718, -0.027236938, -0.030075073, -0.044403076, -0.010147095, -0.049072266, -0.0017194748, -0.007091522, 0.046691895, -0.014305115, 0.021881104, 0.010498047, -0.011436462, -0.028762817, 0.014320374, 0.008918762, ...
http://arxiv.org/abs/2309.04849v1
Speech Emotion Recognition with Distilled Prosodic and Linguistic Affect Representations
Debaditya Shome, Ali Etemad
cs.CL, cs.AI, cs.LG
We propose EmoDistill, a novel speech emotion recognition (SER) framework that leverages cross-modal knowledge distillation during training to learn strong linguistic and prosodic representations of emotion from speech. During inference, our method only uses a stream of speech signals to perform unimodal SER thus reduc...
http://arxiv.org/pdf/2309.04849v1
2023-09-09T17:30:35
2023-09-09T17:30:35
{ "title": [ 0.0020618439, 0.0053482056, -0.039520264, -0.07312012, 0.00076436996, -0.011550903, -0.044891357, 0.006374359, 0.04071045, 0.04031372, -0.02268982, -0.024520874, 0.027633667, -0.027145386, -0.056671143, 0.0019054413, -0.0045394897, 0...
http://arxiv.org/abs/2309.04842v2
Leveraging Large Language Models for Exploiting ASR Uncertainty
Pranay Dighe, Yi Su, Shangshang Zheng, Yunshu Liu, Vineet Garg, Xiaochuan Niu, Ahmed Tewfik
cs.CL, cs.HC, cs.SD, eess.AS
While large language models excel in a variety of natural language processing (NLP) tasks, to perform well on spoken language understanding (SLU) tasks, they must either rely on off-the-shelf automatic speech recognition (ASR) systems for transcription, or be equipped with an in-built speech modality. This work focuses...
http://arxiv.org/pdf/2309.04842v2
2023-09-12T16:46:26
2023-09-09T17:02:33
{ "title": [ -0.028045654, -0.00705719, -0.06365967, -0.07745361, -0.0015449524, -0.005710602, -0.07800293, 0.020858765, -0.005001068, 0.031402588, -0.018753052, -0.01751709, 0.009597778, -0.024917603, -0.042022705, 0.0049705505, 0.002609253, 0.0...
http://arxiv.org/abs/2309.04827v1
Neurons in Large Language Models: Dead, N-gram, Positional
Elena Voita, Javier Ferrando, Christoforos Nalmpantis
cs.CL
We analyze a family of large language models in such a lightweight manner that can be done on a single GPU. Specifically, we focus on the OPT family of models ranging from 125m to 66b parameters and rely only on whether an FFN neuron is activated or not. First, we find that the early part of the network is sparse and r...
http://arxiv.org/pdf/2309.04827v1
2023-09-09T15:51:36
2023-09-09T15:51:36
{ "title": [ 0.0026950836, 0.0053977966, -0.014701843, -0.001578331, -0.028442383, -0.030212402, -0.024169922, -0.037597656, 0.009590149, 0.05130005, -0.032196045, -0.019760132, 0.043518066, -0.021392822, -0.036010742, -0.0014076233, -0.009208679, ...
http://arxiv.org/abs/2309.04823v1
FaNS: a Facet-based Narrative Similarity Metric
Mousumi Akter, Shubhra Kanti Karmaker Santu
cs.CL
Similar Narrative Retrieval is a crucial task since narratives are essential for explaining and understanding events, and multiple related narratives often help to create a holistic view of the event of interest. To accurately identify semantically similar narratives, this paper proposes a novel narrative similarity me...
http://arxiv.org/pdf/2309.04823v1
2023-09-09T15:29:24
2023-09-09T15:29:24
{ "title": [ -0.037963867, 0.0036735535, -0.03010559, -0.04159546, 0.001200676, 0.012664795, -0.03918457, -0.030075073, 0.0014219284, 0.028152466, -0.011123657, 0.04324341, -0.004310608, -0.008804321, -0.018508911, 0.017318726, -0.01309967, 0.017...
http://arxiv.org/abs/2309.04790v1
MMHQA-ICL: Multimodal In-context Learning for Hybrid Question Answering over Text, Tables and Images
Weihao Liu, Fangyu Lei, Tongxu Luo, Jiahe Lei, Shizhu He, Jun Zhao, Kang Liu
cs.CL
In the real world, knowledge often exists in a multimodal and heterogeneous form. Addressing the task of question answering with hybrid data types, including text, tables, and images, is a challenging task (MMHQA). Recently, with the rise of large language models (LLM), in-context learning (ICL) has become the most pop...
http://arxiv.org/pdf/2309.04790v1
2023-09-09T13:35:01
2023-09-09T13:35:01
{ "title": [ -0.006996155, 0.009681702, -0.045410156, -0.031982422, 0.0066184998, -0.024291992, -0.0039405823, -0.0012807846, 0.011329651, 0.040100098, -0.04067993, 0.010398865, 0.020339966, 0.0020503998, -0.045410156, -0.0209198, -0.008735657, 0...
http://arxiv.org/abs/2309.04766v1
SeaEval for Multilingual Foundation Models: From Cross-Lingual Alignment to Cultural Reasoning
Bin Wang, Zhengyuan Liu, Xin Huang, Fangkai Jiao, Yang Ding, Ai Ti Aw, Nancy F. Chen
cs.CL, cs.AI
We present SeaEval, a benchmark for multilingual foundation models. In addition to characterizing how these models understand and reason with natural language, we also investigate how well they comprehend cultural practices, nuances, and values. Alongside standard accuracy metrics, we investigate the brittleness of fou...
http://arxiv.org/pdf/2309.04766v1
2023-09-09T11:42:22
2023-09-09T11:42:22
{ "title": [ -0.02758789, -0.009803772, -0.016662598, -0.051879883, -0.010612488, 0.008766174, -0.07098389, -0.025634766, 0.0012960434, 0.030227661, 0.0037956238, -0.014907837, 0.02027893, -0.046295166, -0.016601562, 0.0064353943, -0.015670776, 0...
http://arxiv.org/abs/2309.04739v1
Data Augmentation for Conversational AI
Heydar Soudani, Evangelos Kanoulas, Faegheh Hasibi
cs.CL, cs.IR
Advancements in conversational systems have revolutionized information access, surpassing the limitations of single queries. However, developing dialogue systems requires a large amount of training data, which is a challenge in low-resource domains and languages. Traditional data collection methods like crowd-sourcing ...
http://arxiv.org/pdf/2309.04739v1
2023-09-09T09:56:35
2023-09-09T09:56:35
{ "title": [ 0.00028443336, 0.0013694763, -0.05923462, -0.056427002, -0.0005583763, -0.023544312, -0.022537231, 0.01361084, 0.00831604, 0.03390503, -0.046569824, 0.0066719055, 0.0079193115, -0.014511108, -0.030273438, 0.0019321442, -0.00005477667, ...
http://arxiv.org/abs/2309.04734v1
Towards Better Multi-modal Keyphrase Generation via Visual Entity Enhancement and Multi-granularity Image Noise Filtering
Yifan Dong, Suhang Wu, Fandong Meng, Jie Zhou, Xiaoli Wang, Jianxin Lin, Jinsong Su
cs.CV, cs.CL, cs.MM
Multi-modal keyphrase generation aims to produce a set of keyphrases that represent the core points of the input text-image pair. In this regard, dominant methods mainly focus on multi-modal fusion for keyphrase generation. Nevertheless, there are still two main drawbacks: 1) only a limited number of sources, such as i...
http://arxiv.org/pdf/2309.04734v1
2023-09-09T09:41:36
2023-09-09T09:41:36
{ "title": [ -0.006225586, 0.008361816, -0.028762817, -0.04324341, -0.007129669, -0.0020885468, -0.049957275, -0.022094727, 0.026931763, 0.057159424, 0.0074005127, 0.015823364, 0.05419922, -0.024398804, -0.037017822, 0.027420044, -0.022857666, 0....
http://arxiv.org/abs/2309.04725v1
EPA: Easy Prompt Augmentation on Large Language Models via Multiple Sources and Multiple Targets
Hongyuan Lu, Wai Lam
cs.CL
Large language models (LLMs) have shown promising performance on various NLP tasks via task prompting. And their performance can be further improved by appending task demonstrations to the head of the prompt. And usually, a better performance can be achieved with more demonstrations. However, asking the users to write ...
http://arxiv.org/pdf/2309.04725v1
2023-09-09T09:03:50
2023-09-09T09:03:50
{ "title": [ -0.002298355, -0.009750366, -0.050109863, -0.08001709, 0.013114929, -0.002773285, -0.06427002, 0.0062789917, 0.0015220642, 0.0066108704, -0.018478394, 0.015083313, 0.0079956055, -0.029830933, -0.0041122437, 0.02722168, -0.011924744, ...
http://arxiv.org/abs/2309.04716v1
Toward Reproducing Network Research Results Using Large Language Models
Qiao Xiang, Yuling Lin, Mingjun Fang, Bang Huang, Siyong Huang, Ridi Wen, Franck Le, Linghe Kong, Jiwu Shu
cs.LG, cs.AI, cs.CL
Reproducing research results in the networking community is important for both academia and industry. The current best practice typically resorts to three approaches: (1) looking for publicly available prototypes; (2) contacting the authors to get a private prototype; and (3) manually implementing a prototype following...
http://arxiv.org/pdf/2309.04716v1
2023-09-09T08:07:54
2023-09-09T08:07:54
{ "title": [ -0.053863525, -0.023483276, -0.060577393, -0.03982544, -0.012237549, 0.00035858154, -0.0713501, 0.0005507469, -0.0051956177, 0.03479004, -0.0010280609, -0.0012664795, 0.0011177063, -0.026779175, -0.04425049, 0.021118164, -0.013641357, ...
http://arxiv.org/abs/2309.04704v1
Analysis of Disinformation and Fake News Detection Using Fine-Tuned Large Language Model
Bohdan M. Pavlyshenko
cs.CL, cs.AI, cs.CY, cs.IR, cs.LG
The paper considers the possibility of fine-tuning Llama 2 large language model (LLM) for the disinformation analysis and fake news detection. For fine-tuning, the PEFT/LoRA based approach was used. In the study, the model was fine-tuned for the following tasks: analysing a text on revealing disinformation and propagan...
http://arxiv.org/pdf/2309.04704v1
2023-09-09T07:10:19
2023-09-09T07:10:19
{ "title": [ -0.005531311, 0.0055999756, -0.043273926, -0.04663086, 0.016952515, -0.012458801, -0.016189575, -0.01146698, -0.025405884, 0.0259552, -0.024017334, 0.008491516, 0.011550903, -0.013633728, -0.025604248, -0.015899658, -0.034851074, 0.0...
http://arxiv.org/abs/2309.04695v1
Code-Style In-Context Learning for Knowledge-Based Question Answering
Zhijie Nie, Richong Zhang, Zhongyuan Wang, Xudong Liu
cs.CL, cs.AI
Current methods for Knowledge-Based Question Answering (KBQA) usually rely on complex training techniques and model frameworks, leading to many limitations in practical applications. Recently, the emergence of In-Context Learning (ICL) capabilities in Large Language Models (LLMs) provides a simple and training-free sem...
http://arxiv.org/pdf/2309.04695v1
2023-09-09T06:27:00
2023-09-09T06:27:00
{ "title": [ -0.00409317, -0.020019531, -0.028198242, -0.048095703, -0.0045776367, -0.018005371, -0.00023281574, 0.020309448, 0.019866943, 0.039611816, -0.04019165, 0.0030136108, 0.0009217262, -0.0015621185, -0.05203247, -0.005592346, 0.010177612, ...
http://arxiv.org/abs/2309.04679v2
Embedding structure matters: Comparing methods to adapt multilingual vocabularies to new languages
C. M. Downey, Terra Blevins, Nora Goldfine, Shane Steinert-Threlkeld
cs.CL
Pre-trained multilingual language models underpin a large portion of modern NLP tools outside of English. A strong baseline for specializing these models for specific languages is Language-Adaptive Pre-Training (LAPT). However, retaining a large cross-lingual vocabulary and embedding matrix comes at considerable excess...
http://arxiv.org/pdf/2309.04679v2
2023-10-26T21:26:16
2023-09-09T04:27:18
{ "title": [ -0.02923584, -0.01474762, -0.020889282, -0.032196045, -0.009895325, -0.00019109249, -0.02482605, -0.0044136047, -0.013755798, 0.023223877, -0.03869629, 0.024978638, -0.009284973, -0.0079956055, -0.01751709, -0.020370483, -0.0067863464, ...
http://arxiv.org/abs/2309.04663v2
FIAT: Fusing learning paradigms with Instruction-Accelerated Tuning
Xinyi Wang, John Wieting, Jonathan H. Clark
cs.CL, cs.AI
Learning paradigms for large language models (LLMs) currently tend to fall within either in-context learning (ICL) or full fine-tuning. Each of these comes with their own trade-offs based on available data, model size, compute cost, ease-of-use, and final quality with neither solution performing well across-the-board. ...
http://arxiv.org/pdf/2309.04663v2
2023-09-12T14:34:03
2023-09-09T02:43:48
{ "title": [ -0.051116943, -0.01965332, -0.044647217, -0.039093018, -0.009819031, -0.010360718, -0.019836426, -0.0006170273, -0.015090942, 0.023727417, -0.017532349, 0.019699097, -0.021621704, -0.02243042, -0.026397705, -0.0021820068, -0.008598328, ...
http://arxiv.org/abs/2309.04662v1
MADLAD-400: A Multilingual And Document-Level Large Audited Dataset
Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat
cs.CL, cs.LG
We introduce MADLAD-400, a manually audited, general domain 3T token monolingual dataset based on CommonCrawl, spanning 419 languages. We discuss the limitations revealed by self-auditing MADLAD-400, and the role data auditing had in the dataset creation process. We then train and release a 10.7B-parameter multilingual...
http://arxiv.org/pdf/2309.04662v1
2023-09-09T02:34:01
2023-09-09T02:34:01
{ "title": [ -0.03265381, -0.02432251, -0.068481445, -0.08068848, 0.009536743, -0.00051927567, -0.06591797, 0.007144928, 0.0044517517, 0.02986145, -0.011543274, 0.007381439, -0.010322571, -0.013824463, -0.030029297, 0.011070251, -0.008850098, 0.0...
http://arxiv.org/abs/2309.04658v1
Exploring Large Language Models for Communication Games: An Empirical Study on Werewolf
Yuzhuang Xu, Shuo Wang, Peng Li, Fuwen Luo, Xiaolong Wang, Weidong Liu, Yang Liu
cs.CL
Communication games, which we refer to as incomplete information games that heavily depend on natural language communication, hold significant research value in fields such as economics, social science, and artificial intelligence. In this work, we explore the problem of how to engage large language models (LLMs) in co...
http://arxiv.org/pdf/2309.04658v1
2023-09-09T01:56:40
2023-09-09T01:56:40
{ "title": [ -0.037078857, -0.0063476562, -0.022384644, -0.0031337738, -0.020690918, -0.014480591, -0.0262146, -0.019500732, -0.012588501, 0.02885437, 0.023086548, -0.0043411255, -0.0031089783, -0.03842163, -0.029327393, 0.004135132, -0.023345947, ...
http://arxiv.org/abs/2309.04646v1
Efficient Finetuning Large Language Models For Vietnamese Chatbot
Vu-Thuan Doan, Quoc-Truong Truong, Duc-Vu Nguyen, Vinh-Tiep Nguyen, Thuy-Ngan Nguyen Luu
cs.CL, cs.AI
Large language models (LLMs), such as GPT-4, PaLM, and LLaMa, have been shown to achieve remarkable performance across a variety of natural language tasks. Recent advancements in instruction tuning bring LLMs with ability in following user's instructions and producing human-like responses. However, the high costs assoc...
http://arxiv.org/pdf/2309.04646v1
2023-09-09T00:11:53
2023-09-09T00:11:53
{ "title": [ -0.010391235, -0.026184082, -0.010566711, -0.06500244, -0.012359619, 0.014671326, -0.041625977, 0.016571045, -0.0054512024, 0.048339844, -0.032226562, -0.008026123, 0.016159058, -0.0037517548, -0.059692383, -0.0008826256, -0.010192871, ...
http://arxiv.org/abs/2309.04635v1
Can NLP Models 'Identify', 'Distinguish', and 'Justify' Questions that Don't have a Definitive Answer?
Ayushi Agarwal, Nisarg Patel, Neeraj Varshney, Mihir Parmar, Pavan Mallina, Aryan Bhavin Shah, Srihari Raju Sangaraju, Tirth Patel, Nihar Thakkar, Chitta Baral
cs.CL
Though state-of-the-art (SOTA) NLP systems have achieved remarkable performance on a variety of language understanding tasks, they primarily focus on questions that have a correct and a definitive answer. However, in real-world applications, users often ask questions that don't have a definitive answer. Incorrectly ans...
http://arxiv.org/pdf/2309.04635v1
2023-09-08T23:12:03
2023-09-08T23:12:03
{ "title": [ -0.008453369, 0.020019531, -0.018432617, -0.009292603, -0.027374268, 0.0002629757, -0.027252197, -0.0446167, -0.0105896, -0.016281128, -0.048736572, 0.00022637844, -0.032562256, -0.023269653, -0.003271103, -0.051116943, -0.019332886, ...
http://arxiv.org/abs/2309.04607v1
Linking Symptom Inventories using Semantic Textual Similarity
Eamonn Kennedy, Shashank Vadlamani, Hannah M Lindsey, Kelly S Peterson, Kristen Dams OConnor, Kenton Murray, Ronak Agarwal, Houshang H Amiri, Raeda K Andersen, Talin Babikian, David A Baron, Erin D Bigler, Karen Caeyenberghs, Lisa Delano-Wood, Seth G Disner, Ekaterina Dobryakova, Blessen C Eapen, Rachel M Edelstein, Ca...
cs.CL, cs.AI
An extensive library of symptom inventories has been developed over time to measure clinical symptoms, but this variety has led to several long standing issues. Most notably, results drawn from different settings and studies are not comparable, which limits reproducibility. Here, we present an artificial intelligence (...
http://arxiv.org/pdf/2309.04607v1
2023-09-08T21:50:10
2023-09-08T21:50:10
{ "title": [ 0.012504578, 0.006587982, -0.017364502, -0.101379395, 0.013244629, 0.018005371, -0.052703857, -0.029785156, 0.00945282, 0.033355713, -0.013313293, 0.015464783, -0.02154541, -0.025650024, -0.029754639, -0.004558563, -0.00026845932, 0....
http://arxiv.org/abs/2309.04564v1
When Less is More: Investigating Data Pruning for Pretraining LLMs at Scale
Max Marion, Ahmet Üstün, Luiza Pozzobon, Alex Wang, Marzieh Fadaee, Sara Hooker
cs.CL, cs.LG
Large volumes of text data have contributed significantly to the development of large language models (LLMs) in recent years. This data is typically acquired by scraping the internet, leading to pretraining datasets comprised of noisy web text. To date, efforts to prune these datasets down to a higher quality subset ha...
http://arxiv.org/pdf/2309.04564v1
2023-09-08T19:34:05
2023-09-08T19:34:05
{ "title": [ -0.047576904, -0.00982666, -0.061584473, -0.03753662, -0.03186035, -0.003955841, -0.06317139, 0.006603241, -0.021392822, -0.009338379, -0.052581787, 0.04852295, -0.031143188, -0.030883789, 0.022109985, -0.0317688, -0.005355835, 0.040...
http://arxiv.org/abs/2309.04561v1
Three Ways to Improve Verbo-visual Fusion for Dense 3D Visual Grounding
Ozan Unal, Christos Sakaridis, Suman Saha, Fisher Yu, Luc Van Gool
cs.CV, cs.CL
3D visual grounding is the task of localizing the object in a 3D scene which is referred by a description in natural language. With a wide range of applications ranging from autonomous indoor robotics to AR/VR, the task has recently risen in popularity. A common formulation to tackle 3D visual grounding is grounding-by...
http://arxiv.org/pdf/2309.04561v1
2023-09-08T19:27:01
2023-09-08T19:27:01
{ "title": [ 0.01852417, 0.030944824, -0.052734375, -0.062194824, 0.027328491, -0.027542114, -0.014320374, -0.01965332, 0.02128601, 0.04284668, -0.040374756, 0.008132935, 0.0340271, -0.016525269, -0.05419922, 0.0061187744, 0.010025024, 0.04208374...
http://arxiv.org/abs/2309.04550v1
Retrieving Evidence from EHRs with LLMs: Possibilities and Challenges
Hiba Ahsan, Denis Jered McInerney, Jisoo Kim, Christopher Potter, Geoffrey Young, Silvio Amir, Byron C. Wallace
cs.CL
Unstructured Electronic Health Record (EHR) data often contains critical information complementary to imaging data that would inform radiologists' diagnoses. However, time constraints and the large volume of notes frequently associated with individual patients renders manual perusal of such data to identify relevant ev...
http://arxiv.org/pdf/2309.04550v1
2023-09-08T18:44:47
2023-09-08T18:44:47
{ "title": [ -0.04019165, -0.013626099, -0.050567627, -0.04928589, -0.015419006, 0.0077056885, -0.05859375, -0.00031232834, -0.04534912, 0.024414062, -0.020431519, 0.03543091, -0.020553589, -0.031311035, -0.0077667236, -0.00026917458, 0.010551453, ...
http://arxiv.org/abs/2309.04461v1
Measuring and Improving Chain-of-Thought Reasoning in Vision-Language Models
Yangyi Chen, Karan Sikka, Michael Cogswell, Heng Ji, Ajay Divakaran
cs.CL, cs.CV, cs.LG
Vision-language models (VLMs) have recently demonstrated strong efficacy as visual assistants that can parse natural queries about the visual content and generate human-like outputs. In this work, we explore the ability of these models to demonstrate human-like reasoning based on the perceived information. To address a...
http://arxiv.org/pdf/2309.04461v1
2023-09-08T17:49:44
2023-09-08T17:49:44
{ "title": [ -0.01928711, -0.019439697, -0.022994995, -0.039611816, -0.018157959, -0.013069153, -0.033172607, -0.011772156, 0.027999878, 0.0077171326, -0.040863037, -0.0107803345, 0.00957489, -0.021469116, -0.04107666, -0.013259888, 0.0041656494, ...
http://arxiv.org/abs/2309.04389v2
CSPRD: A Financial Policy Retrieval Dataset for Chinese Stock Market
Jinyuan Wang, Hai Zhao, Zhong Wang, Zeyang Zhu, Jinhao Xie, Yong Yu, Yongjian Fei, Yue Huang, Dawei Cheng
cs.CL, cs.CE
In recent years, great advances in pre-trained language models (PLMs) have sparked considerable research focus and achieved promising performance on the approach of dense passage retrieval, which aims at retrieving relative passages from massive corpus with given questions. However, most of existing datasets mainly ben...
http://arxiv.org/pdf/2309.04389v2
2023-09-11T05:19:16
2023-09-08T15:40:54
{ "title": [ 0.000024914742, 0.01008606, -0.05050659, -0.050201416, 0.030395508, -0.02279663, 0.0077705383, -0.0065994263, -0.02255249, 0.029937744, -0.037963867, -0.011222839, -0.000014841557, -0.015899658, -0.04147339, -0.033721924, -0.01713562, ...
http://arxiv.org/abs/2309.04372v1
MoEController: Instruction-based Arbitrary Image Manipulation with Mixture-of-Expert Controllers
Sijia Li, Chen Chen, Haonan Lu
cs.CV, cs.CL
Diffusion-model-based text-guided image generation has recently made astounding progress, producing fascinating results in open-domain image manipulation tasks. Few models, however, currently have complete zero-shot capabilities for both global and local image editing due to the complexity and diversity of image manipu...
http://arxiv.org/pdf/2309.04372v1
2023-09-08T15:06:05
2023-09-08T15:06:05
{ "title": [ -0.0074806213, -0.0056419373, -0.032928467, -0.039886475, 0.008003235, -0.021606445, 0.006713867, -0.008354187, -0.00091552734, 0.023498535, -0.015960693, 0.008636475, -0.0011692047, -0.02168274, -0.037902832, -0.0030574799, 0.008102417,...
http://arxiv.org/abs/2309.04369v1
Beyond Static Datasets: A Deep Interaction Approach to LLM Evaluation
Jiatong Li, Rui Li, Qi Liu
cs.CL, cs.AI
Large Language Models (LLMs) have made progress in various real-world tasks, which stimulates requirements for the evaluation of LLMs. Existing LLM evaluation methods are mainly supervised signal-based which depends on static datasets and cannot evaluate the ability of LLMs in dynamic real-world scenarios where deep in...
http://arxiv.org/pdf/2309.04369v1
2023-09-08T15:00:41
2023-09-08T15:00:41
{ "title": [ -0.029907227, 0.0025959015, -0.053588867, -0.072265625, -0.019332886, 0.008590698, -0.084472656, -0.0017557144, 0.027328491, 0.005962372, -0.03579712, 0.016616821, -0.04800415, -0.020980835, -0.02897644, -0.014144897, -0.03616333, 0....
http://arxiv.org/abs/2309.04333v1
Encoding Multi-Domain Scientific Papers by Ensembling Multiple CLS Tokens
Ronald Seoh, Haw-Shiuan Chang, Andrew McCallum
cs.CL, cs.DL, cs.LG
Many useful tasks on scientific documents, such as topic classification and citation prediction, involve corpora that span multiple scientific domains. Typically, such tasks are accomplished by representing the text with a vector embedding obtained from a Transformer's single CLS token. In this paper, we argue that usi...
http://arxiv.org/pdf/2309.04333v1
2023-09-08T14:00:29
2023-09-08T14:00:29
{ "title": [ -0.04916382, -0.0011358261, -0.057739258, -0.04711914, -0.017593384, 0.0090789795, -0.042633057, 0.002298355, -0.016021729, 0.038757324, 0.026428223, 0.007904053, -0.0028877258, -0.02758789, -0.031280518, 0.012619019, -0.017578125, 0...
http://arxiv.org/abs/2309.04292v1
Fuzzy Fingerprinting Transformer Language-Models for Emotion Recognition in Conversations
Patrícia Pereira, Rui Ribeiro, Helena Moniz, Luisa Coheur, Joao Paulo Carvalho
cs.CL, cs.AI
Fuzzy Fingerprints have been successfully used as an interpretable text classification technique, but, like most other techniques, have been largely surpassed in performance by Large Pre-trained Language Models, such as BERT or RoBERTa. These models deliver state-of-the-art results in several Natural Language Processin...
http://arxiv.org/pdf/2309.04292v1
2023-09-08T12:26:01
2023-09-08T12:26:01
{ "title": [ -0.0042648315, 0.017242432, -0.048950195, -0.021530151, 0.0020942688, -0.022842407, -0.033233643, -0.028503418, 0.008354187, 0.05065918, -0.018356323, -0.023895264, 0.06161499, -0.03604126, -0.03050232, 0.0058403015, -0.0037994385, 0...