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/2305.12449v1
Communication Efficient Federated Learning for Multilingual Neural Machine Translation with Adapter
Yi Liu, Xiaohan Bi, Lei Li, Sishuo Chen, Wenkai Yang, Xu Sun
cs.CL, cs.AI
Federated Multilingual Neural Machine Translation (Fed-MNMT) has emerged as a promising paradigm for institutions with limited language resources. This approach allows multiple institutions to act as clients and train a unified model through model synchronization, rather than collecting sensitive data for centralized t...
http://arxiv.org/pdf/2305.12449v1
2023-05-21T12:48:38
2023-05-21T12:48:38
{ "title": [ -0.026519775, 0.009178162, -0.038330078, -0.057556152, 0.01675415, -0.0017871857, -0.024002075, 0.01071167, -0.022491455, 0.05529785, -0.007904053, -0.0023708344, 0.019638062, -0.0073509216, -0.027572632, 0.0034236908, -0.0016040802, ...
http://arxiv.org/abs/2305.12441v2
A Pilot Study on Dialogue-Level Dependency Parsing for Chinese
Gongyao Jiang, Shuang Liu, Meishan Zhang, Min Zhang
cs.CL
Dialogue-level dependency parsing has received insufficient attention, especially for Chinese. To this end, we draw on ideas from syntactic dependency and rhetorical structure theory (RST), developing a high-quality human-annotated corpus, which contains 850 dialogues and 199,803 dependencies. Considering that such tas...
http://arxiv.org/pdf/2305.12441v2
2023-06-01T03:03:43
2023-05-21T12:20:13
{ "title": [ -0.015823364, -0.012901306, -0.030914307, -0.09100342, 0.0044403076, 0.011695862, -0.06945801, 0.03616333, -0.017333984, 0.014465332, -0.006450653, 0.0067253113, -0.021453857, -0.028366089, -0.04244995, -0.010505676, -0.029281616, 0....
http://arxiv.org/abs/2305.11916v1
F-PABEE: Flexible-patience-based Early Exiting for Single-label and Multi-label text Classification Tasks
Xiangxiang Gao, Wei Zhu, Jiasheng Gao, Congrui Yin
cs.CL
Computational complexity and overthinking problems have become the bottlenecks for pre-training language models (PLMs) with millions or even trillions of parameters. A Flexible-Patience-Based Early Exiting method (F-PABEE) has been proposed to alleviate the problems mentioned above for single-label classification (SLC)...
http://arxiv.org/pdf/2305.11916v1
2023-05-21T12:17:27
2023-05-21T12:17:27
{ "title": [ -0.021957397, -0.017120361, -0.0138549805, -0.02293396, -0.01309967, -0.0037879944, -0.0003695488, -0.016723633, 0.006664276, 0.022567749, -0.02772522, 0.014289856, -0.0115356445, -0.01574707, -0.044555664, -0.022781372, -0.023635864, ...
http://arxiv.org/abs/2305.12434v1
BiasAsker: Measuring the Bias in Conversational AI System
Yuxuan Wan, Wenxuan Wang, Pinjia He, Jiazhen Gu, Haonan Bai, Michael Lyu
cs.CL, cs.AI
Powered by advanced Artificial Intelligence (AI) techniques, conversational AI systems, such as ChatGPT and digital assistants like Siri, have been widely deployed in daily life. However, such systems may still produce content containing biases and stereotypes, causing potential social problems. Due to the data-driven,...
http://arxiv.org/pdf/2305.12434v1
2023-05-21T11:25:59
2023-05-21T11:25:59
{ "title": [ -0.014221191, 0.0073280334, -0.057800293, -0.015945435, 0.012924194, -0.03161621, -0.0057907104, -0.03564453, 0.020584106, 0.022583008, -0.041870117, -0.0047683716, 0.017807007, -0.014915466, -0.03918457, -0.003643036, -0.020324707, ...
http://arxiv.org/abs/2305.12421v4
Evaluating Open-QA Evaluation
Cunxiang Wang, Sirui Cheng, Qipeng Guo, Yuanhao Yue, Bowen Ding, Zhikun Xu, Yidong Wang, Xiangkun Hu, Zheng Zhang, Yue Zhang
cs.CL, cs.AI
This study focuses on the evaluation of the Open Question Answering (Open-QA) task, which can directly estimate the factuality of large language models (LLMs). Current automatic evaluation methods have shown limitations, indicating that human evaluation still remains the most reliable approach. We introduce a new task,...
http://arxiv.org/pdf/2305.12421v4
2023-10-23T14:26:26
2023-05-21T10:40:55
{ "title": [ 0.001326561, 0.015777588, -0.013954163, -0.0079422, -0.020217896, -0.018478394, -0.011802673, -0.034362793, -0.031829834, 0.05206299, -0.01687622, -0.020126343, -0.02268982, -0.036956787, -0.020736694, -0.01675415, 0.0068626404, 0.02...
http://arxiv.org/abs/2305.12412v1
EM Pre-training for Multi-party Dialogue Response Generation
Yiyang Li, Hai Zhao
cs.CL
Dialogue response generation requires an agent to generate a response according to the current dialogue history, in terms of which two-party dialogues have been well studied, but leaving a great gap for multi-party dialogues at the same time. Different from two-party dialogues where each response is a direct reply to i...
http://arxiv.org/pdf/2305.12412v1
2023-05-21T09:22:41
2023-05-21T09:22:41
{ "title": [ -0.0011940002, -0.0104904175, -0.026931763, -0.060577393, -0.005935669, -0.011993408, -0.04824829, 0.011444092, -0.026901245, 0.03668213, 0.030181885, 0.013366699, 0.003376007, -0.035858154, -0.03125, 0.009819031, -0.010757446, 0.037...
http://arxiv.org/abs/2305.12401v1
WOT-Class: Weakly Supervised Open-world Text Classification
Tianle Wang, Zihan Wang, Weitang Liu, Jingbo Shang
cs.CL
State-of-the-art weakly supervised text classification methods, while significantly reduced the required human supervision, still requires the supervision to cover all the classes of interest. This is never easy to meet in practice when human explore new, large corpora without complete pictures. In this paper, we work ...
http://arxiv.org/pdf/2305.12401v1
2023-05-21T08:51:24
2023-05-21T08:51:24
{ "title": [ -0.006378174, 0.006084442, -0.024230957, -0.048065186, 0.005584717, 0.00077199936, -0.020233154, -0.027297974, -0.012886047, 0.02406311, -0.0340271, 0.033172607, -0.00819397, -0.0004963875, -0.017929077, -0.0154953, -0.008323669, 0.0...
http://arxiv.org/abs/2305.12394v1
Pruning Pre-trained Language Models with Principled Importance and Self-regularization
Siyu Ren, Kenny Q. Zhu
cs.CL
Iterative pruning is one of the most effective compression methods for pre-trained language models. We discovered that finding the optimal pruning decision is an equality-constrained 0-1 Integer Linear Programming problem. The solution to this optimization problem leads to a principled importance criterion which we use...
http://arxiv.org/pdf/2305.12394v1
2023-05-21T08:15:12
2023-05-21T08:15:12
{ "title": [ -0.016281128, -0.007637024, -0.034301758, -0.07489014, -0.0022296906, 0.007255554, -0.06549072, 0.016098022, -0.0008225441, 0.02696228, 0.008483887, 0.00004941225, 0.0029392242, -0.025482178, -0.03942871, 0.015655518, -0.005207062, 0...
http://arxiv.org/abs/2305.12392v1
PiVe: Prompting with Iterative Verification Improving Graph-based Generative Capability of LLMs
Jiuzhou Han, Nigel Collier, Wray Buntine, Ehsan Shareghi
cs.CL
Large language models (LLMs) have shown great abilities of solving various natural language tasks in different domains. Due to the training objective of LLMs and their pretraining data, LLMs are not very well equipped for tasks involving structured data generation. We propose a framework, Prompting with Iterative Verif...
http://arxiv.org/pdf/2305.12392v1
2023-05-21T08:11:24
2023-05-21T08:11:24
{ "title": [ -0.058166504, -0.037963867, -0.04333496, -0.06732178, -0.04534912, 0.018951416, -0.107055664, 0.01763916, 0.0009531975, -0.0032043457, -0.0234375, 0.023529053, -0.0052337646, -0.040130615, 0.01322937, -0.0014600754, -0.013557434, 0.0...
http://arxiv.org/abs/2305.12389v1
SHINE: Syntax-augmented Hierarchical Interactive Encoder for Zero-shot Cross-lingual Information Extraction
Jun-Yu Ma, Jia-Chen Gu, Zhen-Hua Ling, Quan Liu, Cong Liu, Guoping Hu
cs.CL
Zero-shot cross-lingual information extraction(IE) aims at constructing an IE model for some low-resource target languages, given annotations exclusively in some rich-resource languages. Recent studies based on language-universal features have shown their effectiveness and are attracting increasing attention. However, ...
http://arxiv.org/pdf/2305.12389v1
2023-05-21T08:02:06
2023-05-21T08:02:06
{ "title": [ -0.01247406, -0.0146865845, -0.04269409, -0.083984375, -0.010795593, -0.002708435, -0.048980713, 0.010154724, 0.025177002, 0.038146973, -0.015838623, 0.017242432, 0.0084991455, -0.0041618347, -0.03967285, 0.022766113, 0.0018358231, 0...
http://arxiv.org/abs/2305.12376v1
Measuring Intersectional Biases in Historical Documents
Nadav Borenstein, Karolina Stańczak, Thea Rolskov, Natália da Silva Perez, Natacha Klein Käfer, Isabelle Augenstein
cs.CL, cs.CY, cs.LG
Data-driven analyses of biases in historical texts can help illuminate the origin and development of biases prevailing in modern society. However, digitised historical documents pose a challenge for NLP practitioners as these corpora suffer from errors introduced by optical character recognition (OCR) and are written...
http://arxiv.org/pdf/2305.12376v1
2023-05-21T07:10:31
2023-05-21T07:10:31
{ "title": [ -0.01058197, 0.008460999, -0.035125732, -0.014297485, -0.0028209686, -0.02848816, 0.00035095215, -0.02494812, 0.02078247, 0.03491211, -0.055114746, 0.01663208, 0.036376953, -0.007499695, -0.034118652, 0.008766174, -0.017959595, 0.041...
http://arxiv.org/abs/2305.12371v1
Machine Translation by Projecting Text into the Same Phonetic-Orthographic Space Using a Common Encoding
Amit Kumar, Shantipriya Parida, Ajay Pratap, Anil Kumar Singh
cs.CL
The use of subword embedding has proved to be a major innovation in Neural Machine Translation (NMT). It helps NMT to learn better context vectors for Low Resource Languages (LRLs) so as to predict the target words by better modelling the morphologies of the two languages and also the morphosyntax transfer. Even so, th...
http://arxiv.org/pdf/2305.12371v1
2023-05-21T06:46:33
2023-05-21T06:46:33
{ "title": [ -0.0071525574, -0.009880066, 0.012687683, 0.0012273788, -0.043060303, -0.005332947, -0.0062789917, -0.020446777, -0.032928467, 0.070251465, -0.020721436, -0.0001887083, 0.06414795, -0.035339355, 0.009063721, -0.012763977, 0.0055007935, ...
http://arxiv.org/abs/2305.12351v2
Are Your Explanations Reliable? Investigating the Stability of LIME in Explaining Text Classifiers by Marrying XAI and Adversarial Attack
Christopher Burger, Lingwei Chen, Thai Le
cs.LG, cs.CL
LIME has emerged as one of the most commonly referenced tools in explainable AI (XAI) frameworks that is integrated into critical machine learning applications--e.g., healthcare and finance. However, its stability remains little explored, especially in the context of text data, due to the unique text-space constraints....
http://arxiv.org/pdf/2305.12351v2
2023-10-15T13:19:44
2023-05-21T05:06:46
{ "title": [ -0.030380249, 0.0064353943, -0.013671875, 0.020324707, -0.021133423, -0.010009766, -0.021972656, -0.04446411, -0.050933838, 0.009735107, -0.0068855286, 0.045532227, -0.052947998, -0.01449585, -0.010017395, -0.04574585, -0.022857666, ...
http://arxiv.org/abs/2305.12330v1
Task-agnostic Distillation of Encoder-Decoder Language Models
Chen Zhang, Yang Yang, Jingang Wang, Dawei Song
cs.CL
Finetuning pretrained language models (LMs) have enabled appealing performance on a diverse array of tasks. The intriguing task-agnostic property has driven a shifted focus from task-specific to task-agnostic distillation of LMs. While task-agnostic, compute-efficient, performance-preserved LMs can be yielded by task-a...
http://arxiv.org/pdf/2305.12330v1
2023-05-21T03:35:45
2023-05-21T03:35:45
{ "title": [ -0.03353882, 0.009269714, -0.0635376, -0.06378174, -0.0045280457, -0.0051193237, -0.08294678, 0.0055236816, 0.028335571, 0.04788208, -0.0011138916, -0.032287598, 0.025299072, -0.015777588, -0.06555176, 0.020080566, 0.00044059753, 0.0...
http://arxiv.org/abs/2305.13338v2
Gene Set Summarization using Large Language Models
Marcin P. Joachimiak, J. Harry Caufield, Nomi L. Harris, Hyeongsik Kim, Christopher J. Mungall
q-bio.GN, cs.AI, cs.CL, q-bio.QM
Molecular biologists frequently interpret gene lists derived from high-throughput experiments and computational analysis. This is typically done as a statistical enrichment analysis that measures the over- or under-representation of biological function terms associated with genes or their properties, based on curated a...
http://arxiv.org/pdf/2305.13338v2
2023-05-25T19:10:13
2023-05-21T02:06:33
{ "title": [ -0.043792725, -0.010246277, -0.057922363, -0.0680542, -0.010063171, 0.015525818, -0.08190918, 0.009971619, 0.031677246, 0.03201294, 0.000790596, -0.024780273, 0.0041656494, -0.024047852, -0.05947876, 0.02319336, -0.0042686462, 0.0311...
http://arxiv.org/abs/2305.12311v1
i-Code V2: An Autoregressive Generation Framework over Vision, Language, and Speech Data
Ziyi Yang, Mahmoud Khademi, Yichong Xu, Reid Pryzant, Yuwei Fang, Chenguang Zhu, Dongdong Chen, Yao Qian, Mei Gao, Yi-Ling Chen, Robert Gmyr, Naoyuki Kanda, Noel Codella, Bin Xiao, Yu Shi, Lu Yuan, Takuya Yoshioka, Michael Zeng, Xuedong Huang
cs.CL, cs.AI, cs.CV, cs.LG, eess.AS
The convergence of text, visual, and audio data is a key step towards human-like artificial intelligence, however the current Vision-Language-Speech landscape is dominated by encoder-only models which lack generative abilities. We propose closing this gap with i-Code V2, the first model capable of generating natural la...
http://arxiv.org/pdf/2305.12311v1
2023-05-21T01:25:44
2023-05-21T01:25:44
{ "title": [ -0.017425537, -0.017166138, -0.047851562, -0.058502197, -0.0029315948, -0.0046920776, -0.05126953, 0.004196167, 0.010810852, 0.033691406, -0.030059814, 0.010688782, 0.005622864, -0.010681152, -0.055358887, -0.00566864, 0.006038666, 0...
http://arxiv.org/abs/2305.12307v1
OntoType: Ontology-Guided Zero-Shot Fine-Grained Entity Typing with Weak Supervision from Pre-Trained Language Models
Tanay Komarlu, Minhao Jiang, Xuan Wang, Jiawei Han
cs.CL
Fine-grained entity typing (FET), which assigns entities in text with context-sensitive, fine-grained semantic types, will play an important role in natural language understanding. A supervised FET method, which typically relies on human-annotated corpora for training, is costly and difficult to scale. Recent studies l...
http://arxiv.org/pdf/2305.12307v1
2023-05-21T00:32:37
2023-05-21T00:32:37
{ "title": [ -0.008514404, -0.024490356, -0.03729248, -0.09265137, 0.0018901825, 0.005218506, -0.055541992, 0.018447876, 0.014457703, 0.01651001, -0.021835327, 0.017303467, 0.033233643, -0.020019531, -0.0051498413, 0.0154418945, -0.024246216, 0.0...
http://arxiv.org/abs/2305.12301v1
Sentence Embedder Guided Utterance Encoder (SEGUE) for Spoken Language Understanding
Yi Xuan Tan, Navonil Majumder, Soujanya Poria
cs.CL, cs.AI, cs.SD, eess.AS
The pre-trained speech encoder wav2vec 2.0 performs very well on various spoken language understanding (SLU) tasks. However, on many tasks, it trails behind text encoders with textual input. To improve the understanding capability of SLU encoders, various studies have used knowledge distillation to transfer knowledge f...
http://arxiv.org/pdf/2305.12301v1
2023-05-20T23:55:55
2023-05-20T23:55:55
{ "title": [ -0.0060920715, -0.012374878, -0.025878906, -0.02507019, -0.037994385, -0.014762878, -0.046936035, 0.0023498535, -0.0066337585, 0.031204224, -0.0446167, 0.0053138733, 0.042419434, -0.027420044, -0.001502037, -0.010421753, -0.0052261353, ...
http://arxiv.org/abs/2305.12295v2
Logic-LM: Empowering Large Language Models with Symbolic Solvers for Faithful Logical Reasoning
Liangming Pan, Alon Albalak, Xinyi Wang, William Yang Wang
cs.CL, cs.AI
Large Language Models (LLMs) have shown human-like reasoning abilities but still struggle with complex logical problems. This paper introduces a novel framework, Logic-LM, which integrates LLMs with symbolic solvers to improve logical problem-solving. Our method first utilizes LLMs to translate a natural language probl...
http://arxiv.org/pdf/2305.12295v2
2023-10-19T01:54:27
2023-05-20T22:25:38
{ "title": [ -0.039520264, -0.011955261, -0.054870605, -0.08831787, 0.008934021, 0.021987915, -0.088134766, 0.008071899, -0.010681152, 0.033966064, -0.0113220215, 0.01209259, 0.018356323, -0.017303467, -0.026351929, 0.019439697, -0.0154800415, 0....
http://arxiv.org/abs/2305.12289v1
Revisiting the Architectures like Pointer Networks to Efficiently Improve the Next Word Distribution, Summarization Factuality, and Beyond
Haw-Shiuan Chang, Zonghai Yao, Alolika Gon, Hong Yu, Andrew McCallum
cs.CL
Is the output softmax layer, which is adopted by most language models (LMs), always the best way to compute the next word probability? Given so many attention layers in a modern transformer-based LM, are the pointer networks redundant nowadays? In this study, we discover that the answers to both questions are no. This ...
http://arxiv.org/pdf/2305.12289v1
2023-05-20T21:52:24
2023-05-20T21:52:24
{ "title": [ -0.0033950806, -0.012916565, -0.00081300735, -0.0069770813, -0.046203613, -0.007843018, -0.02218628, -0.022903442, 0.0058784485, 0.07556152, -0.027893066, 0.04425049, 0.020935059, 0.02809143, -0.032226562, 0.00957489, 0.0009717941, -...
http://arxiv.org/abs/2305.12281v1
Lifelong Language Pretraining with Distribution-Specialized Experts
Wuyang Chen, Yanqi Zhou, Nan Du, Yanping Huang, James Laudon, Zhifeng Chen, Claire Cu
cs.CL, cs.LG
Pretraining on a large-scale corpus has become a standard method to build general language models (LMs). Adapting a model to new data distributions targeting different downstream tasks poses significant challenges. Naive fine-tuning may incur catastrophic forgetting when the over-parameterized LMs overfit the new data ...
http://arxiv.org/pdf/2305.12281v1
2023-05-20T21:15:19
2023-05-20T21:15:19
{ "title": [ 0.012138367, 0.008125305, -0.05432129, -0.065979004, -0.028656006, -0.017120361, -0.03591919, 0.02571106, -0.03665161, 0.034576416, 0.014175415, 0.014007568, 0.01878357, -0.064086914, 0.018600464, 0.007980347, 0.0021476746, 0.0524292...
http://arxiv.org/abs/2305.12280v2
Contextualizing Argument Quality Assessment with Relevant Knowledge
Darshan Deshpande, Zhivar Sourati, Filip Ilievski, Fred Morstatter
cs.CL
Automatic assessment of the quality of arguments has been recognized as a challenging task with significant implications for misinformation and targeted speech. While real-world arguments are tightly anchored in context, existing computational methods analyze their quality in isolation, which affects their accuracy and...
http://arxiv.org/pdf/2305.12280v2
2023-11-08T18:41:44
2023-05-20T21:04:58
{ "title": [ 0.0385437, 0.00995636, -0.026016235, -0.09625244, 0.018569946, -0.0070724487, -0.025222778, -0.00793457, -0.020111084, 0.028259277, 0.0032291412, 0.0012779236, -0.034210205, -0.027679443, -0.03665161, -0.013832092, -0.015205383, 0.04...
http://arxiv.org/abs/2305.12276v1
Analogy in Contact: Modeling Maltese Plural Inflection
Sara Court, Andrea D. Sims, Micha Elsner
cs.CL
Maltese is often described as having a hybrid morphological system resulting from extensive contact between Semitic and Romance language varieties. Such a designation reflects an etymological divide as much as it does a larger tradition in the literature to consider concatenative and non-concatenative morphological pat...
http://arxiv.org/pdf/2305.12276v1
2023-05-20T20:16:57
2023-05-20T20:16:57
{ "title": [ 0.020843506, -0.011505127, -0.003276825, -0.04006958, -0.020828247, -0.010437012, -0.008056641, -0.0016765594, 0.02357483, 0.05770874, -0.01473999, -0.019485474, 0.029220581, -0.02218628, -0.05328369, 0.00084257126, -0.011001587, 0.0...
http://arxiv.org/abs/2305.12272v1
Autoregressive Modeling with Lookahead Attention
Li Du, Hongyuan Mei, Jason Eisner
cs.CL, cs.AI, cs.LG
To predict the next token, autoregressive models ordinarily examine the past. Could they also benefit from also examining hypothetical futures? We consider a novel Transformer-based autoregressive architecture that estimates the next-token distribution by extrapolating multiple continuations of the past, according to s...
http://arxiv.org/pdf/2305.12272v1
2023-05-20T19:29:47
2023-05-20T19:29:47
{ "title": [ -0.0057296753, 0.0076065063, -0.048980713, -0.061462402, 0.019058228, 0.0009784698, -0.03765869, -0.01499939, 0.008888245, 0.04107666, -0.022155762, -0.0055656433, -0.008079529, -0.010063171, -0.06530762, 0.0025806427, 0.007045746, 0...
http://arxiv.org/abs/2305.12268v1
Patton: Language Model Pretraining on Text-Rich Networks
Bowen Jin, Wentao Zhang, Yu Zhang, Yu Meng, Xinyang Zhang, Qi Zhu, Jiawei Han
cs.CL, cs.AI, cs.LG
A real-world text corpus sometimes comprises not only text documents but also semantic links between them (e.g., academic papers in a bibliographic network are linked by citations and co-authorships). Text documents and semantic connections form a text-rich network, which empowers a wide range of downstream tasks such ...
http://arxiv.org/pdf/2305.12268v1
2023-05-20T19:17:10
2023-05-20T19:17:10
{ "title": [ -0.030227661, -0.008010864, -0.031051636, -0.044403076, -0.014221191, 0.0090789795, -0.051513672, -0.006061554, -0.014793396, 0.027954102, -0.010643005, 0.027709961, -0.005367279, -0.016021729, -0.027633667, 0.012229919, -0.016708374, ...
http://arxiv.org/abs/2305.12263v2
Self-supervised representations in speech-based depression detection
Wen Wu, Chao Zhang, Philip C. Woodland
cs.CL, cs.SD, eess.AS
This paper proposes handling training data sparsity in speech-based automatic depression detection (SDD) using foundation models pre-trained with self-supervised learning (SSL). An analysis of SSL representations derived from different layers of pre-trained foundation models is first presented for SDD, which provides i...
http://arxiv.org/pdf/2305.12263v2
2023-07-06T13:43:56
2023-05-20T18:41:13
{ "title": [ 0.032073975, 0.0050354004, -0.0026416779, -0.08062744, 0.011459351, -0.0039901733, -0.035461426, 0.0012903214, -0.0029144287, 0.051605225, -0.0005283356, -0.027709961, 0.015716553, -0.033996582, -0.057250977, -0.003074646, -0.002866745, ...
http://arxiv.org/abs/2305.12260v2
Cross2StrA: Unpaired Cross-lingual Image Captioning with Cross-lingual Cross-modal Structure-pivoted Alignment
Shengqiong Wu, Hao Fei, Wei Ji, Tat-Seng Chua
cs.CV, cs.CL
Unpaired cross-lingual image captioning has long suffered from irrelevancy and disfluency issues, due to the inconsistencies of the semantic scene and syntax attributes during transfer. In this work, we propose to address the above problems by incorporating the scene graph (SG) structures and the syntactic constituency...
http://arxiv.org/pdf/2305.12260v2
2023-05-25T04:02:17
2023-05-20T18:30:03
{ "title": [ -0.020095825, -0.0020389557, -0.024246216, -0.030654907, -0.0152282715, -0.0082092285, -0.023452759, -0.03665161, 0.014762878, 0.058044434, -0.015380859, 0.012313843, 0.04208374, -0.016967773, -0.016296387, 0.027038574, -0.006542206, ...
http://arxiv.org/abs/2305.12258v3
Constructing Code-mixed Universal Dependency Forest for Unbiased Cross-lingual Relation Extraction
Hao Fei, Meishan Zhang, Min Zhang, Tat-Seng Chua
cs.CL
Latest efforts on cross-lingual relation extraction (XRE) aggressively leverage the language-consistent structural features from the universal dependency (UD) resource, while they may largely suffer from biased transfer (e.g., either target-biased or source-biased) due to the inevitable linguistic disparity between lan...
http://arxiv.org/pdf/2305.12258v3
2023-06-04T13:29:51
2023-05-20T18:24:06
{ "title": [ -0.010803223, -0.034423828, -0.04373169, -0.094177246, -0.0073623657, 0.010307312, -0.07171631, 0.025817871, 0.024414062, 0.021377563, -0.023284912, 0.026306152, -0.004016876, -0.0068588257, -0.030761719, 0.016448975, -0.024612427, 0...
http://arxiv.org/abs/2305.12257v1
SEntFiN 1.0: Entity-Aware Sentiment Analysis for Financial News
Ankur Sinha, Satishwar Kedas, Rishu Kumar, Pekka Malo
cs.CL, cs.AI, cs.LG, I.2.7
Fine-grained financial sentiment analysis on news headlines is a challenging task requiring human-annotated datasets to achieve high performance. Limited studies have tried to address the sentiment extraction task in a setting where multiple entities are present in a news headline. In an effort to further research in t...
http://arxiv.org/pdf/2305.12257v1
2023-05-20T18:20:39
2023-05-20T18:20:39
{ "title": [ -0.018554688, -0.007987976, -0.05001831, -0.040222168, 0.03100586, -0.008018494, 0.00032687187, -0.004840851, 0.02519226, 0.037902832, -0.023895264, -0.004989624, 0.004749298, 0.0053253174, -0.076171875, 0.005268097, -0.02067566, 0.0...
http://arxiv.org/abs/2305.12256v2
Scene Graph as Pivoting: Inference-time Image-free Unsupervised Multimodal Machine Translation with Visual Scene Hallucination
Hao Fei, Qian Liu, Meishan Zhang, Min Zhang, Tat-Seng Chua
cs.CL
In this work, we investigate a more realistic unsupervised multimodal machine translation (UMMT) setup, inference-time image-free UMMT, where the model is trained with source-text image pairs, and tested with only source-text inputs. First, we represent the input images and texts with the visual and language scene grap...
http://arxiv.org/pdf/2305.12256v2
2023-05-25T04:24:34
2023-05-20T18:17:20
{ "title": [ -0.019958496, -0.0181427, -0.0029468536, -0.03540039, -0.04989624, -0.0011692047, -0.03604126, -0.01399231, 0.012199402, 0.041778564, -0.02027893, 0.017227173, 0.018234253, -0.021255493, -0.01725769, 0.009338379, 0.005004883, 0.03991...
http://arxiv.org/abs/2305.12248v1
Brain encoding models based on multimodal transformers can transfer across language and vision
Jerry Tang, Meng Du, Vy A. Vo, Vasudev Lal, Alexander G. Huth
cs.CL, cs.CV
Encoding models have been used to assess how the human brain represents concepts in language and vision. While language and vision rely on similar concept representations, current encoding models are typically trained and tested on brain responses to each modality in isolation. Recent advances in multimodal pretraining...
http://arxiv.org/pdf/2305.12248v1
2023-05-20T17:38:44
2023-05-20T17:38:44
{ "title": [ 0.020446777, 0.00030946732, -0.01398468, 0.005115509, -0.04537964, -0.037841797, -0.019683838, -0.04058838, -0.0037879944, 0.037017822, 0.003370285, -0.018844604, 0.046813965, -0.057769775, 0.004901886, -0.00018155575, 0.013626099, 0...
http://arxiv.org/abs/2305.12233v1
A Measure of Explanatory Effectiveness
Dylan Cope, Peter McBurney
cs.CL
In most conversations about explanation and AI, the recipient of the explanation (the explainee) is suspiciously absent, despite the problem being ultimately communicative in nature. We pose the problem `explaining AI systems' in terms of a two-player cooperative game in which each agent seeks to maximise our proposed ...
http://arxiv.org/pdf/2305.12233v1
2023-05-20T16:52:30
2023-05-20T16:52:30
{ "title": [ -0.0032997131, -0.003490448, -0.025894165, 0.012664795, -0.041900635, -0.044677734, -0.008888245, -0.03781128, -0.0060539246, 0.007385254, -0.038146973, 0.008903503, -0.022705078, -0.044433594, 0.035339355, -0.016113281, 0.012710571, ...
http://arxiv.org/abs/2305.12228v1
Dynamic Transformers Provide a False Sense of Efficiency
Yiming Chen, Simin Chen, Zexin Li, Wei Yang, Cong Liu, Robby T. Tan, Haizhou Li
cs.CL
Despite much success in natural language processing (NLP), pre-trained language models typically lead to a high computational cost during inference. Multi-exit is a mainstream approach to address this issue by making a trade-off between efficiency and accuracy, where the saving of computation comes from an early exit. ...
http://arxiv.org/pdf/2305.12228v1
2023-05-20T16:41:48
2023-05-20T16:41:48
{ "title": [ 0.04977417, -0.0022068024, 0.024505615, 0.026565552, -0.025634766, -0.017654419, 0.08984375, -0.02230835, -0.041778564, 0.05606079, -0.0018863678, -0.023498535, -0.006126404, -0.03326416, 0.011795044, -0.02911377, 0.020401001, 0.0313...
http://arxiv.org/abs/2305.18243v3
Practical PCG Through Large Language Models
Muhammad U Nasir, Julian Togelius
cs.CL, cs.AI
Large Language Models (LLMs) have proven to be useful tools in various domains outside of the field of their inception, which was natural language processing. In this study, we provide practical directions on how to use LLMs to generate 2D-game rooms for an under-development game, named Metavoidal. Our technique can ha...
http://arxiv.org/pdf/2305.18243v3
2023-07-02T09:11:17
2023-05-20T16:08:45
{ "title": [ -0.023483276, -0.030578613, -0.0357666, -0.086242676, -0.020645142, 0.011558533, -0.09350586, 0.0008802414, 0.007183075, 0.039642334, -0.0019321442, -0.021209717, 0.009284973, -0.04425049, -0.037017822, 0.024490356, 0.005607605, 0.02...
http://arxiv.org/abs/2305.12219v2
Collaborative Development of NLP models
Fereshte Khani, Marco Tulio Ribeiro
cs.LG, cs.AI, cs.CL
Despite substantial advancements, Natural Language Processing (NLP) models often require post-training adjustments to enforce business rules, rectify undesired behavior, and align with user values. These adjustments involve operationalizing "concepts"--dictating desired model responses to certain inputs. However, it's ...
http://arxiv.org/pdf/2305.12219v2
2023-05-24T22:05:16
2023-05-20T15:55:39
{ "title": [ -0.039489746, -0.0096588135, -0.035491943, -0.045837402, -0.03213501, 0.0046691895, -0.06890869, 0.0013542175, 0.001871109, 0.025146484, -0.03253174, -0.016723633, -0.014434814, -0.019439697, -0.04083252, -0.003835678, -0.023132324, ...
http://arxiv.org/abs/2305.12217v1
PromptNER: A Prompting Method for Few-shot Named Entity Recognition via k Nearest Neighbor Search
Mozhi Zhang, Hang Yan, Yaqian Zhou, Xipeng Qiu
cs.CL
Few-shot Named Entity Recognition (NER) is a task aiming to identify named entities via limited annotated samples. Recently, prototypical networks have shown promising performance in few-shot NER. Most of prototypical networks will utilize the entities from the support set to construct label prototypes and use the quer...
http://arxiv.org/pdf/2305.12217v1
2023-05-20T15:47:59
2023-05-20T15:47:59
{ "title": [ -0.021133423, -0.014297485, -0.02772522, -0.045440674, -0.000831604, 0.000011444092, -0.040405273, -0.006576538, 0.016418457, 0.001791954, -0.04849243, 0.023040771, 0.005241394, -0.0038261414, -0.030548096, -0.008995056, 0.015350342, ...
http://arxiv.org/abs/2305.12212v2
Prompting ChatGPT in MNER: Enhanced Multimodal Named Entity Recognition with Auxiliary Refined Knowledge
Jinyuan Li, Han Li, Zhuo Pan, Di Sun, Jiahao Wang, Wenkun Zhang, Gang Pan
cs.CL
Multimodal Named Entity Recognition (MNER) on social media aims to enhance textual entity prediction by incorporating image-based clues. Existing studies mainly focus on maximizing the utilization of pertinent image information or incorporating external knowledge from explicit knowledge bases. However, these methods ei...
http://arxiv.org/pdf/2305.12212v2
2023-10-18T17:05:58
2023-05-20T15:24:38
{ "title": [ -0.0068244934, 0.006286621, -0.04928589, -0.064941406, 0.014091492, -0.010787964, -0.034301758, 0.00016081333, -0.008392334, 0.015266418, -0.026153564, 0.026306152, 0.014923096, -0.009361267, -0.012290955, 0.014549255, -0.010643005, ...
http://arxiv.org/abs/2305.12209v1
Self-Distillation with Meta Learning for Knowledge Graph Completion
Yunshui Li, Junhao Liu, Chengming Li, Min Yang
cs.CL
In this paper, we propose a selfdistillation framework with meta learning(MetaSD) for knowledge graph completion with dynamic pruning, which aims to learn compressed graph embeddings and tackle the longtail samples. Specifically, we first propose a dynamic pruning technique to obtain a small pruned model from a large s...
http://arxiv.org/pdf/2305.12209v1
2023-05-20T15:12:25
2023-05-20T15:12:25
{ "title": [ -0.02720642, 0.0026512146, -0.070251465, -0.087402344, 0.0011730194, 0.015045166, -0.0925293, 0.008163452, 0.018096924, 0.04321289, -0.016433716, 0.0044784546, -0.010734558, -0.004558563, -0.06286621, 0.0003671646, 0.0054893494, 0.04...
http://arxiv.org/abs/2305.12199v1
VNHSGE: VietNamese High School Graduation Examination Dataset for Large Language Models
Xuan-Quy Dao, Ngoc-Bich Le, The-Duy Vo, Xuan-Dung Phan, Bac-Bien Ngo, Van-Tien Nguyen, Thi-My-Thanh Nguyen, Hong-Phuoc Nguyen
cs.CL
The VNHSGE (VietNamese High School Graduation Examination) dataset, developed exclusively for evaluating large language models (LLMs), is introduced in this article. The dataset, which covers nine subjects, was generated from the Vietnamese National High School Graduation Examination and comparable tests. 300 literary ...
http://arxiv.org/pdf/2305.12199v1
2023-05-20T14:13:08
2023-05-20T14:13:08
{ "title": [ -0.0064964294, -0.0075912476, -0.050598145, -0.057556152, -0.0025177002, -0.021240234, -0.030929565, 0.020217896, -0.0031604767, 0.04699707, -0.020996094, -0.024642944, 0.04095459, -0.031051636, -0.036834717, -0.011276245, -0.03414917, ...
http://arxiv.org/abs/2305.12196v1
Experimental results from applying GPT-4 to an unpublished formal language
Gregor vom Scheidt
cs.CL, cs.LO, math.LO, 03B35, F.4.1; I.2.3; F.4.3
Can large language models be used to complete mathematical tasks that are traditionally performed either manually or with the aid of theorem provers? To answer this question, a state-of-the-art system, GPT-4, was provided with a concise natural language specification for a previously unpublished formal system and asked...
http://arxiv.org/pdf/2305.12196v1
2023-05-20T14:00:08
2023-05-20T14:00:08
{ "title": [ -0.012359619, 0.009010315, -0.027999878, -0.03994751, -0.0048332214, -0.019378662, -0.034423828, -0.016830444, -0.0021533966, 0.015029907, -0.0062026978, -0.0022659302, -0.013458252, -0.029190063, -0.01687622, 0.0048065186, -0.026748657,...
http://arxiv.org/abs/2305.12191v1
Pointwise Mutual Information Based Metric and Decoding Strategy for Faithful Generation in Document Grounded Dialogs
Yatin Nandwani, Vineet Kumar, Dinesh Raghu, Sachindra Joshi, Luis A. Lastras
cs.CL
A major concern in using deep learning based generative models for document-grounded dialogs is the potential generation of responses that are not \textit{faithful} to the underlying document. Existing automated metrics used for evaluating the faithfulness of response with respect to the grounding document measure the ...
http://arxiv.org/pdf/2305.12191v1
2023-05-20T13:34:34
2023-05-20T13:34:34
{ "title": [ -0.032165527, -0.011978149, -0.042938232, -0.051208496, -0.029922485, -0.0012273788, -0.06958008, 0.0024375916, -0.007980347, 0.08087158, 0.0063667297, 0.017105103, 0.052124023, -0.015029907, -0.04916382, 0.013885498, -0.0074386597, ...
http://arxiv.org/abs/2305.12190v1
Paragraph-level Citation Recommendation based on Topic Sentences as Queries
Zoran Medić, Jan Šnajder
cs.IR, cs.CL
Citation recommendation (CR) models may help authors find relevant articles at various stages of the paper writing process. Most research has dealt with either global CR, which produces general recommendations suitable for the initial writing stage, or local CR, which produces specific recommendations more fitting for ...
http://arxiv.org/pdf/2305.12190v1
2023-05-20T13:28:22
2023-05-20T13:28:22
{ "title": [ 0.004547119, 0.0067443848, -0.035186768, -0.045715332, 0.005630493, -0.012908936, -0.014076233, -0.020202637, -0.009300232, 0.039855957, -0.031143188, 0.036224365, 0.015213013, 0.0011968613, -0.027359009, 0.0070533752, -0.017944336, ...
http://arxiv.org/abs/2305.12182v2
Glot500: Scaling Multilingual Corpora and Language Models to 500 Languages
Ayyoob Imani, Peiqin Lin, Amir Hossein Kargaran, Silvia Severini, Masoud Jalili Sabet, Nora Kassner, Chunlan Ma, Helmut Schmid, André F. T. Martins, François Yvon, Hinrich Schütze
cs.CL
The NLP community has mainly focused on scaling Large Language Models (LLMs) vertically, i.e., making them better for about 100 languages. We instead scale LLMs horizontally: we create, through continued pretraining, Glot500-m, an LLM that covers 511 predominantly low-resource languages. An important part of this effor...
http://arxiv.org/pdf/2305.12182v2
2023-05-26T11:30:08
2023-05-20T12:26:41
{ "title": [ -0.013267517, -0.011665344, -0.0552063, -0.08129883, 0.020721436, 0.0026340485, -0.0637207, 0.009895325, -0.0027198792, 0.034179688, -0.02784729, 0.016036987, 0.018432617, -0.016586304, -0.024337769, 0.0063552856, -0.008300781, 0.047...
http://arxiv.org/abs/2305.12169v2
Learning to Compose Representations of Different Encoder Layers towards Improving Compositional Generalization
Lei Lin, Shuangtao Li, Yafang Zheng, Biao Fu, Shan Liu, Yidong Chen, Xiaodong Shi
cs.CL
Recent studies have shown that sequence-to-sequence (seq2seq) models struggle with compositional generalization (CG), i.e., the ability to systematically generalize to unseen compositions of seen components. There is mounting evidence that one of the reasons hindering CG is the representation of the encoder uppermost l...
http://arxiv.org/pdf/2305.12169v2
2023-10-18T14:19:40
2023-05-20T11:16:59
{ "title": [ 0.007434845, 0.008651733, -0.043395996, -0.07946777, 0.0073013306, -0.010002136, -0.061309814, 0.010040283, 0.008926392, 0.041015625, 0.005340576, 0.011962891, -0.016647339, -0.017440796, -0.041656494, 0.012832642, -0.017913818, 0.06...
http://arxiv.org/abs/2305.12152v2
Revisiting Automated Topic Model Evaluation with Large Language Models
Dominik Stammbach, Vilém Zouhar, Alexander Hoyle, Mrinmaya Sachan, Elliott Ash
cs.CL
Topic models are used to make sense of large text collections. However, automatically evaluating topic model output and determining the optimal number of topics both have been longstanding challenges, with no effective automated solutions to date. This paper proposes using large language models to evaluate such output....
http://arxiv.org/pdf/2305.12152v2
2023-10-22T09:46:13
2023-05-20T09:42:00
{ "title": [ -0.03616333, 0.005218506, -0.0231781, -0.03036499, -0.00944519, 0.008522034, -0.0463562, -0.037078857, 0.015167236, 0.02645874, -0.014015198, 0.017196655, -0.007549286, -0.013809204, -0.041015625, 0.0181427, -0.019805908, 0.015823364...
http://arxiv.org/abs/2305.12147v2
LogiCoT: Logical Chain-of-Thought Instruction-Tuning
Hanmeng Liu, Zhiyang Teng, Leyang Cui, Chaoli Zhang, Qiji Zhou, Yue Zhang
cs.CL, cs.AI
Generative Pre-trained Transformer 4 (GPT-4) demonstrates impressive chain-of-thought reasoning ability. Recent work on self-instruction tuning, such as Alpaca, has focused on enhancing the general proficiency of models. These instructions enable the model to achieve performance comparable to GPT-3.5 on general tasks l...
http://arxiv.org/pdf/2305.12147v2
2023-10-28T04:22:58
2023-05-20T09:23:09
{ "title": [ -0.02128601, -0.016357422, -0.036224365, -0.02130127, -0.038726807, -0.01235199, -0.041778564, -0.019317627, -0.035827637, 0.050048828, -0.030136108, 0.026489258, 0.027389526, -0.014778137, 0.014442444, 0.00073575974, -0.014823914, 0...
http://arxiv.org/abs/2305.12146v2
Hedges in Bidirectional Translations of Publicity-Oriented Documents
Zhaokun Jiang, Ziyin Zhang
cs.CL
Hedges are widely studied across registers and disciplines, yet research on the translation of hedges in political texts is extremely limited. This contrastive study is dedicated to investigating whether there is a diachronic change in the frequencies of hedging devices in the target texts, to what extent the changing ...
http://arxiv.org/pdf/2305.12146v2
2023-05-23T01:42:06
2023-05-20T09:19:39
{ "title": [ 0.010505676, -0.00015485287, -0.010826111, -0.018051147, -0.027389526, -0.021102905, -0.0138549805, -0.021820068, -0.06008911, 0.07165527, -0.011405945, 0.022323608, 0.044403076, -0.033203125, 0.018325806, -0.0057640076, -0.0178833, ...
http://arxiv.org/abs/2305.12129v1
Lifting the Curse of Capacity Gap in Distilling Language Models
Chen Zhang, Yang Yang, Jiahao Liu, Jingang Wang, Yunsen Xian, Benyou Wang, Dawei Song
cs.CL, cs.LG
Pretrained language models (LMs) have shown compelling performance on various downstream tasks, but unfortunately they require a tremendous amount of inference compute. Knowledge distillation finds a path to compress LMs to small ones with a teacher-student paradigm. However, when the capacity gap between the teacher a...
http://arxiv.org/pdf/2305.12129v1
2023-05-20T07:30:55
2023-05-20T07:30:55
{ "title": [ -0.020217896, -0.003458023, -0.055480957, -0.08880615, -0.0048561096, 0.010322571, -0.09265137, 0.009124756, 0.022384644, 0.027893066, -0.044403076, -0.009941101, 0.003250122, -0.006954193, -0.05657959, -0.0010509491, -0.006511688, 0...
http://arxiv.org/abs/2305.12123v1
Modeling the Q-Diversity in a Min-max Play Game for Robust Optimization
Ting Wu, Rui Zheng, Tao Gui, Qi Zhang, Xuanjing Huang
cs.CL
Models trained with empirical risk minimization (ERM) are revealed to easily rely on spurious correlations, resulting in poor generalization. Group distributionally robust optimization (group DRO) can alleviate this problem by minimizing the worst-case loss over pre-defined groups. While promising, in practice factors ...
http://arxiv.org/pdf/2305.12123v1
2023-05-20T07:02:27
2023-05-20T07:02:27
{ "title": [ -0.012367249, 0.047576904, -0.043884277, 0.016296387, 0.04248047, -0.039398193, 0.024429321, -0.07104492, -0.054870605, 0.029708862, 0.014419556, 0.025360107, -0.06451416, -0.019729614, -0.05053711, -0.031982422, -0.0073432922, 0.005...
http://arxiv.org/abs/2305.12107v1
EE-TTS: Emphatic Expressive TTS with Linguistic Information
Yi Zhong, Chen Zhang, Xule Liu, Chenxi Sun, Weishan Deng, Haifeng Hu, Zhongqian Sun
cs.SD, cs.CL, eess.AS
While Current TTS systems perform well in synthesizing high-quality speech, producing highly expressive speech remains a challenge. Emphasis, as a critical factor in determining the expressiveness of speech, has attracted more attention nowadays. Previous works usually enhance the emphasis by adding intermediate featur...
http://arxiv.org/pdf/2305.12107v1
2023-05-20T05:58:56
2023-05-20T05:58:56
{ "title": [ -0.0036067963, -0.0072021484, -0.05392456, -0.061767578, 0.014350891, -0.021194458, -0.032348633, 0.0035858154, 0.005241394, 0.030685425, -0.017684937, -0.024490356, 0.035888672, -0.027999878, -0.040618896, 0.01499939, -0.0011854172, ...
http://arxiv.org/abs/2305.12096v1
Can NLP Models Correctly Reason Over Contexts that Break the Common Assumptions?
Neeraj Varshney, Mihir Parmar, Nisarg Patel, Divij Handa, Sayantan Sarkar, Man Luo, Chitta Baral
cs.CL
Pre-training on large corpora of text enables the language models to acquire a vast amount of factual and commonsense knowledge which allows them to achieve remarkable performance on a variety of language understanding tasks. They typically acquire this knowledge by learning from the pre-training text and capturing cer...
http://arxiv.org/pdf/2305.12096v1
2023-05-20T05:20:37
2023-05-20T05:20:37
{ "title": [ 0.034729004, -0.0057525635, 0.012023926, -0.026794434, -0.028717041, 0.0062294006, -0.023834229, -0.046447754, -0.0033168793, 0.012863159, -0.055023193, -0.027694702, 0.014137268, -0.027572632, 0.023391724, -0.041656494, -0.017456055, ...
http://arxiv.org/abs/2305.12092v1
ESCOXLM-R: Multilingual Taxonomy-driven Pre-training for the Job Market Domain
Mike Zhang, Rob van der Goot, Barbara Plank
cs.CL
The increasing number of benchmarks for Natural Language Processing (NLP) tasks in the computational job market domain highlights the demand for methods that can handle job-related tasks such as skill extraction, skill classification, job title classification, and de-identification. While some approaches have been deve...
http://arxiv.org/pdf/2305.12092v1
2023-05-20T04:50:20
2023-05-20T04:50:20
{ "title": [ -0.0010480881, -0.039367676, -0.03729248, -0.0519104, 0.00034046173, -0.016067505, -0.012519836, 0.008857727, -0.04385376, 0.020141602, -0.009460449, 0.026397705, -0.022521973, -0.027145386, 0.0034370422, -0.020645142, -0.011489868, ...
http://arxiv.org/abs/2305.12091v2
"What do others think?": Task-Oriented Conversational Modeling with Subjective Knowledge
Chao Zhao, Spandana Gella, Seokhwan Kim, Di Jin, Devamanyu Hazarika, Alexandros Papangelis, Behnam Hedayatnia, Mahdi Namazifar, Yang Liu, Dilek Hakkani-Tur
cs.CL
Task-oriented Dialogue (TOD) Systems aim to build dialogue systems that assist users in accomplishing specific goals, such as booking a hotel or a restaurant. Traditional TODs rely on domain-specific APIs/DBs or external factual knowledge to generate responses, which cannot accommodate subjective user requests (e.g., "...
http://arxiv.org/pdf/2305.12091v2
2023-10-03T03:33:20
2023-05-20T04:43:26
{ "title": [ -0.02204895, -0.0015697479, -0.044189453, -0.060333252, -0.006084442, -0.0031929016, -0.06011963, 0.007762909, 0.020065308, 0.035064697, -0.017288208, -0.010734558, 0.003238678, -0.01322937, -0.08270264, -0.005744934, -0.009719849, 0...
http://arxiv.org/abs/2305.12090v1
UP5: Unbiased Foundation Model for Fairness-aware Recommendation
Wenyue Hua, Yingqiang Ge, Shuyuan Xu, Jianchao Ji, Yongfeng Zhang
cs.IR, cs.AI, cs.CL, cs.LG
Recent advancements in foundation models such as large language models (LLM) have propelled them to the forefront of recommender systems (RS). Moreover, fairness in RS is critical since many users apply it for decision-making and demand fulfillment. However, at present, there is a lack of understanding regarding the le...
http://arxiv.org/pdf/2305.12090v1
2023-05-20T04:32:59
2023-05-20T04:32:59
{ "title": [ 0.0070266724, 0.00062704086, -0.04949951, -0.056365967, 0.02557373, -0.01864624, -0.016204834, -0.017532349, 0.008666992, 0.033416748, -0.024856567, 0.02067566, 0.023345947, -0.01889038, -0.03439331, 0.027877808, -0.019042969, 0.0566...
http://arxiv.org/abs/2305.12086v2
Prefix Propagation: Parameter-Efficient Tuning for Long Sequences
Jonathan Li, Will Aitken, Rohan Bhambhoria, Xiaodan Zhu
cs.CL
Parameter-efficient tuning aims to mitigate the large memory requirements of adapting pretrained language models for downstream tasks. For example, one popular method, prefix-tuning, prepends trainable tokens to sequences while freezing the rest of the model's parameters. Although such models attain comparable performa...
http://arxiv.org/pdf/2305.12086v2
2023-05-24T21:13:24
2023-05-20T04:07:06
{ "title": [ -0.032440186, -0.016571045, -0.02456665, -0.08502197, 0.0025615692, 0.01725769, -0.050048828, 0.0064697266, -0.015098572, 0.03475952, 0.021774292, 0.0013570786, 0.0049552917, -0.03515625, -0.030517578, 0.001789093, 0.00038433075, 0.0...
http://arxiv.org/abs/2305.12084v2
Revisiting Entropy Rate Constancy in Text
Vivek Verma, Nicholas Tomlin, Dan Klein
cs.CL
The uniform information density (UID) hypothesis states that humans tend to distribute information roughly evenly across an utterance or discourse. Early evidence in support of the UID hypothesis came from Genzel & Charniak (2002), which proposed an entropy rate constancy principle based on the probability of English t...
http://arxiv.org/pdf/2305.12084v2
2023-10-18T01:02:56
2023-05-20T03:48:31
{ "title": [ -0.04916382, -0.0053710938, -0.027938843, -0.019729614, -0.012046814, -0.0066871643, -0.029693604, -0.012664795, 0.031402588, 0.0519104, -0.04446411, 0.02154541, 0.013923645, 0.020736694, -0.083740234, 0.023666382, -0.000104784966, 0...
http://arxiv.org/abs/2305.12077v1
Few-Shot Dialogue Summarization via Skeleton-Assisted Prompt Transfer
Kaige Xie, Tong Yu, Haoliang Wang, Junda Wu, Handong Zhao, Ruiyi Zhang, Kanak Mahadik, Ani Nenkova, Mark Riedl
cs.CL
In real-world scenarios, labeled samples for dialogue summarization are usually limited (i.e., few-shot) due to high annotation costs for high-quality dialogue summaries. To efficiently learn from few-shot samples, previous works have utilized massive annotated data from other downstream tasks and then performed prompt...
http://arxiv.org/pdf/2305.12077v1
2023-05-20T03:32:48
2023-05-20T03:32:48
{ "title": [ 0.0009598732, -0.021438599, -0.014884949, -0.050872803, -0.03186035, 0.006916046, -0.034057617, 0.0074653625, 0.00079345703, 0.044921875, 0.023468018, 0.0056266785, -0.00008183718, -0.02468872, -0.04949951, 0.010284424, -0.0050582886, ...
http://arxiv.org/abs/2305.12074v3
DisCo: Distilled Student Models Co-training for Semi-supervised Text Mining
Weifeng Jiang, Qianren Mao, Chenghua Lin, Jianxin Li, Ting Deng, Weiyi Yang, Zheng Wang
cs.CL
Many text mining models are constructed by fine-tuning a large deep pre-trained language model (PLM) in downstream tasks. However, a significant challenge nowadays is maintaining performance when we use a lightweight model with limited labelled samples. We present DisCo, a semi-supervised learning (SSL) framework for f...
http://arxiv.org/pdf/2305.12074v3
2023-10-20T02:11:16
2023-05-20T03:23:16
{ "title": [ -0.020858765, 0.003736496, -0.038848877, -0.049468994, 0.012214661, -0.0009536743, -0.030410767, -0.01335907, 0.012336731, 0.025146484, -0.050750732, 0.015975952, 0.0030155182, -0.0013189316, -0.04663086, -0.019195557, -0.038024902, ...
http://arxiv.org/abs/2305.12057v2
Accurate Knowledge Distillation with n-best Reranking
Hendra Setiawan
cs.CL
We propose utilizing n-best reranking to enhance the Sequence-Level Knowledge Distillation (Kim and Rush, 2016) where we explore hypotheses beyond the top-1 to acquire more accurate pseudo-labels. To accomplish this, we leverage a diverse set of models with different inductive biases, objective functions or architectur...
http://arxiv.org/pdf/2305.12057v2
2023-11-14T21:02:57
2023-05-20T01:53:03
{ "title": [ -0.027175903, -0.007091522, -0.04812622, -0.09863281, 0.0048942566, 0.026611328, -0.07525635, -0.006351471, 0.035003662, 0.008361816, -0.03274536, 0.0072746277, -0.020904541, -0.005783081, -0.035736084, -0.016281128, 0.008491516, -0....
http://arxiv.org/abs/2305.18315v1
CDJUR-BR -- A Golden Collection of Legal Document from Brazilian Justice with Fine-Grained Named Entities
Antonio Mauricio, Vladia Pinheiro, Vasco Furtado, João Araújo Monteiro Neto, Francisco das Chagas Jucá Bomfim, André Câmara Ferreira da Costa, Raquel Silveira, Nilsiton Aragão
cs.CL, cs.AI, cs.IR, cs.LG
A basic task for most Legal Artificial Intelligence (Legal AI) applications is Named Entity Recognition (NER). However, texts produced in the context of legal practice make references to entities that are not trivially recognized by the currently available NERs. There is a lack of categorization of legislation, jurispr...
http://arxiv.org/pdf/2305.18315v1
2023-05-20T00:48:52
2023-05-20T00:48:52
{ "title": [ -0.016906738, -0.030029297, -0.043884277, -0.057922363, -0.017684937, 0.011199951, -0.041656494, 0.040527344, -0.048675537, 0.051727295, 0.0064926147, 0.033447266, 0.03704834, -0.01777649, 0.027954102, 0.0030708313, 0.013122559, 0.01...
http://arxiv.org/abs/2305.12031v2
Clinical Camel: An Open Expert-Level Medical Language Model with Dialogue-Based Knowledge Encoding
Augustin Toma, Patrick R. Lawler, Jimmy Ba, Rahul G. Krishnan, Barry B. Rubin, Bo Wang
cs.CL, cs.AI
We present Clinical Camel, an open large language model (LLM) explicitly tailored for clinical research. Fine-tuned from LLaMA-2 using QLoRA, Clinical Camel achieves state-of-the-art performance across medical benchmarks among openly available medical LLMs. Leveraging efficient single-GPU training, Clinical Camel surpa...
http://arxiv.org/pdf/2305.12031v2
2023-08-17T17:19:02
2023-05-19T23:07:09
{ "title": [ -0.04321289, -0.025436401, -0.045318604, -0.08099365, -0.009933472, 0.026046753, -0.08068848, 0.008628845, -0.0039043427, 0.032196045, 0.027511597, -0.016418457, 0.0062294006, -0.040405273, -0.041259766, 0.01108551, -0.0042266846, 0....
http://arxiv.org/abs/2305.12029v2
MultiTurnCleanup: A Benchmark for Multi-Turn Spoken Conversational Transcript Cleanup
Hua Shen, Vicky Zayats, Johann C. Rocholl, Daniel D. Walker, Dirk Padfield
cs.CL, cs.AI, cs.LG
Current disfluency detection models focus on individual utterances each from a single speaker. However, numerous discontinuity phenomena in spoken conversational transcripts occur across multiple turns, hampering human readability and the performance of downstream NLP tasks. This study addresses these phenomena by prop...
http://arxiv.org/pdf/2305.12029v2
2023-10-27T17:01:50
2023-05-19T22:50:02
{ "title": [ -0.0043792725, 0.00063943863, -0.06707764, -0.091796875, 0.0055007935, -0.005809784, -0.07904053, 0.02607727, 0.002040863, 0.046875, -0.0049209595, -0.0026893616, 0.0149002075, -0.011161804, -0.060180664, 0.0061073303, -0.020126343, ...
http://arxiv.org/abs/2305.12027v2
Polar Ducks and Where to Find Them: Enhancing Entity Linking with Duck Typing and Polar Box Embeddings
Mattia Atzeni, Mikhail Plekhanov, Frédéric A. Dreyer, Nora Kassner, Simone Merello, Louis Martin, Nicola Cancedda
cs.CL, cs.AI
Entity linking methods based on dense retrieval are an efficient and widely used solution in large-scale applications, but they fall short of the performance of generative models, as they are sensitive to the structure of the embedding space. In order to address this issue, this paper introduces DUCK, an approach to in...
http://arxiv.org/pdf/2305.12027v2
2023-10-20T13:58:55
2023-05-19T22:42:16
{ "title": [ 0.02609253, 0.011528015, -0.033294678, -0.059539795, 0.034484863, -0.017501831, -0.0026168823, -0.008514404, -0.0030899048, 0.034301758, 0.00022315979, 0.020767212, 0.033935547, -0.032196045, -0.01134491, 0.00868988, -0.030197144, 0....
http://arxiv.org/abs/2305.12018v1
BOLT: Fast Energy-based Controlled Text Generation with Tunable Biases
Xin Liu, Muhammad Khalifa, Lu Wang
cs.CL
Energy-based models (EBMs) have gained popularity for controlled text generation due to their high applicability to a wide range of constraints. However, sampling from EBMs is non-trivial, as it often requires a large number of iterations to converge to plausible text, which slows down the decoding process and makes it...
http://arxiv.org/pdf/2305.12018v1
2023-05-19T22:02:55
2023-05-19T22:02:55
{ "title": [ -0.044921875, -0.013313293, -0.03112793, -0.036956787, -0.009613037, 0.01008606, -0.033721924, -0.009971619, -0.008049011, 0.042938232, -0.011260986, 0.011695862, 0.021240234, -0.015991211, -0.030136108, 0.009727478, -0.018600464, 0....
http://arxiv.org/abs/2305.12002v1
XuanYuan 2.0: A Large Chinese Financial Chat Model with Hundreds of Billions Parameters
Xuanyu Zhang, Qing Yang, Dongliang Xu
cs.CL
In recent years, pre-trained language models have undergone rapid development with the emergence of large-scale models. However, there is a lack of open-sourced chat models specifically designed for the Chinese language, especially in the field of Chinese finance, at the scale of hundreds of billions. To address this g...
http://arxiv.org/pdf/2305.12002v1
2023-05-19T21:01:20
2023-05-19T21:01:20
{ "title": [ -0.018295288, -0.0071640015, -0.066101074, -0.034301758, -0.0077781677, -0.03753662, -0.037109375, 0.0033397675, -0.017745972, 0.054382324, -0.012802124, -0.008728027, 0.045898438, -0.035583496, -0.016174316, -0.002216339, 0.000005304813...
http://arxiv.org/abs/2305.12001v2
OPT-R: Exploring the Role of Explanations in Finetuning and Prompting for Reasoning Skills of Large Language Models
Badr AlKhamissi, Siddharth Verma, Ping Yu, Zhijing Jin, Asli Celikyilmaz, Mona Diab
cs.CL
In this paper, we conduct a thorough investigation into the reasoning capabilities of Large Language Models (LLMs), focusing specifically on the Open Pretrained Transformers (OPT) models as a representative of such models. Our study entails finetuning three different sizes of OPT on a carefully curated reasoning corpus...
http://arxiv.org/pdf/2305.12001v2
2023-10-24T13:38:19
2023-05-19T20:58:22
{ "title": [ -0.017715454, -0.041015625, -0.0127334595, -0.09234619, -0.025634766, 0.016952515, -0.0569458, 0.022537231, -0.028213501, -0.018218994, -0.025344849, 0.00018715858, -0.013092041, -0.035949707, 0.007209778, -0.037322998, -0.01499176, ...
http://arxiv.org/abs/2305.12000v1
Deep Learning Approaches to Lexical Simplification: A Survey
Kai North, Tharindu Ranasinghe, Matthew Shardlow, Marcos Zampieri
cs.CL
Lexical Simplification (LS) is the task of replacing complex for simpler words in a sentence whilst preserving the sentence's original meaning. LS is the lexical component of Text Simplification (TS) with the aim of making texts more accessible to various target populations. A past survey (Paetzold and Specia, 2017) ha...
http://arxiv.org/pdf/2305.12000v1
2023-05-19T20:56:22
2023-05-19T20:56:22
{ "title": [ -0.0317688, -0.00029039383, -0.038269043, -0.066589355, 0.0048942566, 0.017684937, -0.066833496, 0.008163452, 0.0096588135, 0.050994873, -0.0042762756, 0.015731812, 0.012191772, -0.013031006, -0.047210693, 0.013000488, -0.0077171326, ...
http://arxiv.org/abs/2305.11993v2
Interpretable Word Sense Representations via Definition Generation: The Case of Semantic Change Analysis
Mario Giulianelli, Iris Luden, Raquel Fernandez, Andrey Kutuzov
cs.CL
We propose using automatically generated natural language definitions of contextualised word usages as interpretable word and word sense representations. Given a collection of usage examples for a target word, and the corresponding data-driven usage clusters (i.e., word senses), a definition is generated for each usage...
http://arxiv.org/pdf/2305.11993v2
2023-07-25T11:50:48
2023-05-19T20:36:21
{ "title": [ -0.012313843, 0.00026488304, -0.035980225, -0.038269043, 0.0019292831, 0.001033783, -0.038909912, -0.016586304, 0.010368347, 0.03475952, -0.025344849, 0.015533447, -0.000702858, -0.0021247864, -0.05078125, 0.006439209, -0.00004708767, ...
http://arxiv.org/abs/2305.11991v2
Evaluation of medium-large Language Models at zero-shot closed book generative question answering
René Peinl, Johannes Wirth
cs.CL, I.2.7
Large language models (LLMs) have garnered significant attention, but the definition of "large" lacks clarity. This paper focuses on medium-sized language models (MLMs), defined as having at least six billion parameters but less than 100 billion. The study evaluates MLMs regarding zero-shot generative question answerin...
http://arxiv.org/pdf/2305.11991v2
2023-07-03T18:26:23
2023-05-19T20:33:19
{ "title": [ -0.022918701, -0.018997192, -0.03466797, -0.042022705, -0.025726318, -0.0063171387, -0.039886475, 0.012702942, 0.020858765, 0.02279663, -0.025543213, 0.0033817291, 0.0013694763, -0.013343811, -0.055419922, 0.006679535, -0.0028591156, ...
http://arxiv.org/abs/2305.11979v1
A Weak Supervision Approach for Few-Shot Aspect Based Sentiment
Robert Vacareanu, Siddharth Varia, Kishaloy Halder, Shuai Wang, Giovanni Paolini, Neha Anna John, Miguel Ballesteros, Smaranda Muresan
cs.CL
We explore how weak supervision on abundant unlabeled data can be leveraged to improve few-shot performance in aspect-based sentiment analysis (ABSA) tasks. We propose a pipeline approach to construct a noisy ABSA dataset, and we use it to adapt a pre-trained sequence-to-sequence model to the ABSA tasks. We test the re...
http://arxiv.org/pdf/2305.11979v1
2023-05-19T19:53:54
2023-05-19T19:53:54
{ "title": [ 0.02407837, -0.018478394, 0.0074806213, -0.06921387, -0.004261017, 0.013015747, -0.019592285, 0.015350342, 0.025009155, 0.047912598, -0.01725769, -0.015670776, 0.00674057, -0.010238647, -0.07104492, -0.004863739, -0.0015106201, 0.017...
http://arxiv.org/abs/2305.11952v1
Self-QA: Unsupervised Knowledge Guided Language Model Alignment
Xuanyu Zhang, Qing Yang
cs.CL
Large-scale language models like ChatGPT and GPT-4 have gained attention for their impressive conversational and generative capabilities. However, the creation of supervised paired question-answering data for instruction tuning presents formidable challenges. This endeavor necessitates substantial human effort for data...
http://arxiv.org/pdf/2305.11952v1
2023-05-19T18:26:26
2023-05-19T18:26:26
{ "title": [ -0.048675537, -0.0074424744, -0.063964844, -0.052642822, -0.0059013367, -0.0020637512, -0.08117676, 0.004501343, 0.002243042, 0.046051025, 0.0036773682, -0.0062713623, 0.01637268, -0.0234375, -0.052337646, 0.01625061, -0.0007147789, ...
http://arxiv.org/abs/2305.11948v1
Eye-SpatialNet: Spatial Information Extraction from Ophthalmology Notes
Surabhi Datta, Tasneem Kaochar, Hio Cheng Lam, Nelly Nwosu, Luca Giancardo, Alice Z. Chuang, Robert M. Feldman, Kirk Roberts
cs.CL
We introduce an annotated corpus of 600 ophthalmology notes labeled with detailed spatial and contextual information of ophthalmic entities. We extend our previously proposed frame semantics-based spatial representation schema, Rad-SpatialNet, to represent spatial language in ophthalmology text, resulting in the Eye-Sp...
http://arxiv.org/pdf/2305.11948v1
2023-05-19T18:08:45
2023-05-19T18:08:45
{ "title": [ -0.022842407, -0.022888184, -0.03137207, -0.051330566, -0.0064582825, -0.0011911392, -0.027664185, -0.0067596436, 0.009155273, 0.036987305, -0.050598145, -0.00006633997, 0.0027275085, -0.00058317184, -0.060546875, -0.0064048767, 0.025482...
http://arxiv.org/abs/2305.11944v2
Exploring the Viability of Synthetic Query Generation for Relevance Prediction
Aditi Chaudhary, Karthik Raman, Krishna Srinivasan, Kazuma Hashimoto, Mike Bendersky, Marc Najork
cs.IR, cs.CL
Query-document relevance prediction is a critical problem in Information Retrieval systems. This problem has increasingly been tackled using (pretrained) transformer-based models which are finetuned using large collections of labeled data. However, in specialized domains such as e-commerce and healthcare, the viability...
http://arxiv.org/pdf/2305.11944v2
2023-06-16T22:00:21
2023-05-19T18:03:36
{ "title": [ 0.0030612946, -0.010192871, -0.02507019, -0.07165527, 0.0010967255, -0.0018501282, -0.02796936, 0.0054359436, -0.0042037964, 0.04724121, -0.015541077, -0.007408142, -0.016052246, -0.0057754517, -0.074645996, -0.014785767, -0.008956909, ...
http://arxiv.org/abs/2305.11938v2
XTREME-UP: A User-Centric Scarce-Data Benchmark for Under-Represented Languages
Sebastian Ruder, Jonathan H. Clark, Alexander Gutkin, Mihir Kale, Min Ma, Massimo Nicosia, Shruti Rijhwani, Parker Riley, Jean-Michel A. Sarr, Xinyi Wang, John Wieting, Nitish Gupta, Anna Katanova, Christo Kirov, Dana L. Dickinson, Brian Roark, Bidisha Samanta, Connie Tao, David I. Adelani, Vera Axelrod, Isaac Caswell,...
cs.CL
Data scarcity is a crucial issue for the development of highly multilingual NLP systems. Yet for many under-represented languages (ULs) -- languages for which NLP re-search is particularly far behind in meeting user needs -- it is feasible to annotate small amounts of data. Motivated by this, we propose XTREME-UP, a be...
http://arxiv.org/pdf/2305.11938v2
2023-05-24T06:09:28
2023-05-19T18:00:03
{ "title": [ -0.026519775, -0.013282776, -0.040618896, -0.066833496, -0.00274086, -0.009010315, -0.0637207, 0.0031433105, 0.018630981, 0.020355225, -0.03881836, 0.02407837, -0.023834229, -0.009666443, -0.05343628, -0.003967285, -0.008911133, 0.04...
http://arxiv.org/abs/2305.11864v1
North Sámi Dialect Identification with Self-supervised Speech Models
Sofoklis Kakouros, Katri Hiovain-Asikainen
eess.AS, cs.CL
The North S\'{a}mi (NS) language encapsulates four primary dialectal variants that are related but that also have differences in their phonology, morphology, and vocabulary. The unique geopolitical location of NS speakers means that in many cases they are bilingual in S\'{a}mi as well as in the dominant state language:...
http://arxiv.org/pdf/2305.11864v1
2023-05-19T17:53:12
2023-05-19T17:53:12
{ "title": [ -0.003955841, 0.020568848, -0.026733398, -0.028564453, -0.006626129, -0.0076141357, -0.032989502, -0.01890564, -0.021118164, 0.0491333, -0.024276733, 0.0008277893, 0.026351929, -0.03488159, -0.033447266, 0.011405945, -0.00041937828, ...
http://arxiv.org/abs/2305.11863v2
Scaling laws for language encoding models in fMRI
Richard Antonello, Aditya Vaidya, Alexander G. Huth
cs.CL, cs.AI
Representations from transformer-based unidirectional language models are known to be effective at predicting brain responses to natural language. However, most studies comparing language models to brains have used GPT-2 or similarly sized language models. Here we tested whether larger open-source models such as those ...
http://arxiv.org/pdf/2305.11863v2
2023-05-22T08:05:44
2023-05-19T17:53:03
{ "title": [ -0.037078857, 0.0011730194, -0.03086853, 0.009605408, -0.024276733, -0.011451721, -0.018569946, -0.024017334, -0.010360718, 0.032104492, 0.0054779053, -0.006149292, 0.005428314, -0.03930664, -0.0317688, 0.014137268, -0.0087890625, 0....
http://arxiv.org/abs/2305.11862v2
Reducing Sequence Length by Predicting Edit Operations with Large Language Models
Masahiro Kaneko, Naoaki Okazaki
cs.CL
Large Language Models (LLMs) have demonstrated remarkable performance in various tasks and gained significant attention. LLMs are also used for local sequence transduction tasks, including grammatical error correction (GEC) and formality style transfer, where most tokens in a source text are kept unchanged. However, th...
http://arxiv.org/pdf/2305.11862v2
2023-10-21T00:57:02
2023-05-19T17:51:05
{ "title": [ -0.03186035, -0.02041626, -0.04916382, -0.07232666, -0.017196655, 0.0071029663, -0.08312988, 0.021835327, 0.008583069, 0.033081055, -0.006465912, -0.024765015, 0.021728516, -0.019348145, -0.04324341, 0.010612488, -0.005962372, 0.0405...
http://arxiv.org/abs/2305.11860v1
Let's Sample Step by Step: Adaptive-Consistency for Efficient Reasoning with LLMs
Pranjal Aggarwal, Aman Madaan, Yiming Yang, Mausam
cs.CL
A popular approach for improving the correctness of output from large language models (LLMs) is Self-Consistency - poll the LLM multiple times and output the most frequent solution. Existing Self-Consistency techniques always draw a constant number of samples per question, where a better approach will be to non-uniform...
http://arxiv.org/pdf/2305.11860v1
2023-05-19T17:49:25
2023-05-19T17:49:25
{ "title": [ -0.052124023, -0.01348114, -0.070007324, -0.07434082, -0.014556885, 0.0053520203, -0.064819336, 0.040924072, -0.020355225, 0.025863647, -0.03161621, 0.008934021, 0.01309967, -0.04019165, 0.009925842, -0.00982666, 0.011802673, -0.0051...
http://arxiv.org/abs/2305.11859v1
Complex Claim Verification with Evidence Retrieved in the Wild
Jifan Chen, Grace Kim, Aniruddh Sriram, Greg Durrett, Eunsol Choi
cs.CL
Evidence retrieval is a core part of automatic fact-checking. Prior work makes simplifying assumptions in retrieval that depart from real-world use cases: either no access to evidence, access to evidence curated by a human fact-checker, or access to evidence available long after the claim has been made. In this work, w...
http://arxiv.org/pdf/2305.11859v1
2023-05-19T17:49:19
2023-05-19T17:49:19
{ "title": [ 0.0635376, 0.015411377, -0.024017334, -0.063964844, -0.003107071, -0.036895752, -0.024978638, 0.00043225288, -0.07067871, 0.02470398, -0.01864624, 0.01600647, -0.008331299, -0.0309906, 0.030563354, -0.022872925, 0.002204895, 0.051818...
http://arxiv.org/abs/2305.11853v2
How to Prompt LLMs for Text-to-SQL: A Study in Zero-shot, Single-domain, and Cross-domain Settings
Shuaichen Chang, Eric Fosler-Lussier
cs.CL
Large language models (LLMs) with in-context learning have demonstrated remarkable capability in the text-to-SQL task. Previous research has prompted LLMs with various demonstration-retrieval strategies and intermediate reasoning steps to enhance the performance of LLMs. However, those works often employ varied strateg...
http://arxiv.org/pdf/2305.11853v2
2023-05-23T02:24:02
2023-05-19T17:43:58
{ "title": [ 0.0008416176, 0.0043907166, -0.044677734, -0.04727173, 0.010482788, -0.022445679, -0.033172607, 0.022918701, -0.024978638, 0.013114929, -0.06304932, 0.038970947, -0.0033740997, -0.023834229, 0.036712646, -0.028182983, 0.0077171326, -...
http://arxiv.org/abs/2305.11846v1
Any-to-Any Generation via Composable Diffusion
Zineng Tang, Ziyi Yang, Chenguang Zhu, Michael Zeng, Mohit Bansal
cs.CV, cs.CL, cs.LG, cs.SD, eess.AS
We present Composable Diffusion (CoDi), a novel generative model capable of generating any combination of output modalities, such as language, image, video, or audio, from any combination of input modalities. Unlike existing generative AI systems, CoDi can generate multiple modalities in parallel and its input is not l...
http://arxiv.org/pdf/2305.11846v1
2023-05-19T17:38:32
2023-05-19T17:38:32
{ "title": [ -0.0079422, -0.017669678, -0.044769287, -0.08795166, 0.004425049, 0.009414673, -0.04055786, 0.03414917, 0.026977539, 0.027282715, -0.010414124, 0.009788513, -0.00030589104, -0.03543091, 0.0028686523, 0.02355957, 0.0034999847, 0.01286...
http://arxiv.org/abs/2305.11845v1
RxnScribe: A Sequence Generation Model for Reaction Diagram Parsing
Yujie Qian, Jiang Guo, Zhengkai Tu, Connor W. Coley, Regina Barzilay
cs.CL, cs.AI, cs.CV
Reaction diagram parsing is the task of extracting reaction schemes from a diagram in the chemistry literature. The reaction diagrams can be arbitrarily complex, thus robustly parsing them into structured data is an open challenge. In this paper, we present RxnScribe, a machine learning model for parsing reaction diagr...
http://arxiv.org/pdf/2305.11845v1
2023-05-19T17:37:28
2023-05-19T17:37:28
{ "title": [ -0.03024292, -0.022903442, -0.035461426, -0.083496094, -0.009056091, 0.027648926, -0.07672119, 0.01739502, -0.000080406666, 0.02772522, -0.013214111, 0.009857178, -0.03149414, -0.0184021, -0.035858154, -0.00046491623, -0.0008163452, ...
http://arxiv.org/abs/2305.11841v1
How Does Generative Retrieval Scale to Millions of Passages?
Ronak Pradeep, Kai Hui, Jai Gupta, Adam D. Lelkes, Honglei Zhuang, Jimmy Lin, Donald Metzler, Vinh Q. Tran
cs.IR, cs.CL
Popularized by the Differentiable Search Index, the emerging paradigm of generative retrieval re-frames the classic information retrieval problem into a sequence-to-sequence modeling task, forgoing external indices and encoding an entire document corpus within a single Transformer. Although many different approaches ha...
http://arxiv.org/pdf/2305.11841v1
2023-05-19T17:33:38
2023-05-19T17:33:38
{ "title": [ 0.020263672, -0.021377563, -0.02859497, -0.02331543, -0.03314209, -0.018798828, -0.014053345, 0.0039978027, 0.0075263977, 0.03805542, -0.042999268, 0.0053100586, -0.022277832, -0.0010948181, -0.03918457, -0.022384644, 0.02027893, -0....
http://arxiv.org/abs/2305.11840v1
SeeGULL: A Stereotype Benchmark with Broad Geo-Cultural Coverage Leveraging Generative Models
Akshita Jha, Aida Davani, Chandan K. Reddy, Shachi Dave, Vinodkumar Prabhakaran, Sunipa Dev
cs.CL, cs.CY
Stereotype benchmark datasets are crucial to detect and mitigate social stereotypes about groups of people in NLP models. However, existing datasets are limited in size and coverage, and are largely restricted to stereotypes prevalent in the Western society. This is especially problematic as language technologies gain ...
http://arxiv.org/pdf/2305.11840v1
2023-05-19T17:30:19
2023-05-19T17:30:19
{ "title": [ 0.016235352, -0.012168884, 0.00466156, -0.053131104, -0.006401062, 0.00007367134, -0.018371582, -0.024291992, 0.01763916, 0.021881104, -0.022842407, 0.026947021, -0.015060425, -0.027862549, -0.027542114, 0.008232117, -0.0181427, 0.02...
http://arxiv.org/abs/2305.11828v3
Appraising the Potential Uses and Harms of LLMs for Medical Systematic Reviews
Hye Sun Yun, Iain J. Marshall, Thomas A. Trikalinos, Byron C. Wallace
cs.CL, cs.AI, cs.HC
Medical systematic reviews play a vital role in healthcare decision making and policy. However, their production is time-consuming, limiting the availability of high-quality and up-to-date evidence summaries. Recent advancements in large language models (LLMs) offer the potential to automatically generate literature re...
http://arxiv.org/pdf/2305.11828v3
2023-10-18T13:54:15
2023-05-19T17:09:19
{ "title": [ -0.029815674, -0.00356102, -0.02406311, -0.03881836, -0.02330017, 0.0016412735, -0.049957275, -0.032928467, -0.053771973, 0.017578125, -0.01651001, 0.03189087, -0.027160645, -0.030410767, -0.0026416779, -0.0031051636, 0.0064430237, 0...
http://arxiv.org/abs/2305.11826v2
ReTAG: Reasoning Aware Table to Analytic Text Generation
Deepanway Ghosal, Preksha Nema, Aravindan Raghuveer
cs.CL, cs.AI
The task of table summarization involves generating text that both succinctly and accurately represents the table or a specific set of highlighted cells within a table. While significant progress has been made in table to text generation techniques, models still mostly generate descriptive summaries, which reiterates t...
http://arxiv.org/pdf/2305.11826v2
2023-10-30T03:24:37
2023-05-19T17:03:09
{ "title": [ -0.01550293, -0.015609741, -0.043670654, -0.06225586, -0.010299683, 0.009712219, -0.052642822, 0.005405426, 0.0064888, 0.033111572, -0.025756836, 0.013877869, 0.02067566, -0.012817383, -0.03253174, 0.005176544, -0.011367798, 0.024993...
http://arxiv.org/abs/2305.11808v1
Pseudo-Label Training and Model Inertia in Neural Machine Translation
Benjamin Hsu, Anna Currey, Xing Niu, Maria Nădejde, Georgiana Dinu
cs.CL
Like many other machine learning applications, neural machine translation (NMT) benefits from over-parameterized deep neural models. However, these models have been observed to be brittle: NMT model predictions are sensitive to small input changes and can show significant variation across re-training or incremental mod...
http://arxiv.org/pdf/2305.11808v1
2023-05-19T16:45:19
2023-05-19T16:45:19
{ "title": [ -0.031951904, -0.0057754517, -0.022155762, -0.0069389343, -0.02330017, -0.0063095093, 0.0031719208, -0.0015449524, -0.0129470825, 0.057891846, 0.008834839, -0.0084991455, 0.020843506, -0.029327393, -0.04864502, 0.01361084, -0.006706238, ...
http://arxiv.org/abs/2305.11806v1
The Inside Story: Towards Better Understanding of Machine Translation Neural Evaluation Metrics
Ricardo Rei, Nuno M. Guerreiro, Marcos Treviso, Luisa Coheur, Alon Lavie, André F. T. Martins
cs.CL
Neural metrics for machine translation evaluation, such as COMET, exhibit significant improvements in their correlation with human judgments, as compared to traditional metrics based on lexical overlap, such as BLEU. Yet, neural metrics are, to a great extent, "black boxes" returning a single sentence-level score witho...
http://arxiv.org/pdf/2305.11806v1
2023-05-19T16:42:17
2023-05-19T16:42:17
{ "title": [ -0.04397583, -0.0061912537, -0.0073051453, -0.0019607544, -0.016662598, -0.0027122498, -0.013404846, -0.015457153, -0.017089844, 0.036102295, 0.000116705894, 0.0025691986, 0.013626099, -0.02734375, -0.042999268, 0.009811401, -0.01423645,...
http://arxiv.org/abs/2305.11792v2
Cue-CoT: Chain-of-thought Prompting for Responding to In-depth Dialogue Questions with LLMs
Hongru Wang, Rui Wang, Fei Mi, Yang Deng, Zezhong Wang, Bin Liang, Ruifeng Xu, Kam-Fai Wong
cs.CL, cs.AI
Large Language Models (LLMs), such as \texttt{ChatGPT}, greatly empower dialogue systems with strong language understanding and generation capabilities. However, most of the previous works prompt the LLMs to directly generate a response based on the dialogue context, overlooking the underlying linguistic cues about the...
http://arxiv.org/pdf/2305.11792v2
2023-10-15T12:54:48
2023-05-19T16:27:43
{ "title": [ -0.044769287, -0.045776367, -0.024871826, 0.0029296875, -0.064941406, -0.009269714, -0.031097412, 0.007255554, 0.007194519, -0.0054779053, -0.05319214, 0.018005371, 0.010673523, -0.03652954, -0.0016441345, -0.03010559, -0.0053100586, ...
http://arxiv.org/abs/2305.11791v1
Enhancing Few-shot NER with Prompt Ordering based Data Augmentation
Huiming Wang, Liying Cheng, Wenxuan Zhang, De Wen Soh, Lidong Bing
cs.CL
Recently, data augmentation (DA) methods have been proven to be effective for pre-trained language models (PLMs) in low-resource settings, including few-shot named entity recognition (NER). However, conventional NER DA methods are mostly aimed at sequence labeling models, i.e., token-level classification, and few are c...
http://arxiv.org/pdf/2305.11791v1
2023-05-19T16:25:43
2023-05-19T16:25:43
{ "title": [ 0.04675293, -0.015213013, -0.036193848, -0.062561035, 0.0015716553, 0.0019340515, -0.002407074, 0.013587952, -0.006324768, 0.00008839369, -0.023345947, 0.010726929, -0.004486084, -0.040924072, 0.0029087067, -0.018234253, 0.0057029724, ...
http://arxiv.org/abs/2305.11790v3
Prompting with Pseudo-Code Instructions
Mayank Mishra, Prince Kumar, Riyaz Bhat, Rudra Murthy V, Danish Contractor, Srikanth Tamilselvam
cs.CL
Prompting with natural language instructions has recently emerged as a popular method of harnessing the capabilities of large language models. Given the inherent ambiguity present in natural language, it is intuitive to consider the possible advantages of prompting with less ambiguous prompt styles, such as the use of ...
http://arxiv.org/pdf/2305.11790v3
2023-10-19T05:04:18
2023-05-19T16:25:01
{ "title": [ 0.011009216, -0.016235352, 0.0014762878, -0.018844604, -0.064331055, -0.003944397, -0.03074646, -0.014808655, -0.021743774, 0.022705078, -0.02633667, -0.032287598, -0.008071899, -0.041229248, 0.013069153, -0.02772522, 0.0597229, -0.0...
http://arxiv.org/abs/2305.11789v1
Solving NLP Problems through Human-System Collaboration: A Discussion-based Approach
Masahiro Kaneko, Graham Neubig, Naoaki Okazaki
cs.CL
Humans work together to solve common problems by having discussions, explaining, and agreeing or disagreeing with each other. Similarly, if a system can have discussions with humans when solving tasks, it can improve the system's performance and reliability. In previous research on explainability, it has only been poss...
http://arxiv.org/pdf/2305.11789v1
2023-05-19T16:24:50
2023-05-19T16:24:50
{ "title": [ -0.020843506, -0.018997192, -0.022003174, -0.040771484, -0.012451172, -0.007209778, -0.036712646, 0.00086164474, -0.0011539459, 0.01701355, -0.02772522, -0.00024604797, -0.016616821, -0.017120361, -0.035461426, -0.021728516, -0.015449524...
http://arxiv.org/abs/2305.11779v1
DMDD: A Large-Scale Dataset for Dataset Mentions Detection
Huitong Pan, Qi Zhang, Eduard Dragut, Cornelia Caragea, Longin Jan Latecki
cs.CL, cs.AI, I.2.7
The recognition of dataset names is a critical task for automatic information extraction in scientific literature, enabling researchers to understand and identify research opportunities. However, existing corpora for dataset mention detection are limited in size and naming diversity. In this paper, we introduce the Dat...
http://arxiv.org/pdf/2305.11779v1
2023-05-19T16:18:00
2023-05-19T16:18:00
{ "title": [ 0.009857178, 0.012199402, -0.056274414, -0.064941406, 0.039764404, -0.019073486, -0.016448975, -0.01285553, 0.0053482056, 0.025131226, -0.03942871, 0.02571106, -0.011184692, 0.008903503, -0.042755127, -0.008430481, -0.018508911, 0.04...
http://arxiv.org/abs/2305.11778v1
Cross-Lingual Supervision improves Large Language Models Pre-training
Andrea Schioppa, Xavier Garcia, Orhan Firat
cs.CL, cs.LG
The recent rapid progress in pre-training Large Language Models has relied on using self-supervised language modeling objectives like next token prediction or span corruption. On the other hand, Machine Translation Systems are mostly trained using cross-lingual supervision that requires aligned data between source and ...
http://arxiv.org/pdf/2305.11778v1
2023-05-19T16:14:07
2023-05-19T16:14:07
{ "title": [ -0.011138916, -0.010673523, -0.03805542, -0.079711914, -0.016067505, -0.013870239, -0.0770874, 0.003030777, -0.027130127, 0.015411377, -0.027832031, 0.007286072, 0.012702942, -0.01940918, 0.0037441254, -0.008232117, -0.020233154, 0.0...
http://arxiv.org/abs/2305.11769v2
Enhancing Vision-Language Pre-Training with Jointly Learned Questioner and Dense Captioner
Zikang Liu, Sihan Chen, Longteng Guo, Handong Li, Xingjian He, Jing Liu
cs.CV, cs.AI, cs.CL, cs.MM
Large pre-trained multimodal models have demonstrated significant success in a range of downstream tasks, including image captioning, image-text retrieval, visual question answering (VQA), etc. However, many of these methods rely on image-text pairs collected from the web as pre-training data and unfortunately overlook...
http://arxiv.org/pdf/2305.11769v2
2023-08-06T12:29:21
2023-05-19T15:54:40
{ "title": [ 0.0040016174, -0.005760193, -0.035827637, -0.06665039, -0.0018262863, -0.027954102, -0.030731201, 0.0126953125, 0.003955841, 0.026733398, -0.037109375, 0.022338867, 0.012298584, -0.01991272, -0.020858765, -0.008323669, 0.009284973, 0...
http://arxiv.org/abs/2305.11768v2
Generating Visual Spatial Description via Holistic 3D Scene Understanding
Yu Zhao, Hao Fei, Wei Ji, Jianguo Wei, Meishan Zhang, Min Zhang, Tat-Seng Chua
cs.CV, cs.CL
Visual spatial description (VSD) aims to generate texts that describe the spatial relations of the given objects within images. Existing VSD work merely models the 2D geometrical vision features, thus inevitably falling prey to the problem of skewed spatial understanding of target objects. In this work, we investigate ...
http://arxiv.org/pdf/2305.11768v2
2023-05-25T04:20:46
2023-05-19T15:53:56
{ "title": [ -0.011787415, -0.013092041, -0.041748047, -0.045562744, -0.014961243, -0.008880615, -0.03439331, -0.021072388, 0.017623901, 0.013938904, -0.04421997, -0.0032520294, -0.00856781, -0.010360718, -0.036590576, -0.011177063, 0.014419556, ...
http://arxiv.org/abs/2305.11761v1
ReSeTOX: Re-learning attention weights for toxicity mitigation in machine translation
Javier García Gilabert, Carlos Escolano, Marta R. Costa-Jussà
cs.CL
Our proposed method, ReSeTOX (REdo SEarch if TOXic), addresses the issue of Neural Machine Translation (NMT) generating translation outputs that contain toxic words not present in the input. The objective is to mitigate the introduction of toxic language without the need for re-training. In the case of identified added...
http://arxiv.org/pdf/2305.11761v1
2023-05-19T15:46:08
2023-05-19T15:46:08
{ "title": [ -0.0030078888, -0.008644104, 0.00041627884, -0.06903076, -0.0074539185, 0.012908936, -0.050354004, 0.004562378, 0.0064315796, 0.05923462, -0.02331543, 0.013900757, 0.03857422, 0.0007619858, -0.049438477, 0.0135650635, -0.0051231384, ...
http://arxiv.org/abs/2305.11759v1
Controlling the Extraction of Memorized Data from Large Language Models via Prompt-Tuning
Mustafa Safa Ozdayi, Charith Peris, Jack FitzGerald, Christophe Dupuy, Jimit Majmudar, Haidar Khan, Rahil Parikh, Rahul Gupta
cs.CL, cs.AI
Large Language Models (LLMs) are known to memorize significant portions of their training data. Parts of this memorized content have been shown to be extractable by simply querying the model, which poses a privacy risk. We present a novel approach which uses prompt-tuning to control the extraction rates of memorized co...
http://arxiv.org/pdf/2305.11759v1
2023-05-19T15:45:29
2023-05-19T15:45:29
{ "title": [ -0.021438599, -0.009605408, -0.04168701, -0.06048584, -0.019470215, -0.001329422, -0.047668457, 0.008056641, -0.0020771027, 0.03363037, -0.021377563, -0.0064926147, -0.012016296, -0.007663727, -0.05633545, -0.009223938, -0.0064697266, ...
http://arxiv.org/abs/2305.11747v3
HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models
Junyi Li, Xiaoxue Cheng, Wayne Xin Zhao, Jian-Yun Nie, Ji-Rong Wen
cs.CL
Large language models (LLMs), such as ChatGPT, are prone to generate hallucinations, i.e., content that conflicts with the source or cannot be verified by the factual knowledge. To understand what types of content and to which extent LLMs are apt to hallucinate, we introduce the Hallucination Evaluation benchmark for L...
http://arxiv.org/pdf/2305.11747v3
2023-10-23T01:49:32
2023-05-19T15:36:27
{ "title": [ -0.02609253, -0.0057907104, -0.034210205, -0.04949951, -0.006828308, 0.0010766983, -0.040771484, 0.0017185211, 0.007911682, 0.044647217, -0.019256592, 0.012306213, 0.016555786, -0.0050239563, -0.060302734, 0.008178711, -0.022598267, ...
http://arxiv.org/abs/2305.11746v1
HalOmi: A Manually Annotated Benchmark for Multilingual Hallucination and Omission Detection in Machine Translation
David Dale, Elena Voita, Janice Lam, Prangthip Hansanti, Christophe Ropers, Elahe Kalbassi, Cynthia Gao, Loïc Barrault, Marta R. Costa-jussà
cs.CL, I.2.7
Hallucinations in machine translation are translations that contain information completely unrelated to the input. Omissions are translations that do not include some of the input information. While both cases tend to be catastrophic errors undermining user trust, annotated data with these types of pathologies is extre...
http://arxiv.org/pdf/2305.11746v1
2023-05-19T15:33:50
2023-05-19T15:33:50
{ "title": [ -0.014266968, 0.006690979, 0.0015897751, -0.038330078, -0.016708374, 0.015571594, -0.039489746, -0.02178955, 0.008712769, 0.054870605, -0.0007247925, 0.023605347, 0.042877197, -0.012512207, -0.05529785, 0.034240723, -0.025390625, 0.0...