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/2305.17540v2
Learning from Children: Improving Image-Caption Pretraining via Curriculum
Hammad A. Ayyubi, Rahul Lokesh, Alireza Zareian, Bo Wu, Shih-Fu Chang
cs.CV, cs.CL
Image-caption pretraining has been quite successfully used for downstream vision tasks like zero-shot image classification and object detection. However, image-caption pretraining is still a hard problem -- it requires multiple concepts (nouns) from captions to be aligned to several objects in images. To tackle this pr...
http://arxiv.org/pdf/2305.17540v2
2023-05-30T15:43:50
2023-05-27T17:59:54
{ "title": [ 0.009376526, -0.0066833496, -0.010787964, -0.027267456, 0.0017213821, -0.02897644, 0.015220642, -0.02217102, -0.020645142, 0.029342651, -0.023895264, 0.044311523, -0.03668213, 0.0018434525, -0.023147583, -0.022705078, -0.009880066, 0...
http://arxiv.org/abs/2305.17534v1
Unsupervised Selective Rationalization with Noise Injection
Adam Storek, Melanie Subbiah, Kathleen McKeown
cs.CL, cs.AI, cs.LG
A major issue with using deep learning models in sensitive applications is that they provide no explanation for their output. To address this problem, unsupervised selective rationalization produces rationales alongside predictions by chaining two jointly-trained components, a rationale generator and a predictor. Altho...
http://arxiv.org/pdf/2305.17534v1
2023-05-27T17:34:36
2023-05-27T17:34:36
{ "title": [ -0.012275696, 0.033691406, -0.01763916, -0.053863525, 0.018844604, 0.018356323, -0.037597656, -0.054595947, -0.0006661415, 0.011550903, -0.037353516, 0.0118255615, -0.020950317, -0.018737793, -0.020874023, -0.009597778, -0.009552002, ...
http://arxiv.org/abs/2306.00005v1
A Two-Stage Decoder for Efficient ICD Coding
Thanh-Tung Nguyen, Viktor Schlegel, Abhinav Kashyap, Stefan Winkler
cs.CL
Clinical notes in healthcare facilities are tagged with the International Classification of Diseases (ICD) code; a list of classification codes for medical diagnoses and procedures. ICD coding is a challenging multilabel text classification problem due to noisy clinical document inputs and long-tailed label distributio...
http://arxiv.org/pdf/2306.00005v1
2023-05-27T17:25:13
2023-05-27T17:25:13
{ "title": [ -0.013114929, -0.00079107285, -0.019561768, -0.052886963, 0.0063476562, -0.002111435, -0.018920898, -0.0036773682, -0.011764526, 0.06210327, 0.0058021545, 0.0061569214, 0.03955078, -0.029708862, -0.019897461, 0.041381836, 0.020401001, ...
http://arxiv.org/abs/2305.17530v1
PuMer: Pruning and Merging Tokens for Efficient Vision Language Models
Qingqing Cao, Bhargavi Paranjape, Hannaneh Hajishirzi
cs.CV, cs.AI, cs.CL
Large-scale vision language (VL) models use Transformers to perform cross-modal interactions between the input text and image. These cross-modal interactions are computationally expensive and memory-intensive due to the quadratic complexity of processing the input image and text. We present PuMer: a token reduction fra...
http://arxiv.org/pdf/2305.17530v1
2023-05-27T17:16:27
2023-05-27T17:16:27
{ "title": [ -0.033325195, -0.026245117, -0.0390625, -0.074157715, 0.0032596588, 0.015365601, -0.064453125, 0.012207031, 0.033721924, 0.018341064, -0.027359009, -0.005176544, 0.01727295, -0.0044898987, -0.0418396, 0.0002644062, 0.010810852, 0.018...
http://arxiv.org/abs/2305.17529v1
MeetingBank: A Benchmark Dataset for Meeting Summarization
Yebowen Hu, Tim Ganter, Hanieh Deilamsalehy, Franck Dernoncourt, Hassan Foroosh, Fei Liu
cs.CL
As the number of recorded meetings increases, it becomes increasingly important to utilize summarization technology to create useful summaries of these recordings. However, there is a crucial lack of annotated meeting corpora for developing this technology, as it can be hard to collect meetings, especially when the top...
http://arxiv.org/pdf/2305.17529v1
2023-05-27T17:09:25
2023-05-27T17:09:25
{ "title": [ -0.007858276, 0.01134491, -0.068603516, -0.103027344, 0.02027893, -0.0027446747, -0.08062744, -0.0054740906, 0.032440186, 0.0546875, -0.01675415, 0.02104187, 0.006801605, 0.009925842, -0.07281494, 0.01159668, 0.0027999878, 0.01770019...
http://arxiv.org/abs/2305.17506v2
Backdooring Neural Code Search
Weisong Sun, Yuchen Chen, Guanhong Tao, Chunrong Fang, Xiangyu Zhang, Quanjun Zhang, Bin Luo
cs.SE, cs.AI, cs.CL, 68T01, I.2.2; D.2.13
Reusing off-the-shelf code snippets from online repositories is a common practice, which significantly enhances the productivity of software developers. To find desired code snippets, developers resort to code search engines through natural language queries. Neural code search models are hence behind many such engines....
http://arxiv.org/pdf/2305.17506v2
2023-06-12T08:05:03
2023-05-27T16:00:50
{ "title": [ -0.008583069, 0.0049591064, -0.0029029846, 0.005466461, -0.03326416, -0.02810669, -0.012573242, -0.038513184, -0.02746582, 0.054870605, -0.011413574, -0.0019664764, 0.0079193115, -0.034240723, -0.013259888, 0.005180359, -0.00022864342, ...
http://arxiv.org/abs/2305.17499v1
CIF-PT: Bridging Speech and Text Representations for Spoken Language Understanding via Continuous Integrate-and-Fire Pre-Training
Linhao Dong, Zhecheng An, Peihao Wu, Jun Zhang, Lu Lu, Zejun Ma
cs.CL, cs.MM, eess.AS
Speech or text representation generated by pre-trained models contains modal-specific information that could be combined for benefiting spoken language understanding (SLU) tasks. In this work, we propose a novel pre-training paradigm termed Continuous Integrate-and-Fire Pre-Training (CIF-PT). It relies on a simple but ...
http://arxiv.org/pdf/2305.17499v1
2023-05-27T15:39:13
2023-05-27T15:39:13
{ "title": [ -0.0014533997, -0.003993988, -0.025314331, -0.08984375, 0.004432678, 0.005973816, -0.057128906, 0.02456665, 0.00026726723, 0.015068054, -0.054718018, 0.030395508, -0.0034503937, -0.0015287399, -0.02532959, -0.014968872, -0.016311646, ...
http://arxiv.org/abs/2305.17497v2
FACTUAL: A Benchmark for Faithful and Consistent Textual Scene Graph Parsing
Zhuang Li, Yuyang Chai, Terry Yue Zhuo, Lizhen Qu, Gholamreza Haffari, Fei Li, Donghong Ji, Quan Hung Tran
cs.CL
Textual scene graph parsing has become increasingly important in various vision-language applications, including image caption evaluation and image retrieval. However, existing scene graph parsers that convert image captions into scene graphs often suffer from two types of errors. First, the generated scene graphs fail...
http://arxiv.org/pdf/2305.17497v2
2023-06-01T04:56:26
2023-05-27T15:38:31
{ "title": [ -0.03265381, -0.022018433, -0.028121948, -0.061523438, -0.015083313, 0.006580353, -0.064575195, -0.003314972, -0.005268097, 0.024887085, -0.024032593, 0.016189575, -0.023345947, -0.012741089, -0.034851074, -0.015853882, -0.011924744, ...
http://arxiv.org/abs/2305.17493v2
The Curse of Recursion: Training on Generated Data Makes Models Forget
Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Yarin Gal, Nicolas Papernot, Ross Anderson
cs.LG, cs.AI, cs.CL, cs.CR, cs.CV
Stable Diffusion revolutionised image creation from descriptive text. GPT-2, GPT-3(.5) and GPT-4 demonstrated astonishing performance across a variety of language tasks. ChatGPT introduced such language models to the general public. It is now clear that large language models (LLMs) are here to stay, and will bring abou...
http://arxiv.org/pdf/2305.17493v2
2023-05-31T10:39:26
2023-05-27T15:10:41
{ "title": [ -0.011817932, 0.004776001, -0.04067993, -0.04800415, 0.0126953125, -0.004611969, -0.035339355, -0.03274536, 0.0048675537, 0.017242432, -0.030334473, -0.003921509, -0.017364502, -0.01586914, -0.05505371, -0.010238647, -0.004196167, 0....
http://arxiv.org/abs/2305.17491v1
FERMAT: An Alternative to Accuracy for Numerical Reasoning
Jasivan Alex Sivakumar, Nafise Sadat Moosavi
cs.CL
While pre-trained language models achieve impressive performance on various NLP benchmarks, they still struggle with tasks that require numerical reasoning. Recent advances in improving numerical reasoning are mostly achieved using very large language models that contain billions of parameters and are not accessible to...
http://arxiv.org/pdf/2305.17491v1
2023-05-27T15:00:45
2023-05-27T15:00:45
{ "title": [ -0.01121521, 0.015159607, -0.04586792, -0.031204224, 0.020401001, -0.034454346, 0.020095825, -0.01133728, -0.0033340454, 0.031219482, -0.0035171509, -0.007648468, 0.023742676, -0.039367676, -0.008987427, 0.017684937, 0.0039634705, 0....
http://arxiv.org/abs/2305.17480v2
A Match Made in Heaven: A Multi-task Framework for Hyperbole and Metaphor Detection
Naveen Badathala, Abisek Rajakumar Kalarani, Tejpalsingh Siledar, Pushpak Bhattacharyya
cs.CL, cs.AI
Hyperbole and metaphor are common in day-to-day communication (e.g., "I am in deep trouble": how does trouble have depth?), which makes their detection important, especially in a conversational AI setting. Existing approaches to automatically detect metaphor and hyperbole have studied these language phenomena independe...
http://arxiv.org/pdf/2305.17480v2
2023-05-30T13:35:35
2023-05-27T14:17:59
{ "title": [ 0.017791748, 0.0078086853, -0.024139404, -0.023284912, 0.017990112, -0.023803711, 0.0012273788, -0.032592773, 0.0014715195, 0.04675293, -0.0002975464, 0.00016093254, 0.031143188, -0.016220093, -0.059783936, 0.009025574, -0.019515991, ...
http://arxiv.org/abs/2305.18365v2
What can Large Language Models do in chemistry? A comprehensive benchmark on eight tasks
Taicheng Guo, Kehan Guo, Bozhao Nan, Zhenwen Liang, Zhichun Guo, Nitesh V. Chawla, Olaf Wiest, Xiangliang Zhang
cs.CL, cs.AI
Large Language Models (LLMs) with strong abilities in natural language processing tasks have emerged and have been applied in various kinds of areas such as science, finance and software engineering. However, the capability of LLMs to advance the field of chemistry remains unclear. In this paper, rather than pursuing s...
http://arxiv.org/pdf/2305.18365v2
2023-09-10T16:37:36
2023-05-27T14:17:33
{ "title": [ -0.043151855, -0.017944336, -0.054779053, -0.06982422, -0.0059127808, 0.009147644, -0.07513428, 0.020904541, 0.00044178963, 0.025863647, 0.0061836243, 0.005958557, 0.0019817352, -0.024169922, -0.03112793, 0.020812988, -0.007904053, 0...
http://arxiv.org/abs/2305.17458v1
A Diffusion Model for Event Skeleton Generation
Fangqi Zhu, Lin Zhang, Jun Gao, Bing Qin, Ruifeng Xu, Haiqin Yang
cs.CL
Event skeleton generation, aiming to induce an event schema skeleton graph with abstracted event nodes and their temporal relations from a set of event instance graphs, is a critical step in the temporal complex event schema induction task. Existing methods effectively address this task from a graph generation perspect...
http://arxiv.org/pdf/2305.17458v1
2023-05-27T12:19:21
2023-05-27T12:19:21
{ "title": [ -0.005153656, -0.010726929, -0.029891968, -0.07922363, 0.0055274963, 0.015914917, -0.05670166, -0.009681702, -0.0058517456, 0.015037537, 0.027862549, -0.023529053, -0.0019454956, -0.04598999, -0.034942627, -0.0011940002, -0.015289307, ...
http://arxiv.org/abs/2305.17457v1
Financial misstatement detection: a realistic evaluation
Elias Zavitsanos, Dimitris Mavroeidis, Konstantinos Bougiatiotis, Eirini Spyropoulou, Lefteris Loukas, Georgios Paliouras
cs.CL, cs.LG, q-fin.CP
In this work, we examine the evaluation process for the task of detecting financial reports with a high risk of containing a misstatement. This task is often referred to, in the literature, as ``misstatement detection in financial reports''. We provide an extensive review of the related literature. We propose a new, re...
http://arxiv.org/pdf/2305.17457v1
2023-05-27T12:19:13
2023-05-27T12:19:13
{ "title": [ -0.0031757355, 0.019622803, -0.03427124, -0.024978638, 0.010627747, -0.011131287, -0.02142334, -0.037353516, -0.011245728, 0.0287323, -0.041290283, -0.006072998, -0.008033752, -0.008430481, -0.04949951, -0.019134521, -0.03164673, 0.0...
http://arxiv.org/abs/2305.17455v2
CrossGET: Cross-Guided Ensemble of Tokens for Accelerating Vision-Language Transformers
Dachuan Shi, Chaofan Tao, Anyi Rao, Zhendong Yang, Chun Yuan, Jiaqi Wang
cs.CV, cs.CL
Recent vision-language models have achieved tremendous progress far beyond what we ever expected. However, their computational costs are also dramatically growing with rapid development, especially for the large models. It makes model acceleration exceedingly critical in a scenario of limited resources. Although extens...
http://arxiv.org/pdf/2305.17455v2
2023-10-04T22:11:50
2023-05-27T12:07:21
{ "title": [ -0.018005371, -0.022979736, -0.03857422, -0.073791504, 0.006587982, -0.00023341179, -0.037628174, 0.021820068, 0.02142334, 0.030303955, -0.05532837, 0.0030956268, 0.026397705, 0.0023441315, -0.037841797, 0.0030326843, 0.008758545, 0....
http://arxiv.org/abs/2305.17448v1
Measuring Your ASTE Models in The Wild: A Diversified Multi-domain Dataset For Aspect Sentiment Triplet Extraction
Ting Xu, Huiyun Yang, Zhen Wu, Jiaze Chen, Fei Zhao, Xinyu Dai
cs.CL
Aspect Sentiment Triplet Extraction (ASTE) is widely used in various applications. However, existing ASTE datasets are limited in their ability to represent real-world scenarios, hindering the advancement of research in this area. In this paper, we introduce a new dataset, named DMASTE, which is manually annotated to b...
http://arxiv.org/pdf/2305.17448v1
2023-05-27T11:21:32
2023-05-27T11:21:32
{ "title": [ 0.015007019, 0.008995056, -0.042999268, -0.07141113, 0.01537323, -0.010688782, -0.040222168, -0.019592285, 0.023925781, 0.05529785, -0.003873825, -0.013046265, 0.0069007874, 0.00034570694, -0.07952881, 0.012069702, 0.0043907166, 0.04...
http://arxiv.org/abs/2305.17446v2
Fine-tuning Happens in Tiny Subspaces: Exploring Intrinsic Task-specific Subspaces of Pre-trained Language Models
Zhong Zhang, Bang Liu, Junming Shao
cs.CL, cs.AI, cs.LG, 68T50, I.2.7
Pre-trained language models (PLMs) are known to be overly parameterized and have significant redundancy, indicating a small degree of freedom of the PLMs. Motivated by the observation, in this paper, we study the problem of re-parameterizing and fine-tuning PLMs from a new perspective: Discovery of intrinsic task-speci...
http://arxiv.org/pdf/2305.17446v2
2023-08-01T08:54:06
2023-05-27T11:16:26
{ "title": [ -0.016937256, -0.017425537, -0.024871826, -0.027008057, -0.033843994, -0.0209198, -0.03062439, -0.00022518635, -0.03717041, 0.039276123, -0.030151367, 0.0309906, 0.00095272064, -0.012802124, -0.016799927, -0.007045746, -0.008842468, ...
http://arxiv.org/abs/2305.17444v1
Query-Efficient Black-Box Red Teaming via Bayesian Optimization
Deokjae Lee, JunYeong Lee, Jung-Woo Ha, Jin-Hwa Kim, Sang-Woo Lee, Hwaran Lee, Hyun Oh Song
cs.AI, cs.CL, cs.CR, cs.LG
The deployment of large-scale generative models is often restricted by their potential risk of causing harm to users in unpredictable ways. We focus on the problem of black-box red teaming, where a red team generates test cases and interacts with the victim model to discover a diverse set of failures with limited query...
http://arxiv.org/pdf/2305.17444v1
2023-05-27T11:00:15
2023-05-27T11:00:15
{ "title": [ -0.01222229, 0.0060920715, -0.031951904, -0.054016113, 0.008102417, -0.0084991455, -0.026473999, -0.025726318, -0.009170532, 0.010932922, -0.023391724, 0.02331543, -0.044799805, -0.007987976, -0.016601562, -0.026123047, 0.00072574615, ...
http://arxiv.org/abs/2305.17442v3
Weaker Than You Think: A Critical Look at Weakly Supervised Learning
Dawei Zhu, Xiaoyu Shen, Marius Mosbach, Andreas Stephan, Dietrich Klakow
cs.CL
Weakly supervised learning is a popular approach for training machine learning models in low-resource settings. Instead of requesting high-quality yet costly human annotations, it allows training models with noisy annotations obtained from various weak sources. Recently, many sophisticated approaches have been proposed...
http://arxiv.org/pdf/2305.17442v3
2023-09-17T19:04:44
2023-05-27T10:46:50
{ "title": [ -0.0054092407, 0.0072402954, -0.03463745, -0.04525757, 0.022857666, -0.004611969, -0.016937256, -0.030181885, -0.015411377, 0.019729614, -0.028015137, 0.01979065, -0.017974854, -0.01939392, -0.030975342, -0.012817383, 0.0035495758, 0...
http://arxiv.org/abs/2305.17440v1
Modeling Adversarial Attack on Pre-trained Language Models as Sequential Decision Making
Xuanjie Fang, Sijie Cheng, Yang Liu, Wei Wang
cs.CL
Pre-trained language models (PLMs) have been widely used to underpin various downstream tasks. However, the adversarial attack task has found that PLMs are vulnerable to small perturbations. Mainstream methods adopt a detached two-stage framework to attack without considering the subsequent influence of substitution at...
http://arxiv.org/pdf/2305.17440v1
2023-05-27T10:33:53
2023-05-27T10:33:53
{ "title": [ 0.003326416, -0.007835388, -0.019836426, -0.059448242, -0.00011974573, -0.007980347, -0.016525269, -0.010154724, -0.052947998, 0.017654419, 0.01939392, 0.016738892, -0.037017822, -0.03237915, -0.021194458, -0.01828003, 0.0010185242, ...
http://arxiv.org/abs/2305.17433v1
A Unified Framework for Slot based Response Generation in a Multimodal Dialogue System
Mauajama Firdaus, Avinash Madasu, Asif Ekbal
cs.CV, cs.CL
Natural Language Understanding (NLU) and Natural Language Generation (NLG) are the two critical components of every conversational system that handles the task of understanding the user by capturing the necessary information in the form of slots and generating an appropriate response in accordance with the extracted in...
http://arxiv.org/pdf/2305.17433v1
2023-05-27T10:06:03
2023-05-27T10:06:03
{ "title": [ -0.010757446, -0.01789856, -0.036743164, -0.060699463, -0.0063171387, -0.008659363, -0.04550171, 0.021469116, -0.009468079, 0.037109375, -0.008384705, 0.016174316, -0.030014038, -0.010192871, -0.07080078, -0.019302368, -0.016845703, ...
http://arxiv.org/abs/2305.17422v2
Understanding Emotion Valence is a Joint Deep Learning Task
Gabriel Roccabruna, Seyed Mahed Mousavi, Giuseppe Riccardi
cs.CL, cs.AI
The valence analysis of speakers' utterances or written posts helps to understand the activation and variations of the emotional state throughout the conversation. More recently, the concept of Emotion Carriers (EC) has been introduced to explain the emotion felt by the speaker and its manifestations. In this work, we ...
http://arxiv.org/pdf/2305.17422v2
2023-10-31T13:58:51
2023-05-27T09:07:18
{ "title": [ 0.023269653, 0.027130127, -0.04321289, -0.027008057, -0.010375977, -0.042877197, -0.003145218, -0.018737793, 0.017501831, 0.029190063, -0.029647827, -0.015411377, 0.031234741, -0.03930664, -0.020202637, 0.003545761, 0.009155273, 0.01...
http://arxiv.org/abs/2305.17416v1
A Practical Toolkit for Multilingual Question and Answer Generation
Asahi Ushio, Fernando Alva-Manchego, Jose Camacho-Collados
cs.CL
Generating questions along with associated answers from a text has applications in several domains, such as creating reading comprehension tests for students, or improving document search by providing auxiliary questions and answers based on the query. Training models for question and answer generation (QAG) is not str...
http://arxiv.org/pdf/2305.17416v1
2023-05-27T08:42:37
2023-05-27T08:42:37
{ "title": [ -0.029449463, -0.019454956, -0.024291992, -0.05227661, -0.009712219, 0.01675415, -0.052459717, 0.00472641, 0.015083313, 0.026611328, 0.004016876, 0.014419556, -0.02180481, -0.014480591, -0.05239868, 0.014205933, 0.0032100677, 0.01599...
http://arxiv.org/abs/2305.17415v2
Exploring Better Text Image Translation with Multimodal Codebook
Zhibin Lan, Jiawei Yu, Xiang Li, Wen Zhang, Jian Luan, Bin Wang, Degen Huang, Jinsong Su
cs.CL, cs.AI
Text image translation (TIT) aims to translate the source texts embedded in the image to target translations, which has a wide range of applications and thus has important research value. However, current studies on TIT are confronted with two main bottlenecks: 1) this task lacks a publicly available TIT dataset, 2) do...
http://arxiv.org/pdf/2305.17415v2
2023-06-02T12:38:37
2023-05-27T08:41:18
{ "title": [ -0.039916992, 0.005657196, -0.020217896, -0.019256592, -0.017089844, 0.0033435822, -0.037628174, -0.022598267, 0.023635864, 0.059936523, -0.016036987, 0.03366089, 0.020462036, -0.008102417, -0.039031982, 0.026382446, -0.004787445, 0....
http://arxiv.org/abs/2305.17406v1
Enhancing Translation for Indigenous Languages: Experiments with Multilingual Models
Atnafu Lambebo Tonja, Hellina Hailu Nigatu, Olga Kolesnikova, Grigori Sidorov, Alexander Gelbukh, Jugal Kalita
cs.CL
This paper describes CIC NLP's submission to the AmericasNLP 2023 Shared Task on machine translation systems for indigenous languages of the Americas. We present the system descriptions for three methods. We used two multilingual models, namely M2M-100 and mBART50, and one bilingual (one-to-one) -- Helsinki NLP Spanish...
http://arxiv.org/pdf/2305.17406v1
2023-05-27T08:10:40
2023-05-27T08:10:40
{ "title": [ -0.06402588, -0.006664276, -0.0061569214, -0.021957397, -0.030761719, 0.017868042, -0.064086914, -0.020355225, -0.00642395, 0.058532715, 0.0015745163, 0.017456055, 0.020385742, -0.018585205, -0.06085205, 0.018218994, -0.013786316, 0....
http://arxiv.org/abs/2305.17404v1
Parallel Corpus for Indigenous Language Translation: Spanish-Mazatec and Spanish-Mixtec
Atnafu Lambebo Tonja, Christian Maldonado-Sifuentes, David Alejandro Mendoza Castillo, Olga Kolesnikova, Noé Castro-Sánchez, Grigori Sidorov, Alexander Gelbukh
cs.CL
In this paper, we present a parallel Spanish-Mazatec and Spanish-Mixtec corpus for machine translation (MT) tasks, where Mazatec and Mixtec are two indigenous Mexican languages. We evaluated the usability of the collected corpus using three different approaches: transformer, transfer learning, and fine-tuning pre-train...
http://arxiv.org/pdf/2305.17404v1
2023-05-27T08:03:44
2023-05-27T08:03:44
{ "title": [ -0.057678223, -0.0042381287, -0.057556152, 0.0052757263, -0.04888916, -0.016113281, -0.027633667, 0.011894226, -0.026046753, 0.08709717, 0.006088257, -0.013191223, 0.09265137, -0.04425049, -0.02848816, -0.0008010864, -0.00712204, 0.0...
http://arxiv.org/abs/2305.17401v3
A Framework For Refining Text Classification and Object Recognition from Academic Articles
Jinghong Li, Koichi Ota, Wen Gu, Shinobu Hasegawa
cs.CV, cs.CL, cs.LG
With the widespread use of the internet, it has become increasingly crucial to extract specific information from vast amounts of academic articles efficiently. Data mining techniques are generally employed to solve this issue. However, data mining for academic articles is challenging since it requires automatically ext...
http://arxiv.org/pdf/2305.17401v3
2023-08-15T16:02:43
2023-05-27T07:59:49
{ "title": [ -0.022994995, 0.015991211, -0.037231445, -0.03677368, 0.01637268, -0.009437561, -0.02482605, -0.0390625, 0.004173279, 0.017440796, -0.052215576, 0.013572693, -0.03842163, 0.01423645, -0.06958008, -0.027389526, -0.0033130646, 0.033172...
http://arxiv.org/abs/2305.17393v2
Answering Unanswered Questions through Semantic Reformulations in Spoken QA
Pedro Faustini, Zhiyu Chen, Besnik Fetahu, Oleg Rokhlenko, Shervin Malmasi
cs.CL, cs.AI
Spoken Question Answering (QA) is a key feature of voice assistants, usually backed by multiple QA systems. Users ask questions via spontaneous speech which can contain disfluencies, errors, and informal syntax or phrasing. This is a major challenge in QA, causing unanswered questions or irrelevant answers, and leading...
http://arxiv.org/pdf/2305.17393v2
2023-06-03T05:40:25
2023-05-27T07:19:27
{ "title": [ -0.043518066, -0.0073928833, -0.040100098, -0.05114746, -0.021072388, -0.0013599396, -0.051361084, 0.006755829, 0.0026512146, 0.03857422, -0.021316528, -0.016845703, -0.006412506, -0.008338928, -0.07128906, -0.024093628, -0.00035309792, ...
http://arxiv.org/abs/2305.17390v1
SwiftSage: A Generative Agent with Fast and Slow Thinking for Complex Interactive Tasks
Bill Yuchen Lin, Yicheng Fu, Karina Yang, Prithviraj Ammanabrolu, Faeze Brahman, Shiyu Huang, Chandra Bhagavatula, Yejin Choi, Xiang Ren
cs.CL, cs.AI, cs.LG, cs.MA, cs.RO
We introduce SwiftSage, a novel agent framework inspired by the dual-process theory of human cognition, designed to excel in action planning for complex interactive reasoning tasks. SwiftSage integrates the strengths of behavior cloning and prompting large language models (LLMs) to enhance task completion performance. ...
http://arxiv.org/pdf/2305.17390v1
2023-05-27T07:04:15
2023-05-27T07:04:15
{ "title": [ -0.009651184, -0.009605408, -0.029800415, -0.029296875, -0.014472961, -0.032989502, -0.0035037994, -0.01802063, 0.01751709, 0.028762817, -0.011512756, -0.010314941, -0.027816772, -0.028640747, -0.0647583, -0.0047073364, -0.0067825317, ...
http://arxiv.org/abs/2305.17388v1
MPCHAT: Towards Multimodal Persona-Grounded Conversation
Jaewoo Ahn, Yeda Song, Sangdoo Yun, Gunhee Kim
cs.CL, cs.CV
In order to build self-consistent personalized dialogue agents, previous research has mostly focused on textual persona that delivers personal facts or personalities. However, to fully describe the multi-faceted nature of persona, image modality can help better reveal the speaker's personal characteristics and experien...
http://arxiv.org/pdf/2305.17388v1
2023-05-27T06:46:42
2023-05-27T06:46:42
{ "title": [ -0.005241394, -0.0018005371, -0.04458618, -0.07458496, 0.017715454, -0.0017223358, -0.05218506, 0.0023765564, 0.0072631836, 0.049591064, 0.006210327, 0.00919342, 0.023544312, -0.019104004, -0.04498291, 0.015602112, -0.038848877, 0.04...
http://arxiv.org/abs/2305.17378v1
Improving Generalization in Language Model-Based Text-to-SQL Semantic Parsing: Two Simple Semantic Boundary-Based Techniques
Daking Rai, Bailin Wang, Yilun Zhou, Ziyu Yao
cs.CL, cs.AI, I.2.7
Compositional and domain generalization present significant challenges in semantic parsing, even for state-of-the-art semantic parsers based on pre-trained language models (LMs). In this study, we empirically investigate improving an LM's generalization in semantic parsing with two simple techniques: at the token level...
http://arxiv.org/pdf/2305.17378v1
2023-05-27T06:09:03
2023-05-27T06:09:03
{ "title": [ -0.0035896301, 0.009216309, -0.020217896, -0.07281494, 0.016860962, 0.017089844, -0.0541687, -0.0037612915, 0.0013246536, 0.051116943, -0.053527832, 0.018936157, 0.00944519, -0.0028362274, -0.04006958, -0.018814087, -0.010070801, 0.0...
http://arxiv.org/abs/2305.17373v1
Zero- and Few-Shot Event Detection via Prompt-Based Meta Learning
Zhenrui Yue, Huimin Zeng, Mengfei Lan, Heng Ji, Dong Wang
cs.CL, cs.AI
With emerging online topics as a source for numerous new events, detecting unseen / rare event types presents an elusive challenge for existing event detection methods, where only limited data access is provided for training. To address the data scarcity problem in event detection, we propose MetaEvent, a meta learning...
http://arxiv.org/pdf/2305.17373v1
2023-05-27T05:36:46
2023-05-27T05:36:46
{ "title": [ 0.020477295, 0.00522995, -0.020324707, -0.045715332, 0.011978149, -0.010253906, -0.03829956, -0.04385376, 0.012702942, 0.02545166, -0.015975952, 0.0021915436, -0.0057868958, -0.027374268, -0.041625977, -0.0017366409, 0.00022375584, 0...
http://arxiv.org/abs/2305.17371v1
Towards Better Entity Linking with Multi-View Enhanced Distillation
Yi Liu, Yuan Tian, Jianxun Lian, Xinlong Wang, Yanan Cao, Fang Fang, Wen Zhang, Haizhen Huang, Denvy Deng, Qi Zhang
cs.CL
Dense retrieval is widely used for entity linking to retrieve entities from large-scale knowledge bases. Mainstream techniques are based on a dual-encoder framework, which encodes mentions and entities independently and calculates their relevances via rough interaction metrics, resulting in difficulty in explicitly mod...
http://arxiv.org/pdf/2305.17371v1
2023-05-27T05:15:28
2023-05-27T05:15:28
{ "title": [ 0.02116394, 0.013015747, -0.06213379, -0.12359619, 0.0413208, -0.005142212, -0.055664062, 0.013633728, 0.032989502, 0.028076172, -0.022354126, -0.01285553, 0.022720337, -0.023376465, -0.030639648, 0.0055732727, -0.014549255, 0.027938...
http://arxiv.org/abs/2305.17367v1
Augmenting Large Language Model Translators via Translation Memories
Yongyu Mu, Abudurexiti Reheman, Zhiquan Cao, Yuchun Fan, Bei Li, Yinqiao Li, Tong Xiao, Chunliang Zhang, Jingbo Zhu
cs.CL
Using translation memories (TMs) as prompts is a promising approach to in-context learning of machine translation models. In this work, we take a step towards prompting large language models (LLMs) with TMs and making them better translators. We find that the ability of LLMs to ``understand'' prompts is indeed helpful ...
http://arxiv.org/pdf/2305.17367v1
2023-05-27T04:47:09
2023-05-27T04:47:09
{ "title": [ -0.036376953, -0.022140503, -0.02758789, -0.068359375, -0.014450073, 0.016464233, -0.056732178, 0.014305115, 0.007949829, 0.05038452, -0.0052871704, -0.013069153, 0.022766113, -0.018417358, -0.060302734, 0.0256958, 0.004524231, 0.017...
http://arxiv.org/abs/2305.17364v1
An Investigation of Evaluation Metrics for Automated Medical Note Generation
Asma Ben Abacha, Wen-wai Yim, George Michalopoulos, Thomas Lin
cs.CL, cs.AI, cs.LG
Recent studies on automatic note generation have shown that doctors can save significant amounts of time when using automatic clinical note generation (Knoll et al., 2022). Summarization models have been used for this task to generate clinical notes as summaries of doctor-patient conversations (Krishna et al., 2021; Ca...
http://arxiv.org/pdf/2305.17364v1
2023-05-27T04:34:58
2023-05-27T04:34:58
{ "title": [ -0.03665161, -0.014015198, -0.029052734, -0.047821045, -0.012290955, 0.02027893, -0.057006836, -0.009094238, -0.02420044, 0.040618896, -0.007926941, 0.010246277, -0.0004491806, -0.010986328, -0.050048828, -0.010185242, -0.0024318695, ...
http://arxiv.org/abs/2305.17359v2
DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text
Xianjun Yang, Wei Cheng, Yue Wu, Linda Petzold, William Yang Wang, Haifeng Chen
cs.CL, cs.AI
Large language models (LLMs) have notably enhanced the fluency and diversity of machine-generated text. However, this progress also presents a significant challenge in detecting the origin of a given text, and current research on detection methods lags behind the rapid evolution of LLMs. Conventional training-based met...
http://arxiv.org/pdf/2305.17359v2
2023-10-04T16:36:09
2023-05-27T03:58:29
{ "title": [ -0.019699097, -0.0043411255, -0.03970337, -0.04977417, 0.0016679764, -0.006313324, -0.036132812, -0.00869751, -0.010025024, 0.018051147, -0.031463623, 0.02798462, -0.0135269165, -0.010696411, -0.03390503, -0.00409317, -0.0154953, 0.0...
http://arxiv.org/abs/2305.17358v1
CTC-based Non-autoregressive Speech Translation
Chen Xu, Xiaoqian Liu, Xiaowen Liu, Qingxuan Sun, Yuhao Zhang, Murun Yang, Qianqian Dong, Tom Ko, Mingxuan Wang, Tong Xiao, Anxiang Ma, Jingbo Zhu
cs.CL
Combining end-to-end speech translation (ST) and non-autoregressive (NAR) generation is promising in language and speech processing for their advantages of less error propagation and low latency. In this paper, we investigate the potential of connectionist temporal classification (CTC) for non-autoregressive speech tra...
http://arxiv.org/pdf/2305.17358v1
2023-05-27T03:54:09
2023-05-27T03:54:09
{ "title": [ -0.02796936, -0.01737976, -0.03591919, -0.046813965, -0.025405884, -0.004219055, -0.063964844, 0.015380859, -0.003293991, 0.066711426, -0.022323608, 0.007648468, 0.03314209, -0.0073547363, -0.06100464, 0.014862061, -0.0067634583, 0.0...
http://arxiv.org/abs/2305.17356v1
Bridging the Granularity Gap for Acoustic Modeling
Chen Xu, Yuhao Zhang, Chengbo Jiao, Xiaoqian Liu, Chi Hu, Xin Zeng, Tong Xiao, Anxiang Ma, Huizhen Wang, JingBo Zhu
cs.CL
While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses the acoustic feature...
http://arxiv.org/pdf/2305.17356v1
2023-05-27T03:52:52
2023-05-27T03:52:52
{ "title": [ -0.023132324, 0.0073509216, -0.05001831, -0.11541748, 0.012649536, 0.015357971, -0.11517334, 0.000895977, 0.015335083, 0.03527832, 0.0022583008, -0.026947021, 0.030731201, -0.034332275, -0.043060303, 0.006549835, -0.0048675537, 0.021...
http://arxiv.org/abs/2305.17353v2
Complementary and Integrative Health Lexicon (CIHLex) and Entity Recognition in the Literature
Huixue Zhou, Robin Austin, Sheng-Chieh Lu, Greg Silverman, Yuqi Zhou, Halil Kilicoglu, Hua Xu, Rui Zhang
cs.CL
Objective: Our study aimed to construct an exhaustive Complementary and Integrative Health (CIH) Lexicon (CIHLex) to better represent the often underrepresented physical and psychological CIH approaches in standard terminologies. We also intended to apply advanced Natural Language Processing (NLP) models such as Bidire...
http://arxiv.org/pdf/2305.17353v2
2023-11-14T04:03:18
2023-05-27T03:21:36
{ "title": [ -0.029556274, 0.015670776, -0.07946777, -0.04058838, 0.011024475, -0.03665161, -0.041534424, 0.007522583, -0.042633057, 0.028244019, -0.0027751923, 0.0052223206, 0.009437561, -0.031082153, 0.00592041, -0.021148682, -0.0032615662, 0.0...
http://arxiv.org/abs/2305.17351v1
Disambiguated Lexically Constrained Neural Machine Translation
Jinpeng Zhang, Nini Xiao, Ke Wang, Chuanqi Dong, Xiangyu Duan, Yuqi Zhang, Min Zhang
cs.CL
Lexically constrained neural machine translation (LCNMT), which controls the translation generation with pre-specified constraints, is important in many practical applications. Current approaches to LCNMT typically assume that the pre-specified lexical constraints are contextually appropriate. This assumption limits th...
http://arxiv.org/pdf/2305.17351v1
2023-05-27T03:15:10
2023-05-27T03:15:10
{ "title": [ -0.032043457, 0.0012226105, -0.024963379, -0.04208374, -0.0038089752, 0.019363403, -0.052612305, -0.0053977966, 0.00047135353, 0.04840088, -0.022247314, -0.004337311, 0.025878906, -0.006023407, -0.030273438, 0.0129470825, -0.0056762695, ...
http://arxiv.org/abs/2305.17350v1
How Good is Automatic Segmentation as a Multimodal Discourse Annotation Aid?
Corbyn Terpstra, Ibrahim Khebour, Mariah Bradford, Brett Wisniewski, Nikhil Krishnaswamy, Nathaniel Blanchard
cs.CL
Collaborative problem solving (CPS) in teams is tightly coupled with the creation of shared meaning between participants in a situated, collaborative task. In this work, we assess the quality of different utterance segmentation techniques as an aid in annotating CPS. We (1) manually transcribe utterances in a dataset o...
http://arxiv.org/pdf/2305.17350v1
2023-05-27T03:06:15
2023-05-27T03:06:15
{ "title": [ -0.0041275024, 0.0035858154, -0.0259552, -0.05987549, -0.016998291, -0.004425049, -0.049926758, -0.004005432, -0.020019531, 0.026809692, 0.024765015, 0.01612854, -0.010574341, -0.03010559, -0.025650024, 0.014289856, -0.0067329407, 0....
http://arxiv.org/abs/2305.17347v1
CGELBank Annotation Manual v1.0
Brett Reynolds, Nathan Schneider, Aryaman Arora
cs.CL
CGELBank is a treebank and associated tools based on a syntactic formalism for English derived from the Cambridge Grammar of the English Language. This document lays out the particularities of the CGELBank annotation scheme.
http://arxiv.org/pdf/2305.17347v1
2023-05-27T03:01:53
2023-05-27T03:01:53
{ "title": [ -0.019638062, -0.0036182404, -0.021469116, -0.05407715, -0.00018191338, 0.02420044, -0.07733154, -0.03591919, -0.0038547516, 0.056030273, 0.010475159, -0.01121521, 0.047027588, -0.036621094, -0.02218628, 0.016052246, 0.022903442, -0....
http://arxiv.org/abs/2305.17337v1
Benchmarking Diverse-Modal Entity Linking with Generative Models
Sijia Wang, Alexander Hanbo Li, Henry Zhu, Sheng Zhang, Chung-Wei Hang, Pramuditha Perera, Jie Ma, William Wang, Zhiguo Wang, Vittorio Castelli, Bing Xiang, Patrick Ng
cs.CL, cs.AI
Entities can be expressed in diverse formats, such as texts, images, or column names and cell values in tables. While existing entity linking (EL) models work well on per modality configuration, such as text-only EL, visual grounding, or schema linking, it is more challenging to design a unified model for diverse modal...
http://arxiv.org/pdf/2305.17337v1
2023-05-27T02:38:46
2023-05-27T02:38:46
{ "title": [ -0.00015866756, 0.016311646, -0.042053223, -0.09094238, 0.037628174, -0.0056381226, -0.06958008, -0.019180298, 0.0140686035, 0.04232788, 0.0046577454, 0.023757935, 0.024932861, -0.02746582, -0.019989014, 0.025100708, -0.014320374, 0....
http://arxiv.org/abs/2305.17333v2
Fine-Tuning Language Models with Just Forward Passes
Sadhika Malladi, Tianyu Gao, Eshaan Nichani, Alex Damian, Jason D. Lee, Danqi Chen, Sanjeev Arora
cs.LG, cs.CL
Fine-tuning language models (LMs) has yielded success on diverse downstream tasks, but as LMs grow in size, backpropagation requires a prohibitively large amount of memory. Zeroth-order (ZO) methods can in principle estimate gradients using only two forward passes but are theorized to be catastrophically slow for optim...
http://arxiv.org/pdf/2305.17333v2
2023-10-31T14:57:22
2023-05-27T02:28:10
{ "title": [ -0.021774292, -0.016830444, -0.01651001, -0.08746338, -0.0070648193, 0.017700195, -0.07598877, 0.017593384, -0.0008401871, 0.030761719, -0.01966858, 0.002614975, -0.021453857, -0.0059928894, -0.05606079, 0.0029563904, -0.011833191, 0...
http://arxiv.org/abs/2305.17331v1
Augmentation-Adapted Retriever Improves Generalization of Language Models as Generic Plug-In
Zichun Yu, Chenyan Xiong, Shi Yu, Zhiyuan Liu
cs.CL, cs.LG
Retrieval augmentation can aid language models (LMs) in knowledge-intensive tasks by supplying them with external information. Prior works on retrieval augmentation usually jointly fine-tune the retriever and the LM, making them closely coupled. In this paper, we explore the scheme of generic retrieval plug-in: the ret...
http://arxiv.org/pdf/2305.17331v1
2023-05-27T02:26:52
2023-05-27T02:26:52
{ "title": [ 0.02357483, 0.007980347, -0.025619507, -0.08117676, -0.0073280334, 0.0013923645, -0.06567383, 0.007335663, -0.018997192, 0.043395996, -0.011955261, 0.01449585, 0.033111572, -0.022323608, -0.004688263, 0.023529053, -0.0032577515, 0.02...
http://arxiv.org/abs/2305.17325v1
Why Does Zero-Shot Cross-Lingual Generation Fail? An Explanation and a Solution
Tianjian Li, Kenton Murray
cs.CL
Zero-shot cross-lingual transfer is when a multilingual model is trained to perform a task in one language and then is applied to another language. Although the zero-shot cross-lingual transfer approach has achieved success in various classification tasks, its performance on natural language generation tasks falls shor...
http://arxiv.org/pdf/2305.17325v1
2023-05-27T02:04:19
2023-05-27T02:04:19
{ "title": [ -0.0046157837, -0.018844604, -0.021224976, -0.037506104, -0.0181427, -0.016189575, -0.028335571, -0.018234253, 0.009468079, 0.038513184, 0.0010595322, 0.0025634766, 0.0074157715, -0.034362793, -0.027282715, 0.016799927, -0.013031006, ...
http://arxiv.org/abs/2305.17311v1
Beyond Positive Scaling: How Negation Impacts Scaling Trends of Language Models
Yuhui Zhang, Michihiro Yasunaga, Zhengping Zhou, Jeff Z. HaoChen, James Zou, Percy Liang, Serena Yeung
cs.CL, cs.AI, cs.LG
Language models have been shown to exhibit positive scaling, where performance improves as models are scaled up in terms of size, compute, or data. In this work, we introduce NeQA, a dataset consisting of questions with negation in which language models do not exhibit straightforward positive scaling. We show that this...
http://arxiv.org/pdf/2305.17311v1
2023-05-27T00:07:17
2023-05-27T00:07:17
{ "title": [ -0.009918213, -0.009178162, -0.034729004, -0.061920166, -0.00541687, -0.012039185, -0.04333496, -0.0045204163, 0.0054969788, 0.047607422, -0.03869629, -0.0069084167, 0.032287598, -0.0065689087, -0.060424805, -0.00080156326, -0.014984131,...
http://arxiv.org/abs/2305.17306v1
Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance
Yao Fu, Litu Ou, Mingyu Chen, Yuhao Wan, Hao Peng, Tushar Khot
cs.CL, cs.AI, cs.LG
As large language models (LLMs) are continuously being developed, their evaluation becomes increasingly important yet challenging. This work proposes Chain-of-Thought Hub, an open-source evaluation suite on the multi-step reasoning capabilities of large language models. We are interested in this setting for two reasons...
http://arxiv.org/pdf/2305.17306v1
2023-05-26T23:46:42
2023-05-26T23:46:42
{ "title": [ -0.010223389, -0.01966858, -0.023498535, -0.057159424, -0.02633667, -0.0046539307, -0.049316406, -0.006061554, 0.008460999, 0.020370483, -0.039367676, -0.008422852, 0.04397583, -0.028869629, -0.016662598, -0.000059604645, -0.0077819824, ...
http://arxiv.org/abs/2305.17304v1
External Language Model Integration for Factorized Neural Transducers
Michael Levit, Sarangarajan Parthasarathy, Cem Aksoylar, Mohammad Sadegh Rasooli, Shuangyu Chang
cs.CL
We propose an adaptation method for factorized neural transducers (FNT) with external language models. We demonstrate that both neural and n-gram external LMs add significantly more value when linearly interpolated with predictor output compared to shallow fusion, thus confirming that FNT forces the predictor to act li...
http://arxiv.org/pdf/2305.17304v1
2023-05-26T23:30:21
2023-05-26T23:30:21
{ "title": [ -0.0009741783, 0.007965088, -0.022003174, -0.053863525, 0.0009727478, 0.0069999695, -0.05404663, -0.014411926, -0.020828247, 0.035949707, 0.0035152435, -0.015213013, 0.012489319, -0.032348633, -0.026016235, 0.006072998, -0.005176544, ...
http://arxiv.org/abs/2305.17280v1
Improved Instruction Ordering in Recipe-Grounded Conversation
Duong Minh Le, Ruohao Guo, Wei Xu, Alan Ritter
cs.CL
In this paper, we study the task of instructional dialogue and focus on the cooking domain. Analyzing the generated output of the GPT-J model, we reveal that the primary challenge for a recipe-grounded dialog system is how to provide the instructions in the correct order. We hypothesize that this is due to the model's ...
http://arxiv.org/pdf/2305.17280v1
2023-05-26T21:57:11
2023-05-26T21:57:11
{ "title": [ -0.013671875, 0.0060424805, -0.054901123, -0.07312012, 0.011505127, -0.0076522827, -0.055511475, 0.018585205, -0.02015686, 0.039367676, -0.00043535233, 0.014503479, 0.019042969, -0.030563354, -0.01033783, -0.00035595894, -0.0209198, ...
http://arxiv.org/abs/2306.00739v3
SQL-PaLM: Improved Large Language Model Adaptation for Text-to-SQL
Ruoxi Sun, Sercan O. Arik, Hootan Nakhost, Hanjun Dai, Rajarishi Sinha, Pengcheng Yin, Tomas Pfister
cs.CL, cs.AI, cs.DB
One impressive emergent capability of large language models (LLMs) is generation of code, including Structured Query Language (SQL) for databases. For the task of converting natural language text to SQL queries, Text-to-SQL, adaptation of LLMs is of paramount importance, both in in-context learning and fine-tuning sett...
http://arxiv.org/pdf/2306.00739v3
2023-06-25T06:44:48
2023-05-26T21:39:05
{ "title": [ -0.03286743, -0.020370483, -0.04977417, -0.0758667, 0.013298035, 0.011787415, -0.050567627, 0.025360107, 0.028656006, 0.043426514, -0.024841309, 0.0015001297, 0.018051147, -0.00062179565, -0.06713867, 0.00522995, -0.0017528534, 0.016...
http://arxiv.org/abs/2305.17273v1
Slide, Constrain, Parse, Repeat: Synchronous SlidingWindows for Document AMR Parsing
Sadhana Kumaravel, Tahira Naseem, Ramon Fernandez Astudillo, Radu Florian, Salim Roukos
cs.CL, cs.AI
The sliding window approach provides an elegant way to handle contexts of sizes larger than the Transformer's input window, for tasks like language modeling. Here we extend this approach to the sequence-to-sequence task of document parsing. For this, we exploit recent progress in transition-based parsing to implement a...
http://arxiv.org/pdf/2305.17273v1
2023-05-26T21:38:08
2023-05-26T21:38:08
{ "title": [ -0.011100769, -0.039367676, -0.031463623, -0.07745361, -0.04727173, 0.002986908, -0.07702637, 0.020889282, -0.02104187, 0.05407715, -0.031143188, 0.021347046, 0.01789856, -0.022094727, -0.034484863, -0.033477783, -0.0039749146, 0.028...
http://arxiv.org/abs/2305.17268v1
Metaphor Detection via Explicit Basic Meanings Modelling
Yucheng Li, Shun Wang, Chenghua Lin, Guerin Frank
cs.CL
One noticeable trend in metaphor detection is the embrace of linguistic theories such as the metaphor identification procedure (MIP) for model architecture design. While MIP clearly defines that the metaphoricity of a lexical unit is determined based on the contrast between its \textit{contextual meaning} and its \text...
http://arxiv.org/pdf/2305.17268v1
2023-05-26T21:25:05
2023-05-26T21:25:05
{ "title": [ -0.008621216, 0.004749298, -0.0030117035, -0.008277893, 0.0090408325, 0.0038166046, -0.018096924, -0.047729492, 0.019638062, 0.04284668, -0.014709473, -0.005168915, 0.028137207, -0.009651184, -0.0552063, 0.021591187, -0.019012451, 0....
http://arxiv.org/abs/2305.17267v1
CODET: A Benchmark for Contrastive Dialectal Evaluation of Machine Translation
Md Mahfuz Ibn Alam, Sina Ahmadi, Antonios Anastasopoulos
cs.CL
Neural machine translation (NMT) systems exhibit limited robustness in handling source-side linguistic variations. Their performance tends to degrade when faced with even slight deviations in language usage, such as different domains or variations introduced by second-language speakers. It is intuitive to extend this o...
http://arxiv.org/pdf/2305.17267v1
2023-05-26T21:24:00
2023-05-26T21:24:00
{ "title": [ -0.020462036, -0.007095337, 0.011260986, -0.040863037, -0.011886597, 0.010185242, -0.032165527, -0.012367249, 0.01977539, 0.053863525, -0.018371582, 0.020095825, 0.02772522, -0.009262085, -0.068847656, 0.025054932, -0.0049324036, 0.0...
http://arxiv.org/abs/2305.17266v2
Honey, I Shrunk the Language: Language Model Behavior at Reduced Scale
Vijeta Deshpande, Dan Pechi, Shree Thatte, Vladislav Lialin, Anna Rumshisky
cs.CL
In recent years, language models have drastically grown in size, and the abilities of these models have been shown to improve with scale. The majority of recent scaling laws studies focused on high-compute high-parameter count settings, leaving the question of when these abilities begin to emerge largely unanswered. In...
http://arxiv.org/pdf/2305.17266v2
2023-05-30T18:37:32
2023-05-26T21:22:10
{ "title": [ -0.048583984, -0.015655518, -0.054718018, -0.032165527, -0.003156662, -0.008285522, -0.030181885, 0.0058250427, -0.00793457, 0.032806396, -0.02128601, -0.007106781, 0.020812988, -0.027618408, -0.044036865, 0.010391235, -0.010383606, ...
http://arxiv.org/abs/2305.17256v2
Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learning
Ruixiang Tang, Dehan Kong, Longtao Huang, Hui Xue
cs.CL, cs.AI, cs.LG
Large language models (LLMs) have recently shown great potential for in-context learning, where LLMs learn a new task simply by conditioning on a few input-label pairs (prompts). Despite their potential, our understanding of the factors influencing end-task performance and the robustness of in-context learning remains ...
http://arxiv.org/pdf/2305.17256v2
2023-09-09T18:32:00
2023-05-26T20:56:30
{ "title": [ -0.009056091, 0.0011386871, -0.028915405, -0.009162903, -0.0056495667, -0.037994385, 0.024093628, -0.0072402954, -0.029449463, 0.03048706, -0.041046143, -0.008407593, -0.018173218, -0.02394104, -0.018569946, -0.05606079, -0.0004582405, ...
http://arxiv.org/abs/2305.17221v1
Federated Learning for Semantic Parsing: Task Formulation, Evaluation Setup, New Algorithms
Tianshu Zhang, Changchang Liu, Wei-Han Lee, Yu Su, Huan Sun
cs.CL, cs.AI, cs.DB, cs.LG
This paper studies a new task of federated learning (FL) for semantic parsing, where multiple clients collaboratively train one global model without sharing their semantic parsing data. By leveraging data from multiple clients, the FL paradigm can be especially beneficial for clients that have little training data to d...
http://arxiv.org/pdf/2305.17221v1
2023-05-26T19:25:49
2023-05-26T19:25:49
{ "title": [ -0.02243042, 0.011169434, -0.062927246, -0.10021973, 0.02609253, 0.0064430237, -0.056915283, 0.021316528, -0.032226562, 0.031082153, -0.039733887, 0.035247803, -0.024337769, 0.006248474, -0.033721924, -0.024795532, -0.017303467, 0.04...
http://arxiv.org/abs/2305.17219v1
GVdoc: Graph-based Visual Document Classification
Fnu Mohbat, Mohammed J. Zaki, Catherine Finegan-Dollak, Ashish Verma
cs.CV, cs.CL, cs.LG
The robustness of a model for real-world deployment is decided by how well it performs on unseen data and distinguishes between in-domain and out-of-domain samples. Visual document classifiers have shown impressive performance on in-distribution test sets. However, they tend to have a hard time correctly classifying an...
http://arxiv.org/pdf/2305.17219v1
2023-05-26T19:23:20
2023-05-26T19:23:20
{ "title": [ -0.020599365, -0.008056641, -0.03817749, -0.044189453, -0.0035953522, -0.00166893, -0.032196045, -0.01411438, 0.0076942444, 0.008087158, -0.042175293, 0.016998291, -0.025253296, -0.0059394836, -0.040771484, -0.021072388, -0.008903503, ...
http://arxiv.org/abs/2305.17216v3
Generating Images with Multimodal Language Models
Jing Yu Koh, Daniel Fried, Ruslan Salakhutdinov
cs.CL, cs.CV, cs.LG
We propose a method to fuse frozen text-only large language models (LLMs) with pre-trained image encoder and decoder models, by mapping between their embedding spaces. Our model demonstrates a wide suite of multimodal capabilities: image retrieval, novel image generation, and multimodal dialogue. Ours is the first appr...
http://arxiv.org/pdf/2305.17216v3
2023-10-13T15:35:42
2023-05-26T19:22:03
{ "title": [ -0.016647339, 0.0025196075, -0.025039673, -0.05026245, -0.002090454, 0.004142761, -0.056518555, -0.025543213, 0.02659607, 0.038879395, -0.0015439987, 0.0026302338, 0.0005364418, -0.020584106, -0.03955078, 0.017074585, -0.004432678, 0...
http://arxiv.org/abs/2305.17211v1
Coping with low data availability for social media crisis message categorisation
Congcong Wang
cs.CL
During crisis situations, social media allows people to quickly share information, including messages requesting help. This can be valuable to emergency responders, who need to categorise and prioritise these messages based on the type of assistance being requested. However, the high volume of messages makes it difficu...
http://arxiv.org/pdf/2305.17211v1
2023-05-26T19:08:24
2023-05-26T19:08:24
{ "title": [ 0.033355713, 0.010154724, -0.022003174, -0.050872803, -0.0028038025, -0.027954102, -0.004650116, -0.014297485, 0.0013713837, 0.035095215, -0.02670288, -0.002922058, 0.012809753, -0.042510986, -0.03286743, -0.025405884, 0.0009660721, ...
http://arxiv.org/abs/2305.17202v1
BIG-C: a Multimodal Multi-Purpose Dataset for Bemba
Claytone Sikasote, Eunice Mukonde, Md Mahfuz Ibn Alam, Antonios Anastasopoulos
cs.CL
We present BIG-C (Bemba Image Grounded Conversations), a large multimodal dataset for Bemba. While Bemba is the most populous language of Zambia, it exhibits a dearth of resources which render the development of language technologies or language processing research almost impossible. The dataset is comprised of multi-t...
http://arxiv.org/pdf/2305.17202v1
2023-05-26T18:49:55
2023-05-26T18:49:55
{ "title": [ -0.029373169, -0.03994751, -0.030059814, -0.08703613, -0.03692627, 0.024612427, -0.10510254, -0.006477356, 0.018661499, 0.043762207, -0.0013847351, -0.015113831, 0.0048942566, -0.04446411, -0.035858154, 0.017547607, 0.004776001, 0.04...
http://arxiv.org/abs/2305.19280v1
Large language models improve Alzheimer's disease diagnosis using multi-modality data
Yingjie Feng, Jun Wang, Xianfeng Gu, Xiaoyin Xu, Min Zhang
cs.LG, cs.AI, cs.CL, cs.CV
In diagnosing challenging conditions such as Alzheimer's disease (AD), imaging is an important reference. Non-imaging patient data such as patient information, genetic data, medication information, cognitive and memory tests also play a very important role in diagnosis. Effect. However, limited by the ability of artifi...
http://arxiv.org/pdf/2305.19280v1
2023-05-26T18:42:19
2023-05-26T18:42:19
{ "title": [ -0.016784668, 0.005870819, -0.06222534, -0.068725586, -0.002981186, -0.012313843, -0.06951904, -0.0024318695, 0.0045433044, 0.03149414, -0.030044556, -0.010803223, 0.015625, -0.032806396, -0.032073975, -0.00043845177, -0.0022659302, ...
http://arxiv.org/abs/2305.17197v1
Entailment as Robust Self-Learner
Jiaxin Ge, Hongyin Luo, Yoon Kim, James Glass
cs.CL
Entailment has been recognized as an important metric for evaluating natural language understanding (NLU) models, and recent studies have found that entailment pretraining benefits weakly supervised fine-tuning. In this work, we design a prompting strategy that formulates a number of different NLU tasks as contextual e...
http://arxiv.org/pdf/2305.17197v1
2023-05-26T18:41:23
2023-05-26T18:41:23
{ "title": [ 0.011688232, 0.014381409, -0.043701172, -0.091552734, 0.022033691, -0.004673004, -0.037384033, -0.000562191, -0.00019431114, 0.015899658, -0.009269714, 0.011077881, -0.025817871, -0.016189575, -0.03765869, 0.015975952, -0.0042152405, ...
http://arxiv.org/abs/2305.17182v2
On the Copying Problem of Unsupervised NMT: A Training Schedule with a Language Discriminator Loss
Yihong Liu, Alexandra Chronopoulou, Hinrich Schütze, Alexander Fraser
cs.CL, cs.AI
Although unsupervised neural machine translation (UNMT) has achieved success in many language pairs, the copying problem, i.e., directly copying some parts of the input sentence as the translation, is common among distant language pairs, especially when low-resource languages are involved. We find this issue is closely...
http://arxiv.org/pdf/2305.17182v2
2023-06-04T09:41:35
2023-05-26T18:14:23
{ "title": [ -0.0093307495, 0.024765015, -0.0068511963, -0.036102295, 0.009399414, 0.0047798157, -0.036346436, -0.04537964, -0.023757935, 0.018127441, -0.010177612, 0.030975342, -0.009109497, -0.029006958, 0.0154953, -0.003206253, -0.016494751, 0...
http://arxiv.org/abs/2305.17179v1
Tokenization Impacts Multilingual Language Modeling: Assessing Vocabulary Allocation and Overlap Across Languages
Tomasz Limisiewicz, Jiří Balhar, David Mareček
cs.CL
Multilingual language models have recently gained attention as a promising solution for representing multiple languages in a single model. In this paper, we propose new criteria to evaluate the quality of lexical representation and vocabulary overlap observed in sub-word tokenizers. Our findings show that the overlap o...
http://arxiv.org/pdf/2305.17179v1
2023-05-26T18:06:49
2023-05-26T18:06:49
{ "title": [ -0.036254883, 0.010269165, -0.04473877, -0.025802612, -0.017791748, -0.006580353, -0.059020996, -0.015411377, -0.026519775, 0.033233643, -0.033447266, 0.022155762, 0.023910522, -0.026168823, -0.0019130707, 0.0015058517, -0.02368164, ...
http://arxiv.org/abs/2305.18357v1
DeepSI: Interactive Deep Learning for Semantic Interaction
Yali Bian, Chris North
cs.LG, cs.AI, cs.CL, cs.HC
In this paper, we design novel interactive deep learning methods to improve semantic interactions in visual analytics applications. The ability of semantic interaction to infer analysts' precise intents during sensemaking is dependent on the quality of the underlying data representation. We propose the $\text{DeepSI}_{...
http://arxiv.org/pdf/2305.18357v1
2023-05-26T18:05:57
2023-05-26T18:05:57
{ "title": [ -0.015052795, -0.014076233, -0.040405273, -0.056030273, 0.0033130646, -0.0062561035, -0.03781128, -0.011978149, 0.014335632, 0.0054016113, -0.024627686, -0.010261536, -0.015991211, -0.027618408, -0.03729248, -0.006313324, -0.0015354156, ...
http://arxiv.org/abs/2305.17174v1
From Dogwhistles to Bullhorns: Unveiling Coded Rhetoric with Language Models
Julia Mendelsohn, Ronan Le Bras, Yejin Choi, Maarten Sap
cs.CL, cs.CY
Dogwhistles are coded expressions that simultaneously convey one meaning to a broad audience and a second one, often hateful or provocative, to a narrow in-group; they are deployed to evade both political repercussions and algorithmic content moderation. For example, in the sentence 'we need to end the cosmopolitan exp...
http://arxiv.org/pdf/2305.17174v1
2023-05-26T18:00:57
2023-05-26T18:00:57
{ "title": [ -0.012832642, -0.004776001, -0.00806427, -0.026473999, -0.004032135, 0.0045394897, -0.015960693, -0.02507019, -0.02229309, 0.03062439, 0.008346558, 0.00932312, -0.0005121231, -0.02128601, -0.04107666, -0.004421234, -0.026016235, 0.06...
http://arxiv.org/abs/2305.17131v1
RAMP: Retrieval and Attribute-Marking Enhanced Prompting for Attribute-Controlled Translation
Gabriele Sarti, Phu Mon Htut, Xing Niu, Benjamin Hsu, Anna Currey, Georgiana Dinu, Maria Nadejde
cs.CL
Attribute-controlled translation (ACT) is a subtask of machine translation that involves controlling stylistic or linguistic attributes (like formality and gender) of translation outputs. While ACT has garnered attention in recent years due to its usefulness in real-world applications, progress in the task is currently...
http://arxiv.org/pdf/2305.17131v1
2023-05-26T17:56:53
2023-05-26T17:56:53
{ "title": [ -0.031982422, -0.007133484, -0.00037956238, -0.04727173, -0.021255493, 0.01889038, -0.04550171, -0.020324707, -0.02734375, 0.033294678, 0.016052246, 0.02355957, 0.02609253, -0.04135132, -0.020950317, 0.021240234, -0.00033140182, -0.0...
http://arxiv.org/abs/2305.17127v1
Characterizing and Measuring Linguistic Dataset Drift
Tyler A. Chang, Kishaloy Halder, Neha Anna John, Yogarshi Vyas, Yassine Benajiba, Miguel Ballesteros, Dan Roth
cs.CL
NLP models often degrade in performance when real world data distributions differ markedly from training data. However, existing dataset drift metrics in NLP have generally not considered specific dimensions of linguistic drift that affect model performance, and they have not been validated in their ability to predict ...
http://arxiv.org/pdf/2305.17127v1
2023-05-26T17:50:51
2023-05-26T17:50:51
{ "title": [ -0.015312195, 0.0023269653, -0.04058838, -0.037017822, 0.009803772, -0.01701355, -0.019317627, -0.010131836, 0.0027103424, 0.024810791, -0.026535034, 0.0030174255, -0.025817871, -0.009422302, -0.06213379, -0.014801025, -0.0064086914, ...
http://arxiv.org/abs/2305.17126v1
Large Language Models as Tool Makers
Tianle Cai, Xuezhi Wang, Tengyu Ma, Xinyun Chen, Denny Zhou
cs.LG, cs.AI, cs.CL, stat.ML
Recent research shows the potential of enhancing the problem-solving ability of large language models (LLMs) through the use of external tools. However, prior work along this line depends on the availability of existing tools. In this work, we take an initial step towards removing this dependency by proposing a closed-...
http://arxiv.org/pdf/2305.17126v1
2023-05-26T17:50:11
2023-05-26T17:50:11
{ "title": [ -0.032104492, -0.02041626, -0.03387451, -0.0703125, -0.03668213, 0.009521484, -0.08325195, 0.016921997, -0.0128479, 0.031021118, -0.0079956055, -0.017349243, -0.0053482056, -0.02722168, -0.024810791, 0.00919342, 0.007369995, 0.035827...
http://arxiv.org/abs/2305.17118v2
Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time
Zichang Liu, Aditya Desai, Fangshuo Liao, Weitao Wang, Victor Xie, Zhaozhuo Xu, Anastasios Kyrillidis, Anshumali Shrivastava
cs.LG, cs.CL
Large language models(LLMs) have sparked a new wave of exciting AI applications. Hosting these models at scale requires significant memory resources. One crucial memory bottleneck for the deployment stems from the context window. It is commonly recognized that model weights are memory hungry; however, the size of key-v...
http://arxiv.org/pdf/2305.17118v2
2023-08-28T22:48:46
2023-05-26T17:39:58
{ "title": [ -0.0690918, -0.022460938, -0.034057617, -0.026489258, -0.044830322, 0.007663727, -0.025878906, 0.0390625, -0.0030441284, 0.070007324, 0.009864807, 0.011375427, 0.029541016, -0.023727417, -0.03451538, 0.012390137, 0.011405945, 0.03268...
http://arxiv.org/abs/2305.17116v2
Improving accuracy of GPT-3/4 results on biomedical data using a retrieval-augmented language model
David Soong, Sriram Sridhar, Han Si, Jan-Samuel Wagner, Ana Caroline Costa Sá, Christina Y Yu, Kubra Karagoz, Meijian Guan, Hisham Hamadeh, Brandon W Higgs
cs.CL, cs.AI
Large language models (LLMs) have made significant advancements in natural language processing (NLP). Broad corpora capture diverse patterns but can introduce irrelevance, while focused corpora enhance reliability by reducing misleading information. Training LLMs on focused corpora poses computational challenges. An al...
http://arxiv.org/pdf/2305.17116v2
2023-05-30T15:37:45
2023-05-26T17:33:05
{ "title": [ 0.000579834, 0.010787964, -0.024658203, -0.045532227, -0.007549286, 0.0028533936, -0.05883789, -0.023849487, 0.015533447, 0.030456543, -0.012237549, -0.0031528473, -0.0012016296, -0.029525757, -0.053649902, 0.011177063, -0.015235901, ...
http://arxiv.org/abs/2305.17104v1
PromptNER: Prompt Locating and Typing for Named Entity Recognition
Yongliang Shen, Zeqi Tan, Shuhui Wu, Wenqi Zhang, Rongsheng Zhang, Yadong Xi, Weiming Lu, Yueting Zhuang
cs.CL
Prompt learning is a new paradigm for utilizing pre-trained language models and has achieved great success in many tasks. To adopt prompt learning in the NER task, two kinds of methods have been explored from a pair of symmetric perspectives, populating the template by enumerating spans to predict their entity types or...
http://arxiv.org/pdf/2305.17104v1
2023-05-26T17:16:11
2023-05-26T17:16:11
{ "title": [ -0.03363037, 0.0007610321, -0.047027588, -0.047027588, 0.009010315, -0.0005559921, -0.04208374, -0.017593384, 0.012229919, 0.013031006, -0.046722412, 0.023513794, 0.020065308, -0.010063171, -0.024353027, 0.0049476624, -0.011741638, 0...
http://arxiv.org/abs/2305.17100v1
BiomedGPT: A Unified and Generalist Biomedical Generative Pre-trained Transformer for Vision, Language, and Multimodal Tasks
Kai Zhang, Jun Yu, Zhiling Yan, Yixin Liu, Eashan Adhikarla, Sunyang Fu, Xun Chen, Chen Chen, Yuyin Zhou, Xiang Li, Lifang He, Brian D. Davison, Quanzheng Li, Yong Chen, Hongfang Liu, Lichao Sun
cs.CL, cs.AI
In this paper, we introduce a unified and generalist Biomedical Generative Pre-trained Transformer (BiomedGPT) model, which leverages self-supervision on large and diverse datasets to accept multi-modal inputs and perform a range of downstream tasks. Our experiments demonstrate that BiomedGPT delivers expansive and inc...
http://arxiv.org/pdf/2305.17100v1
2023-05-26T17:14:43
2023-05-26T17:14:43
{ "title": [ 0.010017395, -0.00217247, -0.04348755, -0.06298828, 0.004901886, -0.022277832, -0.055725098, -0.0072784424, 0.002779007, 0.0357666, -0.0039978027, 0.0054359436, 0.024169922, -0.031951904, -0.01576233, 0.014511108, 0.0062026978, 0.039...
http://arxiv.org/abs/2305.17080v1
Expand, Rerank, and Retrieve: Query Reranking for Open-Domain Question Answering
Yung-Sung Chuang, Wei Fang, Shang-Wen Li, Wen-tau Yih, James Glass
cs.CL
We propose EAR, a query Expansion And Reranking approach for improving passage retrieval, with the application to open-domain question answering. EAR first applies a query expansion model to generate a diverse set of queries, and then uses a query reranker to select the ones that could lead to better retrieval results....
http://arxiv.org/pdf/2305.17080v1
2023-05-26T16:41:03
2023-05-26T16:41:03
{ "title": [ -0.00035381317, 0.00504303, -0.029724121, -0.06402588, 0.019851685, 0.0022907257, -0.011245728, -0.010627747, 0.009567261, 0.035980225, -0.024307251, 0.022903442, 0.010726929, 0.0015382767, -0.048309326, 0.003540039, -0.0013341904, 0...
http://arxiv.org/abs/2305.17077v1
Learning and Leveraging Verifiers to Improve Planning Capabilities of Pre-trained Language Models
Daman Arora, Subbarao Kambhampati
cs.CL, cs.AI
There have been wide spread claims in the literature about the emergent reasoning capabilities of Pretrained Large Language Models. However, recent studies, have found that their ability to plan remains questionable. Through our experiments using GPT-2, we empirically demonstrate that the performance of a finetuned bas...
http://arxiv.org/pdf/2305.17077v1
2023-05-26T16:36:55
2023-05-26T16:36:55
{ "title": [ -0.003326416, -0.016464233, -0.039154053, -0.095336914, -0.008003235, -0.001964569, -0.085754395, -0.0038433075, -0.019241333, 0.013572693, -0.031082153, 0.007621765, -0.010345459, -0.015808105, 0.0030403137, -0.019241333, -0.010635376, ...
http://arxiv.org/abs/2305.17075v1
CREST: A Joint Framework for Rationalization and Counterfactual Text Generation
Marcos Treviso, Alexis Ross, Nuno M. Guerreiro, André F. T. Martins
cs.CL
Selective rationales and counterfactual examples have emerged as two effective, complementary classes of interpretability methods for analyzing and training NLP models. However, prior work has not explored how these methods can be integrated to combine their complementary advantages. We overcome this limitation by intr...
http://arxiv.org/pdf/2305.17075v1
2023-05-26T16:34:58
2023-05-26T16:34:58
{ "title": [ -0.004497528, -0.005004883, -0.01449585, -0.038146973, -0.022781372, 0.017211914, -0.053833008, -0.04135132, 0.0056915283, 0.024261475, -0.017288208, 0.03414917, -0.0129470825, -0.014839172, -0.038238525, -0.00049209595, -0.0284729, ...
http://arxiv.org/abs/2305.17073v1
NeuroX Library for Neuron Analysis of Deep NLP Models
Fahim Dalvi, Hassan Sajjad, Nadir Durrani
cs.CL
Neuron analysis provides insights into how knowledge is structured in representations and discovers the role of neurons in the network. In addition to developing an understanding of our models, neuron analysis enables various applications such as debiasing, domain adaptation and architectural search. We present NeuroX,...
http://arxiv.org/pdf/2305.17073v1
2023-05-26T16:32:56
2023-05-26T16:32:56
{ "title": [ -0.020019531, -0.010757446, -0.04611206, -0.027786255, -0.0010223389, -0.023452759, -0.021453857, -0.014564514, 0.003660202, 0.022735596, -0.028198242, -0.006011963, 0.019821167, -0.02458191, -0.03463745, 0.0124435425, -0.0004312992, ...
http://arxiv.org/abs/2305.18354v1
LLMs and the Abstraction and Reasoning Corpus: Successes, Failures, and the Importance of Object-based Representations
Yudong Xu, Wenhao Li, Pashootan Vaezipoor, Scott Sanner, Elias B. Khalil
cs.CL, cs.AI
Can a Large Language Model (LLM) solve simple abstract reasoning problems? We explore this broad question through a systematic analysis of GPT on the Abstraction and Reasoning Corpus (ARC), a representative benchmark of abstract reasoning ability from limited examples in which solutions require some "core knowledge" of...
http://arxiv.org/pdf/2305.18354v1
2023-05-26T16:32:17
2023-05-26T16:32:17
{ "title": [ -0.003921509, -0.002981186, -0.041107178, -0.0552063, -0.009925842, -0.013412476, -0.04055786, -0.005569458, -0.006549835, 0.020126343, -0.061767578, 0.0013513565, 0.022583008, -0.02885437, -0.00045084953, -0.024230957, -0.01777649, ...
http://arxiv.org/abs/2305.17072v1
Stereotypes and Smut: The (Mis)representation of Non-cisgender Identities by Text-to-Image Models
Eddie L. Ungless, Björn Ross, Anne Lauscher
cs.CL, cs.CY
Cutting-edge image generation has been praised for producing high-quality images, suggesting a ubiquitous future in a variety of applications. However, initial studies have pointed to the potential for harm due to predictive bias, reflecting and potentially reinforcing cultural stereotypes. In this work, we are the fir...
http://arxiv.org/pdf/2305.17072v1
2023-05-26T16:28:49
2023-05-26T16:28:49
{ "title": [ 0.020858765, 0.024749756, -0.02772522, -0.0035171509, 0.017929077, -0.032989502, 0.0335083, -0.039245605, 0.010726929, 0.050628662, -0.03326416, 0.0013656616, 0.022537231, -0.014297485, -0.04638672, -0.015022278, -0.033599854, 0.0396...
http://arxiv.org/abs/2305.17066v1
Mindstorms in Natural Language-Based Societies of Mind
Mingchen Zhuge, Haozhe Liu, Francesco Faccio, Dylan R. Ashley, Róbert Csordás, Anand Gopalakrishnan, Abdullah Hamdi, Hasan Abed Al Kader Hammoud, Vincent Herrmann, Kazuki Irie, Louis Kirsch, Bing Li, Guohao Li, Shuming Liu, Jinjie Mai, Piotr Piękos, Aditya Ramesh, Imanol Schlag, Weimin Shi, Aleksandar Stanić, Wenyi Wan...
cs.AI, cs.CL, cs.CV, cs.LG, cs.MA, 68T07, I.2.6; I.2.11
Both Minsky's "society of mind" and Schmidhuber's "learning to think" inspire diverse societies of large multimodal neural networks (NNs) that solve problems by interviewing each other in a "mindstorm." Recent implementations of NN-based societies of minds consist of large language models (LLMs) and other NN-based expe...
http://arxiv.org/pdf/2305.17066v1
2023-05-26T16:21:25
2023-05-26T16:21:25
{ "title": [ -0.019851685, -0.019226074, -0.015335083, -0.043884277, -0.00491333, 0.006526947, -0.0033931732, 0.00042152405, 0.0050811768, 0.03237915, -0.057769775, 0.0032520294, -0.0121154785, 0.0015439987, -0.05545044, -0.0071754456, -0.013801575, ...
http://arxiv.org/abs/2305.17055v1
Counterfactuals of Counterfactuals: a back-translation-inspired approach to analyse counterfactual editors
Giorgos Filandrianos, Edmund Dervakos, Orfeas Menis-Mastromichalakis, Chrysoula Zerva, Giorgos Stamou
cs.CL
In the wake of responsible AI, interpretability methods, which attempt to provide an explanation for the predictions of neural models have seen rapid progress. In this work, we are concerned with explanations that are applicable to natural language processing (NLP) models and tasks, and we focus specifically on the ana...
http://arxiv.org/pdf/2305.17055v1
2023-05-26T16:04:28
2023-05-26T16:04:28
{ "title": [ -0.02128601, -0.0031147003, -0.00007420778, -0.0014333725, -0.026779175, -0.0057754517, -0.041503906, -0.061462402, 0.007896423, 0.04019165, 0.0036029816, -0.002790451, 0.009437561, -0.04107666, -0.04949951, 0.016021729, -0.018997192, ...
http://arxiv.org/abs/2305.17050v1
Exploiting Abstract Meaning Representation for Open-Domain Question Answering
Cunxiang Wang, Zhikun Xu, Qipeng Guo, Xiangkun Hu, Xuefeng Bai, Zheng Zhang, Yue Zhang
cs.CL
The Open-Domain Question Answering (ODQA) task involves retrieving and subsequently generating answers from fine-grained relevant passages within a database. Current systems leverage Pretrained Language Models (PLMs) to model the relationship between questions and passages. However, the diversity in surface form expres...
http://arxiv.org/pdf/2305.17050v1
2023-05-26T16:00:16
2023-05-26T16:00:16
{ "title": [ 0.0012397766, 0.0025863647, -0.017959595, -0.054260254, 0.0053634644, 0.0041770935, -0.03604126, -0.017425537, 0.022476196, 0.030288696, -0.005970001, 0.015792847, -0.0014076233, -0.0049972534, -0.05456543, 0.01676941, -0.010627747, ...
http://arxiv.org/abs/2305.17041v1
RFiD: Towards Rational Fusion-in-Decoder for Open-Domain Question Answering
Cunxiang Wang, Haofei Yu, Yue Zhang
cs.CL, cs.AI
Open-Domain Question Answering (ODQA) systems necessitate a reader model capable of generating answers by simultaneously referring to multiple passages. Although representative models like Fusion-in-Decoder (FiD) have been proposed to address this challenge, these systems can inadvertently rely on spurious features ins...
http://arxiv.org/pdf/2305.17041v1
2023-05-26T15:51:25
2023-05-26T15:51:25
{ "title": [ -0.006259918, 0.0039482117, -0.025039673, -0.05505371, 0.004447937, -0.0043640137, -0.0413208, -0.01525116, 0.00015926361, 0.029067993, -0.029510498, -0.007472992, 0.019546509, -0.012916565, -0.04776001, -0.013572693, -0.011604309, 0...
http://arxiv.org/abs/2305.17040v1
A Mechanism for Sample-Efficient In-Context Learning for Sparse Retrieval Tasks
Jacob Abernethy, Alekh Agarwal, Teodor V. Marinov, Manfred K. Warmuth
cs.LG, cs.CL
We study the phenomenon of \textit{in-context learning} (ICL) exhibited by large language models, where they can adapt to a new learning task, given a handful of labeled examples, without any explicit parameter optimization. Our goal is to explain how a pre-trained transformer model is able to perform ICL under reasona...
http://arxiv.org/pdf/2305.17040v1
2023-05-26T15:49:43
2023-05-26T15:49:43
{ "title": [ 0.005973816, -0.0069885254, -0.034179688, -0.046875, 0.01727295, -0.016815186, 0.00541687, 0.006149292, 0.037872314, 0.03353882, -0.03604126, 0.0036182404, -0.0011281967, -0.0056266785, -0.04156494, 0.012718201, 0.013458252, -0.00896...
http://arxiv.org/abs/2305.17026v2
On the Computational Power of Decoder-Only Transformer Language Models
Jesse Roberts
cs.CL, cs.LG
This article presents a theoretical evaluation of the computational universality of decoder-only transformer models. We extend the theoretical literature on transformer models and show that decoder-only transformer architectures (even with only a single layer and single attention head) are Turing complete under reasona...
http://arxiv.org/pdf/2305.17026v2
2023-05-30T21:58:24
2023-05-26T15:35:43
{ "title": [ -0.031219482, -0.00065755844, -0.027893066, -0.056732178, -0.0052871704, 0.0005273819, -0.06591797, -0.0037231445, 0.00017380714, 0.050628662, -0.002292633, -0.010726929, 0.03778076, -0.02078247, -0.043060303, 0.037261963, 0.0036087036, ...
http://arxiv.org/abs/2305.17020v3
Diable: Efficient Dialogue State Tracking as Operations on Tables
Pietro Lesci, Yoshinari Fujinuma, Momchil Hardalov, Chao Shang, Yassine Benajiba, Lluis Marquez
cs.CL, cs.LG
Sequence-to-sequence state-of-the-art systems for dialogue state tracking (DST) use the full dialogue history as input, represent the current state as a list with all the slots, and generate the entire state from scratch at each dialogue turn. This approach is inefficient, especially when the number of slots is large a...
http://arxiv.org/pdf/2305.17020v3
2023-11-01T20:36:38
2023-05-26T15:26:12
{ "title": [ -0.007575989, -0.025497437, -0.038848877, -0.09063721, 0.0026836395, 0.00024938583, -0.05532837, 0.00013768673, 0.005470276, 0.021972656, -0.009742737, 0.014381409, 0.009727478, -0.014144897, -0.026763916, 0.006515503, -0.014289856, ...
http://arxiv.org/abs/2305.17019v1
Commonsense Knowledge Graph Completion Via Contrastive Pretraining and Node Clustering
Siwei Wu, Xiangqing Shen, Rui Xia
cs.CL
The nodes in the commonsense knowledge graph (CSKG) are normally represented by free-form short text (e.g., word or phrase). Different nodes may represent the same concept. This leads to the problems of edge sparsity and node redundancy, which challenges CSKG representation and completion. On the one hand, edge sparsit...
http://arxiv.org/pdf/2305.17019v1
2023-05-26T15:24:32
2023-05-26T15:24:32
{ "title": [ 0.01638794, -0.01876831, -0.033569336, -0.07824707, -0.012565613, 0.0011167526, -0.045928955, 0.007209778, 0.019226074, 0.02961731, -0.031463623, 0.023605347, -0.018844604, -0.008476257, -0.03994751, -0.003955841, -0.0045776367, 0.05...
http://arxiv.org/abs/2305.17013v1
D-CALM: A Dynamic Clustering-based Active Learning Approach for Mitigating Bias
Sabit Hassan, Malihe Alikhani
cs.CL
Despite recent advancements, NLP models continue to be vulnerable to bias. This bias often originates from the uneven distribution of real-world data and can propagate through the annotation process. Escalated integration of these models in our lives calls for methods to mitigate bias without overbearing annotation cos...
http://arxiv.org/pdf/2305.17013v1
2023-05-26T15:17:43
2023-05-26T15:17:43
{ "title": [ 0.01826477, 0.0030117035, -0.035858154, -0.03503418, 0.03012085, -0.020965576, 0.007881165, -0.034362793, -0.013145447, 0.02142334, 0.012466431, -0.0060272217, -0.031021118, -0.04699707, -0.016448975, -0.0038661957, -0.0020885468, 0....
http://arxiv.org/abs/2305.17008v2
NormBank: A Knowledge Bank of Situational Social Norms
Caleb Ziems, Jane Dwivedi-Yu, Yi-Chia Wang, Alon Halevy, Diyi Yang
cs.CL
We present NormBank, a knowledge bank of 155k situational norms. This resource is designed to ground flexible normative reasoning for interactive, assistive, and collaborative AI systems. Unlike prior commonsense resources, NormBank grounds each inference within a multivalent sociocultural frame, which includes the set...
http://arxiv.org/pdf/2305.17008v2
2023-07-24T19:18:25
2023-05-26T15:09:11
{ "title": [ -0.0059013367, 0.0034866333, -0.049987793, -0.06817627, 0.02029419, -0.01701355, -0.030838013, -0.014701843, 0.022506714, 0.023605347, -0.046569824, 0.013900757, 0.0038890839, -0.005783081, -0.0496521, 0.006401062, -0.010215759, 0.03...
http://arxiv.org/abs/2305.17006v1
Zero-shot Visual Question Answering with Language Model Feedback
Yifan Du, Junyi Li, Tianyi Tang, Wayne Xin Zhao, Ji-Rong Wen
cs.CV, cs.CL
In this paper, we propose a novel language model guided captioning approach, LAMOC, for knowledge-based visual question answering (VQA). Our approach employs the generated captions by a captioning model as the context of an answer prediction model, which is a Pre-trained Language model (PLM). As the major contribution,...
http://arxiv.org/pdf/2305.17006v1
2023-05-26T15:04:20
2023-05-26T15:04:20
{ "title": [ -0.017593384, -0.016174316, -0.006679535, -0.048095703, -0.03591919, 0.0030784607, -0.05328369, -0.00044345856, 0.048095703, 0.03491211, -0.028533936, -0.0011434555, -0.0055770874, 0.0007634163, -0.07672119, 0.012619019, 0.027374268, ...
http://arxiv.org/abs/2305.17002v1
An Empirical Comparison of LM-based Question and Answer Generation Methods
Asahi Ushio, Fernando Alva-Manchego, Jose Camacho-Collados
cs.CL
Question and answer generation (QAG) consists of generating a set of question-answer pairs given a context (e.g. a paragraph). This task has a variety of applications, such as data augmentation for question answering (QA) models, information retrieval and education. In this paper, we establish baselines with three diff...
http://arxiv.org/pdf/2305.17002v1
2023-05-26T14:59:53
2023-05-26T14:59:53
{ "title": [ -0.027526855, -0.004676819, -0.027679443, -0.018447876, -0.01171875, 0.0029315948, -0.023773193, -0.015533447, 0.003124237, 0.0062217712, -0.016296387, 0.023010254, -0.040496826, -0.015625, -0.039001465, -0.002784729, -0.005558014, 0...
http://arxiv.org/abs/2305.16986v3
NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models
Gengze Zhou, Yicong Hong, Qi Wu
cs.CV, cs.AI, cs.CL, cs.RO
Trained with an unprecedented scale of data, large language models (LLMs) like ChatGPT and GPT-4 exhibit the emergence of significant reasoning abilities from model scaling. Such a trend underscored the potential of training LLMs with unlimited language data, advancing the development of a universal embodied agent. In ...
http://arxiv.org/pdf/2305.16986v3
2023-10-19T17:59:43
2023-05-26T14:41:06
{ "title": [ -0.009849548, -0.026107788, -0.010894775, -0.058746338, -0.0024356842, -0.003643036, -0.030639648, -0.0104904175, 0.024978638, 0.00894165, -0.03994751, -0.009902954, -0.010398865, -0.020217896, -0.0262146, -0.012626648, 0.0036373138, ...
http://arxiv.org/abs/2305.16982v1
TranSFormer: Slow-Fast Transformer for Machine Translation
Bei Li, Yi Jing, Xu Tan, Zhen Xing, Tong Xiao, Jingbo Zhu
cs.CL, cs.AI
Learning multiscale Transformer models has been evidenced as a viable approach to augmenting machine translation systems. Prior research has primarily focused on treating subwords as basic units in developing such systems. However, the incorporation of fine-grained character-level features into multiscale Transformer h...
http://arxiv.org/pdf/2305.16982v1
2023-05-26T14:37:38
2023-05-26T14:37:38
{ "title": [ -0.043426514, -0.0065574646, -0.03189087, -0.028244019, -0.013038635, -0.0072364807, -0.02784729, -0.011695862, -0.0060539246, 0.0491333, -0.023208618, 0.008934021, 0.052886963, -0.02255249, -0.02407837, 0.023620605, 0.0044517517, 0....
http://arxiv.org/abs/2305.16967v3
Evaluating Open-Domain Dialogues in Latent Space with Next Sentence Prediction and Mutual Information
Kun Zhao, Bohao Yang, Chenghua Lin, Wenge Rong, Aline Villavicencio, Xiaohui Cui
cs.CL
The long-standing one-to-many issue of the open-domain dialogues poses significant challenges for automatic evaluation methods, i.e., there may be multiple suitable responses which differ in semantics for a given conversational context. To tackle this challenge, we propose a novel learning-based automatic evaluation me...
http://arxiv.org/pdf/2305.16967v3
2023-06-10T13:23:41
2023-05-26T14:21:54
{ "title": [ -0.0034294128, 0.013694763, -0.029434204, -0.03640747, -0.0037231445, -0.013648987, -0.031799316, -0.020614624, -0.00944519, 0.022155762, -0.015106201, 0.025253296, -0.015235901, -0.007965088, -0.046020508, -0.019073486, -0.014312744, ...
http://arxiv.org/abs/2305.16960v3
Training Socially Aligned Language Models on Simulated Social Interactions
Ruibo Liu, Ruixin Yang, Chenyan Jia, Ge Zhang, Denny Zhou, Andrew M. Dai, Diyi Yang, Soroush Vosoughi
cs.CL, cs.AI, cs.CY, cs.HC
Social alignment in AI systems aims to ensure that these models behave according to established societal values. However, unlike humans, who derive consensus on value judgments through social interaction, current language models (LMs) are trained to rigidly replicate their training corpus in isolation, leading to subpa...
http://arxiv.org/pdf/2305.16960v3
2023-10-28T09:02:39
2023-05-26T14:17:36
{ "title": [ -0.02168274, 0.0003669262, -0.038024902, -0.040008545, -0.010169983, -0.010353088, -0.057800293, -0.009063721, -0.012825012, 0.028503418, 0.0022621155, 0.0007138252, 0.0018520355, -0.022354126, -0.024993896, 0.007293701, -0.015357971, ...
http://arxiv.org/abs/2305.16958v1
MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies
Shiyue Zhang, Shijie Wu, Ozan Irsoy, Steven Lu, Mohit Bansal, Mark Dredze, David Rosenberg
cs.CL, cs.AI, cs.LG
Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P -- that is, minimizing the forward cross-entropy, which is equivalent to maximum likelihood estimation (MLE). We have observed that models trained in this way may "over-generalize",...
http://arxiv.org/pdf/2305.16958v1
2023-05-26T14:14:51
2023-05-26T14:14:51
{ "title": [ -0.03189087, -0.0031776428, -0.053619385, -0.054779053, 0.013923645, 0.00074243546, -0.0385437, 0.007549286, -0.00031781197, 0.040283203, -0.010536194, 0.030181885, 0.01008606, -0.007358551, -0.049316406, 0.018112183, -0.012573242, 0...
http://arxiv.org/abs/2305.16957v1
DisfluencyFixer: A tool to enhance Language Learning through Speech To Speech Disfluency Correction
Vineet Bhat, Preethi Jyothi, Pushpak Bhattacharyya
cs.CL, cs.LG, cs.SD, eess.AS
Conversational speech often consists of deviations from the speech plan, producing disfluent utterances that affect downstream NLP tasks. Removing these disfluencies is necessary to create fluent and coherent speech. This paper presents DisfluencyFixer, a tool that performs speech-to-speech disfluency correction in Eng...
http://arxiv.org/pdf/2305.16957v1
2023-05-26T14:13:38
2023-05-26T14:13:38
{ "title": [ 0.014312744, 0.0026474, -0.053771973, -0.029403687, -0.0071411133, -0.032043457, -0.027877808, 0.0016307831, 0.012016296, 0.042144775, -0.019378662, -0.027511597, 0.05908203, -0.035461426, -0.029144287, -0.008399963, -0.018478394, 0....
http://arxiv.org/abs/2305.16954v1
Compositional Generalization without Trees using Multiset Tagging and Latent Permutations
Matthias Lindemann, Alexander Koller, Ivan Titov
cs.CL
Seq2seq models have been shown to struggle with compositional generalization in semantic parsing, i.e. generalizing to unseen compositions of phenomena that the model handles correctly in isolation. We phrase semantic parsing as a two-step process: we first tag each input token with a multiset of output tokens. Then ...
http://arxiv.org/pdf/2305.16954v1
2023-05-26T14:09:35
2023-05-26T14:09:35
{ "title": [ -0.0024051666, 0.0007534027, -0.034210205, -0.099365234, -0.004371643, 0.016403198, -0.0892334, 0.011962891, 0.0039787292, 0.035369873, -0.018478394, 0.03955078, -0.04800415, -0.009429932, -0.04159546, -0.008224487, -0.018554688, 0.0...
http://arxiv.org/abs/2305.16947v1
Sentence-Incremental Neural Coreference Resolution
Matt Grenander, Shay B. Cohen, Mark Steedman
cs.CL
We propose a sentence-incremental neural coreference resolution system which incrementally builds clusters after marking mention boundaries in a shift-reduce method. The system is aimed at bridging two recent approaches at coreference resolution: (1) state-of-the-art non-incremental models that incur quadratic complexi...
http://arxiv.org/pdf/2305.16947v1
2023-05-26T14:00:25
2023-05-26T14:00:25
{ "title": [ 0.021560669, 0.0063972473, -0.022857666, -0.08557129, 0.010116577, -0.0087890625, -0.041046143, 0.0021419525, 0.011619568, 0.042907715, -0.030303955, 0.007160187, 0.023162842, -0.004058838, -0.03439331, 0.025726318, -0.010185242, 0.0...