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/2305.14282v3
INSTRUCTSCORE: Explainable Text Generation Evaluation with Finegrained Feedback
Wenda Xu, Danqing Wang, Liangming Pan, Zhenqiao Song, Markus Freitag, William Yang Wang, Lei Li
cs.CL, cs.AI
Automatically evaluating the quality of language generation is critical. Although recent learned metrics show high correlation with human judgement, these metrics can not explain their verdict or associate the scores with defects in generated text. To address this limitation, we present InstructScore, an explainable ev...
http://arxiv.org/pdf/2305.14282v3
2023-10-26T18:21:30
2023-05-23T17:27:22
{ "title": [ -0.03652954, -0.015289307, -0.014343262, -0.053497314, -0.017059326, 0.027816772, -0.0423584, -0.011619568, -0.0018939972, 0.024169922, -0.027389526, 0.028381348, -0.027740479, 0.0027618408, -0.052642822, -0.015655518, -0.019760132, ...
http://arxiv.org/abs/2305.14281v2
Weakly-Supervised Learning of Visual Relations in Multimodal Pretraining
Emanuele Bugliarello, Aida Nematzadeh, Lisa Anne Hendricks
cs.CL, cs.CV
Recent work in vision-and-language pretraining has investigated supervised signals from object detection data to learn better, fine-grained multimodal representations. In this work, we take a step further and explore how we can tap into supervision from small-scale visual relation data. In particular, we propose two pr...
http://arxiv.org/pdf/2305.14281v2
2023-10-19T17:46:34
2023-05-23T17:27:12
{ "title": [ 0.013694763, 0.009941101, -0.04928589, -0.06933594, 0.0004210472, -0.017959595, -0.03930664, -0.008293152, 0.006614685, 0.013587952, -0.00894928, 0.018112183, -0.0054359436, -0.035339355, -0.012557983, -0.0062713623, 0.008995056, 0.0...
http://arxiv.org/abs/2305.14280v2
Multilingual Pixel Representations for Translation and Effective Cross-lingual Transfer
Elizabeth Salesky, Neha Verma, Philipp Koehn, Matt Post
cs.CL
We introduce and demonstrate how to effectively train multilingual machine translation models with pixel representations. We experiment with two different data settings with a variety of language and script coverage, demonstrating improved performance compared to subword embeddings. We explore various properties of pix...
http://arxiv.org/pdf/2305.14280v2
2023-10-24T13:36:49
2023-05-23T17:26:50
{ "title": [ -0.024795532, -0.0051574707, 0.003583908, -0.03302002, -0.0104599, 0.013076782, -0.02406311, -0.02659607, 0.019226074, 0.05596924, -0.013374329, 0.016540527, 0.020233154, -0.014228821, -0.0440979, 0.0234375, 0.01348114, 0.016448975, ...
http://arxiv.org/abs/2305.14279v3
Two Failures of Self-Consistency in the Multi-Step Reasoning of LLMs
Angelica Chen, Jason Phang, Alicia Parrish, Vishakh Padmakumar, Chen Zhao, Samuel R. Bowman, Kyunghyun Cho
cs.CL
Large language models (LLMs) have achieved widespread success on a variety of in-context few-shot tasks, but this success is typically evaluated via correctness rather than consistency. We argue that self-consistency is an important criteria for valid multi-step reasoning in tasks where the solution is composed of the ...
http://arxiv.org/pdf/2305.14279v3
2023-10-02T15:09:41
2023-05-23T17:25:59
{ "title": [ -0.03137207, -0.021072388, -0.034973145, -0.034454346, -0.05291748, 0.0045547485, -0.023666382, 0.01852417, -0.03237915, 0.037322998, -0.023956299, -0.0064926147, 0.0039863586, -0.052337646, -0.0076293945, -0.025558472, 0.005218506, ...
http://arxiv.org/abs/2305.14268v1
Masked Path Modeling for Vision-and-Language Navigation
Zi-Yi Dou, Feng Gao, Nanyun Peng
cs.CV, cs.CL
Vision-and-language navigation (VLN) agents are trained to navigate in real-world environments by following natural language instructions. A major challenge in VLN is the limited availability of training data, which hinders the models' ability to generalize effectively. Previous approaches have attempted to address thi...
http://arxiv.org/pdf/2305.14268v1
2023-05-23T17:20:20
2023-05-23T17:20:20
{ "title": [ -0.034057617, -0.0012187958, -0.024246216, -0.015640259, -0.02192688, -0.010238647, -0.033447266, -0.026611328, 0.019348145, 0.011878967, -0.023834229, -0.022369385, -0.0063209534, -0.043670654, -0.021057129, -0.0054244995, 0.00039672852...
http://arxiv.org/abs/2305.14264v1
Active Learning Principles for In-Context Learning with Large Language Models
Katerina Margatina, Timo Schick, Nikolaos Aletras, Jane Dwivedi-Yu
cs.CL, cs.AI
The remarkable advancements in large language models (LLMs) have significantly enhanced the performance in few-shot learning settings. By using only a small number of labeled examples, referred to as demonstrations, LLMs can effectively grasp the task at hand through in-context learning. However, the process of selecti...
http://arxiv.org/pdf/2305.14264v1
2023-05-23T17:16:04
2023-05-23T17:16:04
{ "title": [ -0.010505676, -0.0053215027, -0.02796936, -0.057556152, 0.014808655, -0.011932373, -0.010505676, -0.0007328987, 0.0044784546, 0.034484863, -0.014678955, 0.0076675415, -0.008712769, -0.021148682, -0.026412964, 0.012992859, -0.0029621124, ...
http://arxiv.org/abs/2305.14263v2
LIMIT: Language Identification, Misidentification, and Translation using Hierarchical Models in 350+ Languages
Milind Agarwal, Md Mahfuz Ibn Alam, Antonios Anastasopoulos
cs.CL, cs.AI
Knowing the language of an input text/audio is a necessary first step for using almost every NLP tool such as taggers, parsers, or translation systems. Language identification is a well-studied problem, sometimes even considered solved; in reality, due to lack of data and computational challenges, current systems canno...
http://arxiv.org/pdf/2305.14263v2
2023-11-06T16:29:21
2023-05-23T17:15:43
{ "title": [ -0.00022304058, 0.015930176, -0.050750732, -0.031234741, -0.003282547, -0.013938904, -0.056640625, -0.011817932, -0.05529785, 0.043670654, -0.024749756, 0.017532349, 0.026245117, -0.046051025, -0.0011377335, -0.014907837, 0.0036392212, ...
http://arxiv.org/abs/2305.14260v2
R2H: Building Multimodal Navigation Helpers that Respond to Help Requests
Yue Fan, Jing Gu, Kaizhi Zheng, Xin Eric Wang
cs.CL
Intelligent navigation-helper agents are critical as they can navigate users in unknown areas through environmental awareness and conversational ability, serving as potential accessibility tools for individuals with disabilities. In this work, we first introduce a novel benchmark, Respond to Help Requests (R2H), to pro...
http://arxiv.org/pdf/2305.14260v2
2023-10-17T17:46:41
2023-05-23T17:12:09
{ "title": [ -0.005607605, -0.00036621094, -0.04272461, -0.061706543, 0.018615723, -0.0104599, -0.02104187, -0.010620117, -0.014549255, 0.029418945, -0.012573242, 0.014205933, 0.0011043549, -0.02432251, -0.014175415, 0.0031967163, 0.0034484863, 0...
http://arxiv.org/abs/2305.14259v3
Learning to Generate Novel Scientific Directions with Contextualized Literature-based Discovery
Qingyun Wang, Doug Downey, Heng Ji, Tom Hope
cs.CL, cs.AI, cs.LG
Literature-Based Discovery (LBD) aims to discover new scientific knowledge by mining papers and generating hypotheses. Standard LBD is limited to predicting pairwise relations between discrete concepts (e.g., drug-disease links), and ignores critical contexts like experimental settings (e.g., a specific patient populat...
http://arxiv.org/pdf/2305.14259v3
2023-10-12T16:10:51
2023-05-23T17:12:08
{ "title": [ -0.023880005, 0.0021362305, -0.04876709, -0.045013428, 0.0013904572, -0.004146576, -0.037231445, -0.019744873, -0.013000488, 0.029144287, 0.024383545, 0.024871826, -0.028533936, -0.025390625, -0.034606934, 0.01878357, -0.004005432, 0...
http://arxiv.org/abs/2305.14257v3
Hierarchical Prompting Assists Large Language Model on Web Navigation
Abishek Sridhar, Robert Lo, Frank F. Xu, Hao Zhu, Shuyan Zhou
cs.CL, cs.AI, cs.LG
Large language models (LLMs) struggle on processing complicated observations in interactive decision making tasks. To alleviate this issue, we propose a simple hierarchical prompting approach. Diverging from previous prompting approaches that always put the full observation (e.g. a web page) to the prompt, we propose t...
http://arxiv.org/pdf/2305.14257v3
2023-10-30T00:55:53
2023-05-23T17:10:39
{ "title": [ 0.026748657, -0.0005044937, -0.029937744, -0.0635376, -0.02003479, -0.025482178, -0.038146973, 0.0010118484, -0.012023926, 0.029876709, -0.028930664, 0.00035429, 0.030380249, -0.03656006, 0.016723633, -0.0007381439, 0.022079468, -0.0...
http://arxiv.org/abs/2305.14256v1
Linear Cross-Lingual Mapping of Sentence Embeddings
Oleg Vasilyev, Fumika Isono, John Bohannon
cs.CL
Semantics of a sentence is defined with much less ambiguity than semantics of a single word, and it should be better preserved by translation to another language. If multilingual sentence embeddings intend to represent sentence semantics, then the similarity between embeddings of any two sentences must be invariant wit...
http://arxiv.org/pdf/2305.14256v1
2023-05-23T17:10:37
2023-05-23T17:10:37
{ "title": [ -0.0049209595, -0.0049476624, -0.004966736, -0.047729492, -0.0005168915, 0.0082473755, -0.04345703, -0.02355957, 0.015007019, 0.022094727, -0.010055542, 0.020202637, 0.009407043, -0.01461792, -0.028503418, 0.021774292, -0.013374329, ...
http://arxiv.org/abs/2305.14251v2
FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation
Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Wei Koh, Mohit Iyyer, Luke Zettlemoyer, Hannaneh Hajishirzi
cs.CL, cs.AI, cs.LG
Evaluating the factuality of long-form text generated by large language models (LMs) is non-trivial because (1) generations often contain a mixture of supported and unsupported pieces of information, making binary judgments of quality inadequate, and (2) human evaluation is time-consuming and costly. In this paper, we ...
http://arxiv.org/pdf/2305.14251v2
2023-10-11T05:27:50
2023-05-23T17:06:00
{ "title": [ -0.035095215, -0.016479492, -0.039886475, -0.050598145, -0.011329651, -0.0069274902, -0.03793335, -0.005382538, 0.0028152466, 0.045166016, -0.009521484, 0.01878357, 0.014915466, -0.015792847, -0.057250977, 0.0050468445, -0.02053833, ...
http://arxiv.org/abs/2305.14250v2
Language Models with Rationality
Nora Kassner, Oyvind Tafjord, Ashish Sabharwal, Kyle Richardson, Hinrich Schuetze, Peter Clark
cs.CL, cs.AI
While large language models (LLMs) are proficient at question-answering (QA), it is not always clear how (or even if) an answer follows from their latent "beliefs". This lack of interpretability is a growing impediment to widespread use of LLMs. To address this, our goals are to make model beliefs and their inferential...
http://arxiv.org/pdf/2305.14250v2
2023-10-29T14:51:48
2023-05-23T17:04:25
{ "title": [ -0.03189087, 0.0034294128, -0.039154053, -0.057769775, -0.019943237, 0.0010709763, -0.077697754, 0.002105713, 0.00843811, 0.026229858, -0.01525116, -0.023925781, 0.0073394775, -0.025741577, -0.039611816, 0.014862061, -0.011749268, 0....
http://arxiv.org/abs/2305.14246v1
Modeling Empathic Similarity in Personal Narratives
Jocelyn Shen, Maarten Sap, Pedro Colon-Hernandez, Hae Won Park, Cynthia Breazeal
cs.CL
The most meaningful connections between people are often fostered through expression of shared vulnerability and emotional experiences in personal narratives. We introduce a new task of identifying similarity in personal stories based on empathic resonance, i.e., the extent to which two people empathize with each other...
http://arxiv.org/pdf/2305.14246v1
2023-05-23T17:00:45
2023-05-23T17:00:45
{ "title": [ -0.025650024, -0.00017797947, -0.025299072, -0.043518066, -0.0025577545, -0.0065574646, -0.01889038, -0.014709473, -0.0067481995, 0.025756836, -0.010940552, 0.0048713684, 0.0006990433, -0.026168823, -0.029922485, -0.0008420944, -0.026062...
http://arxiv.org/abs/2305.14240v1
Revisiting Machine Translation for Cross-lingual Classification
Mikel Artetxe, Vedanuj Goswami, Shruti Bhosale, Angela Fan, Luke Zettlemoyer
cs.CL, cs.AI, cs.LG
Machine Translation (MT) has been widely used for cross-lingual classification, either by translating the test set into English and running inference with a monolingual model (translate-test), or translating the training set into the target languages and finetuning a multilingual model (translate-train). However, most ...
http://arxiv.org/pdf/2305.14240v1
2023-05-23T16:56:10
2023-05-23T16:56:10
{ "title": [ -0.023590088, -0.009132385, -0.0059814453, -0.0340271, 0.0013885498, 0.00027871132, -0.027832031, -0.023284912, -0.008529663, 0.04534912, -0.015281677, 0.011451721, 0.019302368, -0.006160736, -0.0491333, 0.009010315, -0.017074585, 0....
http://arxiv.org/abs/2305.14239v1
On Learning to Summarize with Large Language Models as References
Yixin Liu, Alexander R. Fabbri, Pengfei Liu, Dragomir Radev, Arman Cohan
cs.CL
Recent studies have found that summaries generated by large language models (LLMs) are favored by human annotators over the original reference summaries in commonly used summarization datasets. Therefore, we investigate a new learning paradigm of text summarization models that considers the LLMs as the reference or the...
http://arxiv.org/pdf/2305.14239v1
2023-05-23T16:56:04
2023-05-23T16:56:04
{ "title": [ -0.006324768, 0.004211426, -0.026031494, -0.097961426, -0.0036849976, 0.019210815, -0.0892334, -0.0013723373, 0.0031204224, 0.043670654, -0.019332886, 0.012077332, 0.002374649, 0.006252289, -0.052520752, 0.016357422, -0.0024471283, 0...
http://arxiv.org/abs/2305.14237v1
HOP, UNION, GENERATE: Explainable Multi-hop Reasoning without Rationale Supervision
Wenting Zhao, Justin T. Chiu, Claire Cardie, Alexander M. Rush
cs.CL, cs.AI
Explainable multi-hop question answering (QA) not only predicts answers but also identifies rationales, i. e. subsets of input sentences used to derive the answers. This problem has been extensively studied under the supervised setting, where both answer and rationale annotations are given. Because rationale annotation...
http://arxiv.org/pdf/2305.14237v1
2023-05-23T16:53:49
2023-05-23T16:53:49
{ "title": [ -0.018569946, -0.03186035, -0.044281006, -0.09100342, -0.00013899803, 0.012214661, -0.038726807, 0.034332275, 0.0007209778, -0.0027809143, -0.03149414, 0.02027893, -0.022949219, -0.01171875, -0.011207581, -0.009605408, -0.015220642, ...
http://arxiv.org/abs/2305.14235v2
Multilingual Large Language Models Are Not (Yet) Code-Switchers
Ruochen Zhang, Samuel Cahyawijaya, Jan Christian Blaise Cruz, Genta Indra Winata, Alham Fikri Aji
cs.CL, cs.AI
Multilingual Large Language Models (LLMs) have recently shown great capabilities in a wide range of tasks, exhibiting state-of-the-art performance through zero-shot or few-shot prompting methods. While there have been extensive studies on their abilities in monolingual tasks, the investigation of their potential in the...
http://arxiv.org/pdf/2305.14235v2
2023-10-23T15:17:59
2023-05-23T16:50:48
{ "title": [ -0.023757935, -0.0009326935, -0.025787354, -0.0340271, -0.018432617, -0.003200531, -0.03665161, -0.010269165, -0.018127441, 0.036712646, -0.012062073, -0.024719238, 0.019042969, -0.043823242, -0.023284912, 0.0001295805, -0.009246826, ...
http://arxiv.org/abs/2305.14233v1
Enhancing Chat Language Models by Scaling High-quality Instructional Conversations
Ning Ding, Yulin Chen, Bokai Xu, Yujia Qin, Zhi Zheng, Shengding Hu, Zhiyuan Liu, Maosong Sun, Bowen Zhou
cs.CL, cs.AI
Fine-tuning on instruction data has been widely validated as an effective practice for implementing chat language models like ChatGPT. Scaling the diversity and quality of such data, although straightforward, stands a great chance of leading to improved performance. This paper aims to improve the upper bound of open-so...
http://arxiv.org/pdf/2305.14233v1
2023-05-23T16:49:14
2023-05-23T16:49:14
{ "title": [ -0.0184021, -0.008331299, -0.023162842, -0.051940918, -0.009277344, 0.0042304993, -0.058044434, 0.0028190613, 0.004924774, 0.035614014, -0.01928711, 0.0025806427, 0.0044441223, -0.0104904175, -0.04525757, 0.005142212, -0.014717102, 0...
http://arxiv.org/abs/2305.14232v2
Pre-training Multi-task Contrastive Learning Models for Scientific Literature Understanding
Yu Zhang, Hao Cheng, Zhihong Shen, Xiaodong Liu, Ye-Yi Wang, Jianfeng Gao
cs.CL, cs.DL, cs.IR, cs.LG
Scientific literature understanding tasks have gained significant attention due to their potential to accelerate scientific discovery. Pre-trained language models (LMs) have shown effectiveness in these tasks, especially when tuned via contrastive learning. However, jointly utilizing pre-training data across multiple h...
http://arxiv.org/pdf/2305.14232v2
2023-10-19T01:18:49
2023-05-23T16:47:22
{ "title": [ -0.012893677, 0.010856628, -0.02708435, -0.051239014, 0.0011291504, -0.014884949, -0.02722168, -0.020431519, 0.016052246, 0.033172607, -0.0003080368, 0.006755829, -0.017608643, -0.016662598, -0.053131104, 0.0007853508, -0.0135269165, ...
http://arxiv.org/abs/2305.14230v1
Exploring Representational Disparities Between Multilingual and Bilingual Translation Models
Neha Verma, Kenton Murray, Kevin Duh
cs.CL
Multilingual machine translation has proven immensely useful for low-resource and zero-shot language pairs. However, language pairs in multilingual models sometimes see worse performance than in bilingual models, especially when translating in a one-to-many setting. To understand why, we examine the geometric differenc...
http://arxiv.org/pdf/2305.14230v1
2023-05-23T16:46:18
2023-05-23T16:46:18
{ "title": [ -0.04574585, -0.0029830933, -0.012252808, -0.02482605, -0.026443481, 0.011833191, -0.04446411, -0.021591187, 0.0048179626, 0.05984497, -0.008163452, 0.0023479462, 0.032684326, -0.019836426, -0.05496216, 0.008415222, -0.008911133, 0.0...
http://arxiv.org/abs/2305.14225v1
ManiTweet: A New Benchmark for Identifying Manipulation of News on Social Media
Kung-Hsiang Huang, Hou Pong Chan, Kathleen McKeown, Heng Ji
cs.CL
Considerable advancements have been made to tackle the misrepresentation of information derived from reference articles in the domains of fact-checking and faithful summarization. However, an unaddressed aspect remains - the identification of social media posts that manipulate information within associated news article...
http://arxiv.org/pdf/2305.14225v1
2023-05-23T16:40:07
2023-05-23T16:40:07
{ "title": [ 0.0073013306, 0.0021800995, -0.023254395, -0.040618896, 0.024002075, -0.0082473755, -0.0009274483, -0.030761719, -0.053527832, 0.032684326, -0.013824463, 0.062683105, -0.004787445, -0.0022678375, -0.01838684, -0.0024299622, -0.029327393,...
http://arxiv.org/abs/2305.14224v1
mmT5: Modular Multilingual Pre-Training Solves Source Language Hallucinations
Jonas Pfeiffer, Francesco Piccinno, Massimo Nicosia, Xinyi Wang, Machel Reid, Sebastian Ruder
cs.CL
Multilingual sequence-to-sequence models perform poorly with increased language coverage and fail to consistently generate text in the correct target language in few-shot settings. To address these challenges, we propose mmT5, a modular multilingual sequence-to-sequence model. mmT5 utilizes language-specific modules du...
http://arxiv.org/pdf/2305.14224v1
2023-05-23T16:38:01
2023-05-23T16:38:01
{ "title": [ 0.012435913, -0.008110046, -0.020950317, -0.07287598, -0.03866577, 0.004245758, -0.07751465, 0.0012302399, -0.04901123, 0.03591919, 0.021194458, 0.012374878, 0.050323486, -0.057434082, 0.028671265, 0.015159607, 0.0042800903, 0.024780...
http://arxiv.org/abs/2305.14221v3
Question Answering as Programming for Solving Time-Sensitive Questions
Xinyu Zhu, Cheng Yang, Bei Chen, Siheng Li, Jian-Guang Lou, Yujiu Yang
cs.CL
Question answering plays a pivotal role in human daily life because it involves our acquisition of knowledge about the world. However, due to the dynamic and ever-changing nature of real-world facts, the answer can be completely different when the time constraint in the question changes. Recently, Large Language Models...
http://arxiv.org/pdf/2305.14221v3
2023-10-20T13:21:53
2023-05-23T16:35:16
{ "title": [ 0.0011787415, -0.021438599, -0.02670288, -0.058044434, -0.005783081, -0.02709961, -0.005207062, 0.012916565, 0.007369995, 0.015510559, -0.009780884, -0.0181427, -0.01928711, -0.02558899, -0.064575195, -0.015945435, 0.0021438599, 0.00...
http://arxiv.org/abs/2305.14215v2
Exploring Chain-of-Thought Style Prompting for Text-to-SQL
Chang-You Tai, Ziru Chen, Tianshu Zhang, Xiang Deng, Huan Sun
cs.CL
In-context learning with large language models (LLMs) has recently caught increasing attention due to its superior few-shot performance on various tasks. However, its performance on text-to-SQL parsing still has much room for improvement. In this paper, we hypothesize that a crucial aspect of LLMs to improve for text-t...
http://arxiv.org/pdf/2305.14215v2
2023-10-27T15:21:38
2023-05-23T16:32:36
{ "title": [ 0.0029087067, -0.013809204, -0.036834717, -0.047668457, 0.00016820431, -0.016052246, -0.0090408325, 0.019195557, 0.023971558, 0.03479004, -0.046325684, 0.009765625, 0.019424438, -0.01473999, -0.04498291, -0.008651733, 0.0042877197, 0...
http://arxiv.org/abs/2305.14214v2
CompoundPiece: Evaluating and Improving Decompounding Performance of Language Models
Benjamin Minixhofer, Jonas Pfeiffer, Ivan Vulić
cs.CL
While many languages possess processes of joining two or more words to create compound words, previous studies have been typically limited only to languages with excessively productive compound formation (e.g., German, Dutch) and there is no public dataset containing compound and non-compound words across a large numbe...
http://arxiv.org/pdf/2305.14214v2
2023-10-23T11:17:53
2023-05-23T16:32:27
{ "title": [ -0.03768921, -0.015808105, -0.030471802, -0.087890625, -0.001958847, 0.029342651, -0.097961426, 0.0074157715, 0.0070991516, 0.0423584, -0.009124756, 0.00021457672, 0.018005371, -0.006702423, -0.061462402, 0.024810791, -0.0034770966, ...
http://arxiv.org/abs/2305.14211v1
Towards Graph-hop Retrieval and Reasoning in Complex Question Answering over Textual Database
Minjun Zhu, Yixuan Weng, Shizhu He, Kang Liu, Jun Zhao
cs.CL
In Textual question answering (TQA) systems, complex questions often require retrieving multiple textual fact chains with multiple reasoning steps. While existing benchmarks are limited to single-chain or single-hop retrieval scenarios. In this paper, we propose to conduct Graph-Hop -- a novel multi-chains and multi-ho...
http://arxiv.org/pdf/2305.14211v1
2023-05-23T16:28:42
2023-05-23T16:28:42
{ "title": [ -0.0070228577, -0.02532959, -0.028579712, -0.07067871, -0.004627228, 0.0048294067, -0.023635864, 0.042999268, 0.028152466, 0.04434204, -0.03479004, 0.0040740967, 0.012199402, -0.0027751923, -0.07092285, -0.0016212463, -0.003768921, 0...
http://arxiv.org/abs/2305.14210v2
Skill-Based Few-Shot Selection for In-Context Learning
Shengnan An, Bo Zhou, Zeqi Lin, Qiang Fu, Bei Chen, Nanning Zheng, Weizhu Chen, Jian-Guang Lou
cs.CL, cs.AI
In-context learning is the paradigm that adapts large language models to downstream tasks by providing a few examples. Few-shot selection -- selecting appropriate examples for each test instance separately -- is important for in-context learning. In this paper, we propose Skill-KNN, a skill-based few-shot selection met...
http://arxiv.org/pdf/2305.14210v2
2023-10-10T16:23:33
2023-05-23T16:28:29
{ "title": [ 0.03237915, -0.00504303, -0.021057129, -0.006099701, 0.008857727, -0.039031982, 0.04144287, -0.017349243, -0.0019836426, 0.018707275, -0.029754639, 0.016906738, -0.030792236, -0.019332886, -0.022369385, -0.028198242, 0.0030555725, 0....
http://arxiv.org/abs/2305.14208v1
Domain Private Transformers
Anmol Kabra, Ethan R. Elenberg
cs.CL, cs.LG
Large, general purpose language models have demonstrated impressive performance across many different conversational domains. While multi-domain language models achieve low overall perplexity, their outputs are not guaranteed to stay within the domain of a given input prompt. This paper proposes domain privacy as a nov...
http://arxiv.org/pdf/2305.14208v1
2023-05-23T16:27:12
2023-05-23T16:27:12
{ "title": [ 0.041229248, -0.010055542, -0.023284912, -0.05456543, -0.038238525, -0.032196045, -0.06677246, 0.002067566, -0.045440674, 0.018615723, -0.024383545, 0.010627747, -0.00674057, -0.032836914, 0.04333496, -0.02619934, 0.022369385, 0.0327...
http://arxiv.org/abs/2305.14205v1
$μ$PLAN: Summarizing using a Content Plan as Cross-Lingual Bridge
Fantine Huot, Joshua Maynez, Chris Alberti, Reinald Kim Amplayo, Priyanka Agrawal, Constanza Fierro, Shashi Narayan, Mirella Lapata
cs.CL
Cross-lingual summarization consists of generating a summary in one language given an input document in a different language, allowing for the dissemination of relevant content across speakers of other languages. However, this task remains challenging, mainly because of the need for cross-lingual datasets and the compo...
http://arxiv.org/pdf/2305.14205v1
2023-05-23T16:25:21
2023-05-23T16:25:21
{ "title": [ -0.020721436, -0.017959595, -0.038513184, -0.06427002, -0.030532837, -0.006385803, -0.054260254, 0.0030117035, -0.03866577, 0.04611206, -0.016220093, 0.042510986, 0.03643799, -0.011543274, 0.0005059242, 0.010215759, -0.0010843277, 0....
http://arxiv.org/abs/2305.14202v2
Fine-tuned LLMs Know More, Hallucinate Less with Few-Shot Sequence-to-Sequence Semantic Parsing over Wikidata
Silei Xu, Shicheng Liu, Theo Culhane, Elizaveta Pertseva, Meng-Hsi Wu, Sina J. Semnani, Monica S. Lam
cs.CL
While large language models (LLMs) can answer many questions correctly, they can also hallucinate and give wrong answers. Wikidata, with its over 12 billion facts, can be used to ground LLMs to improve their factuality. This paper presents WikiWebQuestions, a high-quality question answering benchmark for Wikidata. Port...
http://arxiv.org/pdf/2305.14202v2
2023-11-05T19:26:17
2023-05-23T16:20:43
{ "title": [ 0.004688263, -0.0090408325, -0.046783447, -0.068847656, -0.028884888, -0.017242432, -0.055419922, 0.01902771, -0.008117676, 0.010917664, -0.06488037, 0.041534424, -0.029876709, -0.004878998, -0.0011949539, -0.042633057, 0.023620605, ...
http://arxiv.org/abs/2305.14201v1
Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks
Tiedong Liu, Bryan Kian Hsiang Low
cs.LG, cs.AI, cs.CL
We introduce Goat, a fine-tuned LLaMA model that significantly outperforms GPT-4 on a range of arithmetic tasks. Fine-tuned on a synthetically generated dataset, Goat achieves state-of-the-art performance on BIG-bench arithmetic sub-task. In particular, the zero-shot Goat-7B matches or even surpasses the accuracy achie...
http://arxiv.org/pdf/2305.14201v1
2023-05-23T16:20:30
2023-05-23T16:20:30
{ "title": [ 0.01184845, -0.008796692, -0.0077934265, -0.004714966, -0.015823364, -0.013160706, 0.018127441, -0.021911621, -0.025558472, 0.0184021, -0.0042877197, 0.0031547546, -0.01260376, -0.044921875, -0.0017852783, -0.022903442, -0.015792847, ...
http://arxiv.org/abs/2305.14200v1
Accessing Higher Dimensions for Unsupervised Word Translation
Sida I. Wang
cs.CL, cs.LG, I.2.7
The striking ability of unsupervised word translation has been demonstrated with the help of word vectors / pretraining; however, they require large amounts of data and usually fails if the data come from different domains. We propose coocmap, a method that can use either high-dimensional co-occurrence counts or their ...
http://arxiv.org/pdf/2305.14200v1
2023-05-23T16:19:30
2023-05-23T16:19:30
{ "title": [ -0.047698975, -0.007835388, -0.0146865845, -0.029144287, -0.023803711, 0.010154724, -0.03842163, -0.013366699, 0.006629944, 0.052703857, -0.030441284, -0.0002901554, 0.031066895, -0.0038166046, -0.04257202, 0.0022602081, 0.010765076, ...
http://arxiv.org/abs/2305.14196v2
ZeroSCROLLS: A Zero-Shot Benchmark for Long Text Understanding
Uri Shaham, Maor Ivgi, Avia Efrat, Jonathan Berant, Omer Levy
cs.CL, cs.AI, cs.LG, stat.ML
We introduce ZeroSCROLLS, a zero-shot benchmark for natural language understanding over long texts, which contains only test and small validation sets, without training data. We adapt six tasks from the SCROLLS benchmark, and add four new datasets, including two novel information fusing tasks, such as aggregating the p...
http://arxiv.org/pdf/2305.14196v2
2023-10-14T16:06:15
2023-05-23T16:15:31
{ "title": [ -0.012802124, -0.017364502, -0.015945435, -0.064941406, -0.012382507, -0.00029492378, -0.039794922, -0.0009469986, 0.033355713, 0.023513794, -0.01739502, -0.0015773773, -0.009880066, -0.014350891, -0.059783936, 0.0023231506, -0.005729675...
http://arxiv.org/abs/2305.14195v2
How Old is GPT?: The HumBEL Framework for Evaluating Language Models using Human Demographic Data
Anthony Sicilia, Jennifer C. Gates, Malihe Alikhani
cs.CL, cs.AI
While large pre-trained language models (LMs) find greater use across NLP, existing evaluation protocols do not consider how LM language use aligns with particular human demographic groups, which can be an important consideration in conversational AI applications. To remedy this gap, we consider how LM language skills ...
http://arxiv.org/pdf/2305.14195v2
2023-05-24T02:55:24
2023-05-23T16:15:24
{ "title": [ -0.01737976, 0.008613586, -0.034179688, -0.045013428, 0.007637024, -0.0044403076, -0.058135986, -0.022201538, -0.00390625, 0.018676758, -0.015701294, -0.005214691, 0.005470276, -0.027664185, -0.03375244, 0.0027122498, -0.008659363, 0...
http://arxiv.org/abs/2305.14189v2
Beyond Shared Vocabulary: Increasing Representational Word Similarities across Languages for Multilingual Machine Translation
Di Wu, Christof Monz
cs.CL, cs.LG
Using a vocabulary that is shared across languages is common practice in Multilingual Neural Machine Translation (MNMT). In addition to its simple design, shared tokens play an important role in positive knowledge transfer, assuming that shared tokens refer to similar meanings across languages. However, when word overl...
http://arxiv.org/pdf/2305.14189v2
2023-10-10T11:27:49
2023-05-23T16:11:00
{ "title": [ -0.025665283, -0.0049476624, -0.0024986267, -0.046722412, -0.0071258545, 0.015327454, -0.047088623, -0.013290405, -0.004585266, 0.058898926, -0.02029419, 0.036621094, 0.03677368, -0.008415222, -0.032440186, 0.028335571, -0.009460449, ...
http://arxiv.org/abs/2305.14171v2
Probing in Context: Toward Building Robust Classifiers via Probing Large Language Models
Afra Amini, Massimiliano Ciaramita
cs.CL
Large language models are able to learn new tasks in context, where they are provided with instructions and a few annotated examples. However, the effectiveness of in-context learning is dependent on the provided context, and the performance on a downstream task can vary considerably, depending on the instruction. Impo...
http://arxiv.org/pdf/2305.14171v2
2023-09-14T13:58:01
2023-05-23T15:43:04
{ "title": [ -0.011947632, -0.0015277863, -0.020401001, -0.048583984, 0.014419556, -0.0038700104, -0.03378296, -0.022521973, -0.003540039, 0.009262085, -0.033294678, 0.0129470825, -0.012237549, -0.020339966, -0.026275635, -0.017990112, -0.018188477, ...
http://arxiv.org/abs/2305.14169v1
EASE: An Easily-Customized Annotation System Powered by Efficiency Enhancement Mechanisms
Naihao Deng, Yikai Liu, Mingye Chen, Winston Wu, Siyang Liu, Yulong Chen, Yue Zhang, Rada Mihalcea
cs.HC, cs.CL
The performance of current supervised AI systems is tightly connected to the availability of annotated datasets. Annotations are usually collected through annotation tools, which are often designed for specific tasks and are difficult to customize. Moreover, existing annotation tools with an active learning mechanism o...
http://arxiv.org/pdf/2305.14169v1
2023-05-23T15:38:37
2023-05-23T15:38:37
{ "title": [ -0.0006079674, -0.0024147034, -0.03918457, -0.09710693, 0.007030487, -0.013381958, -0.0496521, -0.004371643, 0.026031494, 0.037475586, 0.0017538071, -0.007106781, 0.008331299, -0.03643799, -0.048980713, 0.018096924, 0.008262634, 0.03...
http://arxiv.org/abs/2305.14163v1
Leveraging Open Information Extraction for Improving Few-Shot Trigger Detection Domain Transfer
David Dukić, Kiril Gashteovski, Goran Glavaš, Jan Šnajder
cs.CL, cs.LG
Event detection is a crucial information extraction task in many domains, such as Wikipedia or news. The task typically relies on trigger detection (TD) -- identifying token spans in the text that evoke specific events. While the notion of triggers should ideally be universal across domains, domain transfer for TD from...
http://arxiv.org/pdf/2305.14163v1
2023-05-23T15:27:35
2023-05-23T15:27:35
{ "title": [ 0.030059814, 0.005569458, -0.021331787, -0.056640625, 0.006866455, -0.02520752, -0.015052795, -0.027038574, -0.028579712, 0.014434814, -0.002828598, 0.01713562, 0.0128479, -0.035003662, 0.012481689, -0.0046653748, 0.00894165, -0.0121...
http://arxiv.org/abs/2305.14160v1
Label Words are Anchors: An Information Flow Perspective for Understanding In-Context Learning
Lean Wang, Lei Li, Damai Dai, Deli Chen, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun
cs.CL, cs.LG
In-context learning (ICL) emerges as a promising capability of large language models (LLMs) by providing them with demonstration examples to perform diverse tasks. However, the underlying mechanism of how LLMs learn from the provided context remains under-explored. In this paper, we investigate the working mechanism of...
http://arxiv.org/pdf/2305.14160v1
2023-05-23T15:26:20
2023-05-23T15:26:20
{ "title": [ -0.012107849, -0.018615723, -0.0309906, -0.028839111, -0.0070724487, -0.028701782, -0.0021343231, -0.011734009, -0.030670166, 0.019592285, -0.019104004, 0.016113281, -0.0211792, -0.023376465, -0.021850586, -0.021194458, -0.008552551, ...
http://arxiv.org/abs/2305.14150v1
WYWEB: A NLP Evaluation Benchmark For Classical Chinese
Bo Zhou, Qianglong Chen, Tianyu Wang, Xiaomi Zhong, Yin Zhang
cs.CL, cs.AI
To fully evaluate the overall performance of different NLP models in a given domain, many evaluation benchmarks are proposed, such as GLUE, SuperGLUE and CLUE. The fi eld of natural language understanding has traditionally focused on benchmarks for various tasks in languages such as Chinese, English, and multilingua, h...
http://arxiv.org/pdf/2305.14150v1
2023-05-23T15:15:11
2023-05-23T15:15:11
{ "title": [ -0.029708862, 0.004009247, -0.026565552, -0.039611816, 0.014129639, 0.0011091232, -0.031311035, -0.0058250427, 0.020980835, 0.019134521, -0.016113281, 0.008964539, 0.0065612793, -0.021453857, -0.048980713, 0.019927979, -0.024505615, ...
http://arxiv.org/abs/2305.14128v1
Dr.ICL: Demonstration-Retrieved In-context Learning
Man Luo, Xin Xu, Zhuyun Dai, Panupong Pasupat, Mehran Kazemi, Chitta Baral, Vaiva Imbrasaite, Vincent Y Zhao
cs.CL, cs.AI
In-context learning (ICL), teaching a large language model (LLM) to perform a task with few-shot demonstrations rather than adjusting the model parameters, has emerged as a strong paradigm for using LLMs. While early studies primarily used a fixed or random set of demonstrations for all test queries, recent research su...
http://arxiv.org/pdf/2305.14128v1
2023-05-23T14:55:25
2023-05-23T14:55:25
{ "title": [ 0.011360168, 0.016418457, -0.052764893, -0.05609131, 0.010398865, -0.026916504, -0.035827637, -0.009651184, -0.020492554, 0.017364502, -0.034423828, -0.0019245148, -0.017105103, -0.03704834, 0.0023212433, -0.038269043, -0.0020332336, ...
http://arxiv.org/abs/2305.14126v1
To Copy Rather Than Memorize: A Vertical Learning Paradigm for Knowledge Graph Completion
Rui Li, Xu Chen, Chaozhuo Li, Yanming Shen, Jianan Zhao, Yujing Wang, Weihao Han, Hao Sun, Weiwei Deng, Qi Zhang, Xing Xie
cs.CL, cs.AI
Embedding models have shown great power in knowledge graph completion (KGC) task. By learning structural constraints for each training triple, these methods implicitly memorize intrinsic relation rules to infer missing links. However, this paper points out that the multi-hop relation rules are hard to be reliably memor...
http://arxiv.org/pdf/2305.14126v1
2023-05-23T14:53:20
2023-05-23T14:53:20
{ "title": [ -0.013771057, -0.013404846, -0.07647705, -0.09082031, 0.0115737915, -0.0074501038, -0.045928955, 0.028762817, -0.006664276, 0.028579712, -0.025527954, 0.03274536, -0.009552002, 0.006511688, -0.05255127, -0.015197754, -0.0015363693, 0...
http://arxiv.org/abs/2305.14124v2
When Does Monolingual Data Help Multilingual Translation: The Role of Domain and Model Scale
Christos Baziotis, Biao Zhang, Alexandra Birch, Barry Haddow
cs.CL
Multilingual machine translation (MMT), trained on a mixture of parallel and monolingual data, is key for improving translation in low-resource language pairs. However, the literature offers conflicting results on the performance of different methods of including monolingual data. To resolve this, we examine how denois...
http://arxiv.org/pdf/2305.14124v2
2023-10-18T09:17:37
2023-05-23T14:48:42
{ "title": [ -0.035247803, -0.002506256, -0.04034424, -0.04840088, -0.007286072, -0.008369446, -0.044769287, -0.015220642, -0.0289917, 0.045196533, 0.0009202957, 0.019104004, 0.01914978, -0.029190063, -0.028274536, 0.012756348, -0.010154724, 0.02...
http://arxiv.org/abs/2305.14106v1
Better Zero-Shot Reasoning with Self-Adaptive Prompting
Xingchen Wan, Ruoxi Sun, Hanjun Dai, Sercan O. Arik, Tomas Pfister
cs.CL, cs.AI, cs.LG
Modern large language models (LLMs) have demonstrated impressive capabilities at sophisticated tasks, often through step-by-step reasoning similar to humans. This is made possible by their strong few and zero-shot abilities -- they can effectively learn from a handful of handcrafted, completed responses ("in-context ex...
http://arxiv.org/pdf/2305.14106v1
2023-05-23T14:27:16
2023-05-23T14:27:16
{ "title": [ 0.008964539, -0.01486969, -0.039398193, -0.07293701, -0.0026454926, -0.01058197, -0.027236938, 0.018066406, 0.016296387, 0.010185242, -0.0064468384, -0.0020561218, -0.03314209, -0.025772095, -0.03540039, 0.008987427, 0.02722168, -0.0...
http://arxiv.org/abs/2305.14105v2
CTQScorer: Combining Multiple Features for In-context Example Selection for Machine Translation
Aswanth Kumar, Ratish Puduppully, Raj Dabre, Anoop Kunchukuttan
cs.CL, cs.AI
Large language models have demonstrated the capability to perform on machine translation when the input is prompted with a few examples (in-context learning). Translation quality depends on various features of the selected examples, such as their quality and relevance, but previous work has predominantly focused on ind...
http://arxiv.org/pdf/2305.14105v2
2023-10-21T14:22:02
2023-05-23T14:26:17
{ "title": [ -0.003660202, -0.02279663, -0.0055503845, -0.032104492, -0.012550354, 0.0021419525, -0.0289917, -0.022445679, 0.01751709, 0.047973633, -0.015304565, 0.014175415, 0.038604736, -0.024154663, -0.046569824, 0.009109497, -0.0034255981, 0....
http://arxiv.org/abs/2305.14104v1
Out-of-Distribution Generalization in Text Classification: Past, Present, and Future
Linyi Yang, Yaoxiao Song, Xuan Ren, Chenyang Lyu, Yidong Wang, Lingqiao Liu, Jindong Wang, Jennifer Foster, Yue Zhang
cs.CL, cs.AI
Machine learning (ML) systems in natural language processing (NLP) face significant challenges in generalizing to out-of-distribution (OOD) data, where the test distribution differs from the training data distribution. This poses important questions about the robustness of NLP models and their high accuracy, which may ...
http://arxiv.org/pdf/2305.14104v1
2023-05-23T14:26:11
2023-05-23T14:26:11
{ "title": [ 0.0010280609, 0.0014181137, -0.013496399, -0.03552246, 0.014877319, -0.009132385, 0.010887146, -0.02494812, 0.013656616, 0.028457642, -0.03326416, 0.031951904, -0.029525757, -0.004524231, -0.05795288, -0.016677856, -0.02861023, 0.067...
http://arxiv.org/abs/2305.14091v3
Revisiting Acceptability Judgements
Hai Hu, Ziyin Zhang, Weifang Huang, Jackie Yan-Ki Lai, Aini Li, Yina Patterson, Jiahui Huang, Peng Zhang, Chien-Jer Charles Lin, Rui Wang
cs.CL, cs.AI
In this work, we revisit linguistic acceptability in the context of large language models. We introduce CoLAC - Corpus of Linguistic Acceptability in Chinese, the first large-scale acceptability dataset for a non-Indo-European language. It is verified by native speakers and is the first acceptability dataset that comes...
http://arxiv.org/pdf/2305.14091v3
2023-09-28T02:58:48
2023-05-23T14:16:22
{ "title": [ 0.024032593, 0.016906738, 0.0009036064, -0.008651733, -0.024475098, -0.016296387, 0.012336731, -0.013282776, 0.01713562, 0.017044067, -0.04260254, -0.053649902, -0.03829956, -0.024642944, -0.047027588, -0.032958984, -0.0011911392, 0....
http://arxiv.org/abs/2305.14087v1
BM25 Query Augmentation Learned End-to-End
Xiaoyin Chen, Sam Wiseman
cs.CL, cs.IR
Given BM25's enduring competitiveness as an information retrieval baseline, we investigate to what extent it can be even further improved by augmenting and re-weighting its sparse query-vector representation. We propose an approach to learning an augmentation and a re-weighting end-to-end, and we find that our approach...
http://arxiv.org/pdf/2305.14087v1
2023-05-23T14:11:42
2023-05-23T14:11:42
{ "title": [ 0.04333496, 0.0041122437, -0.05001831, -0.0569458, -0.0020122528, -0.029418945, -0.034057617, 0.0154800415, 0.01399231, 0.033477783, -0.016952515, -0.019302368, 0.021148682, -0.040924072, 0.009353638, 0.0015087128, 0.0013618469, 0.02...
http://arxiv.org/abs/2305.14083v2
Counterfactual Augmentation for Multimodal Learning Under Presentation Bias
Victoria Lin, Louis-Philippe Morency, Dimitrios Dimitriadis, Srinagesh Sharma
cs.LG, cs.CL
In real-world machine learning systems, labels are often derived from user behaviors that the system wishes to encourage. Over time, new models must be trained as new training examples and features become available. However, feedback loops between users and models can bias future user behavior, inducing a presentation ...
http://arxiv.org/pdf/2305.14083v2
2023-10-30T22:05:11
2023-05-23T14:09:47
{ "title": [ 0.01625061, 0.019454956, -0.03869629, -0.020217896, 0.004535675, -0.032592773, 0.00024485588, -0.035308838, 0.016326904, 0.01927185, -0.028396606, 0.016662598, -0.008483887, -0.024902344, -0.017929077, -0.016708374, -0.0069618225, 0....
http://arxiv.org/abs/2305.14081v1
How to Solve Few-Shot Abusive Content Detection Using the Data We Actually Have
Viktor Hangya, Alexander Fraser
cs.CL
Due to the broad range of social media platforms and their user groups, the requirements of abusive language detection systems are varied and ever-changing. Already a large set of annotated corpora with different properties and label sets were created, such as hate or misogyny detection, but the form and targets of abu...
http://arxiv.org/pdf/2305.14081v1
2023-05-23T14:04:12
2023-05-23T14:04:12
{ "title": [ 0.018035889, -0.01576233, -0.032836914, 0.00014030933, 0.006942749, -0.045684814, 0.044433594, 0.00059843063, -0.016723633, -0.008392334, -0.050750732, 0.0026378632, -0.0231781, -0.02798462, 0.010368347, -0.062408447, 0.011680603, 0....
http://arxiv.org/abs/2305.14071v1
Disentangled Variational Autoencoder for Emotion Recognition in Conversations
Kailai Yang, Tianlin Zhang, Sophia Ananiadou
cs.CL, cs.SD, eess.AS
In Emotion Recognition in Conversations (ERC), the emotions of target utterances are closely dependent on their context. Therefore, existing works train the model to generate the response of the target utterance, which aims to recognise emotions leveraging contextual information. However, adjacent response generation i...
http://arxiv.org/pdf/2305.14071v1
2023-05-23T13:50:06
2023-05-23T13:50:06
{ "title": [ -0.004512787, 0.007133484, -0.0569458, -0.05999756, 0.0075416565, -0.01525116, -0.046081543, 0.0015220642, 0.040130615, 0.039794922, -0.011451721, -0.014228821, 0.026428223, -0.02305603, -0.056732178, 0.014846802, -0.008712769, 0.027...
http://arxiv.org/abs/2305.14070v2
Assessing Linguistic Generalisation in Language Models: A Dataset for Brazilian Portuguese
Rodrigo Wilkens, Leonardo Zilio, Aline Villavicencio
cs.CL
Much recent effort has been devoted to creating large-scale language models. Nowadays, the most prominent approaches are based on deep neural networks, such as BERT. However, they lack transparency and interpretability, and are often seen as black boxes. This affects not only their applicability in downstream tasks but...
http://arxiv.org/pdf/2305.14070v2
2023-06-07T08:53:08
2023-05-23T13:49:14
{ "title": [ -0.05709839, -0.0017738342, -0.035064697, -0.040924072, -0.014343262, 0.023239136, -0.06933594, 0.0068893433, -0.010353088, 0.04522705, -0.021224976, 0.0066337585, 0.008544922, -0.010353088, -0.06298828, -0.00121212, -0.0256958, 0.05...
http://arxiv.org/abs/2305.14069v2
Evaluating Factual Consistency of Summaries with Large Language Models
Shiqi Chen, Siyang Gao, Junxian He
cs.CL
Detecting factual errors in summaries has been an important and challenging subject in summarization research. Inspired by the emergent ability of large language models (LLMs), we explore evaluating factual consistency of summaries by directly prompting LLMs. We present a comprehensive empirical study to assess the abi...
http://arxiv.org/pdf/2305.14069v2
2023-10-12T06:20:42
2023-05-23T13:48:32
{ "title": [ -0.019439697, 0.0037784576, -0.047943115, -0.07788086, -0.003320694, 0.0007843971, -0.06750488, 0.00094747543, 0.0029296875, 0.04586792, -0.015464783, -0.0046157837, 0.0034713745, 0.006881714, -0.07232666, 0.0043640137, -0.020019531, ...
http://arxiv.org/abs/2305.14066v2
One-stop Training of Multiple Capacity Models
Lan Jiang, Haoyang Huang, Dongdong Zhang, Rui Jiang, Furu Wei
cs.CL
Training models with varying capacities can be advantageous for deploying them in different scenarios. While high-capacity models offer better performance, low-capacity models require fewer computing resources for training and inference. In this work, we propose a novel one-stop training framework to jointly train high...
http://arxiv.org/pdf/2305.14066v2
2023-05-24T09:37:47
2023-05-23T13:44:09
{ "title": [ -0.013015747, 0.0011339188, -0.036712646, -0.06488037, -0.01878357, 0.011016846, -0.051971436, -0.005332947, 0.00052785873, 0.021194458, -0.026794434, 0.013717651, -0.027023315, -0.02355957, -0.0048294067, 0.0059547424, -0.0023174286, ...
http://arxiv.org/abs/2305.14057v1
Can Language Models Understand Physical Concepts?
Lei Li, Jingjing Xu, Qingxiu Dong, Ce Zheng, Qi Liu, Lingpeng Kong, Xu Sun
cs.CL, cs.CV
Language models~(LMs) gradually become general-purpose interfaces in the interactive and embodied world, where the understanding of physical concepts is an essential prerequisite. However, it is not yet clear whether LMs can understand physical concepts in the human world. To investigate this, we design a benchmark VEC...
http://arxiv.org/pdf/2305.14057v1
2023-05-23T13:36:55
2023-05-23T13:36:55
{ "title": [ -0.020111084, 0.0025482178, -0.002828598, -0.02355957, -0.06008911, 0.0063476562, -0.08288574, -0.0335083, 0.0010557175, 0.044830322, -0.016998291, -0.019515991, 0.052978516, -0.030914307, -0.0036563873, 0.00957489, 0.0013065338, 0.0...
http://arxiv.org/abs/2305.14049v1
Rethinking Speech Recognition with A Multimodal Perspective via Acoustic and Semantic Cooperative Decoding
Tian-Hao Zhang, Hai-Bo Qin, Zhi-Hao Lai, Song-Lu Chen, Qi Liu, Feng Chen, Xinyuan Qian, Xu-Cheng Yin
cs.CL, cs.SD, eess.AS
Attention-based encoder-decoder (AED) models have shown impressive performance in ASR. However, most existing AED methods neglect to simultaneously leverage both acoustic and semantic features in decoder, which is crucial for generating more accurate and informative semantic states. In this paper, we propose an Acousti...
http://arxiv.org/pdf/2305.14049v1
2023-05-23T13:25:44
2023-05-23T13:25:44
{ "title": [ -0.035736084, 0.005756378, -0.056884766, -0.0592041, -0.0068969727, -0.0090789795, -0.06616211, 0.012046814, -0.013214111, 0.05316162, -0.022598267, 0.004058838, 0.014709473, -0.009864807, -0.054626465, -0.0027446747, -0.016937256, 0...
http://arxiv.org/abs/2305.14045v2
The CoT Collection: Improving Zero-shot and Few-shot Learning of Language Models via Chain-of-Thought Fine-Tuning
Seungone Kim, Se June Joo, Doyoung Kim, Joel Jang, Seonghyeon Ye, Jamin Shin, Minjoon Seo
cs.CL, cs.AI, cs.LG
Language models (LMs) with less than 100B parameters are known to perform poorly on chain-of-thought (CoT) reasoning in contrast to large LMs when solving unseen tasks. In this work, we aim to equip smaller LMs with the step-by-step reasoning capability by instruction tuning with CoT rationales. In order to achieve thi...
http://arxiv.org/pdf/2305.14045v2
2023-10-14T10:46:55
2023-05-23T13:14:59
{ "title": [ -0.0029201508, -0.029968262, -0.011009216, -0.048797607, -0.021438599, -0.0025234222, -0.03994751, 0.020629883, -0.0050201416, 0.026947021, -0.031021118, 0.02558899, -0.0060272217, -0.009346008, -0.046691895, 0.0013628006, -0.0012140274,...
http://arxiv.org/abs/2305.14044v1
Process-To-Text: A Framework for the Quantitative Description of Processes in Natural Language
Yago Fontenla-Seco, Alberto Bugarín-Diz, Manuel Lama
cs.CL
In this paper we present the Process-To-Text (P2T) framework for the automatic generation of textual descriptive explanations of processes. P2T integrates three AI paradigms: process mining for extracting temporal and structural information from a process, fuzzy linguistic protoforms for modelling uncertain terms, and ...
http://arxiv.org/pdf/2305.14044v1
2023-05-23T13:14:34
2023-05-23T13:14:34
{ "title": [ -0.024856567, -0.01335907, -0.02645874, -0.05911255, 0.010131836, 0.0069999695, -0.045959473, -0.02583313, 0.0021591187, 0.03036499, -0.014404297, 0.022216797, 0.008644104, -0.015289307, -0.040893555, 0.017822266, -0.01374054, 0.0301...
http://arxiv.org/abs/2305.14042v1
Improving speech translation by fusing speech and text
Wenbiao Yin, Zhicheng Liu, Chengqi Zhao, Tao Wang, Jian Tong, Rong Ye
cs.CL, cs.SD, eess.AS
In speech translation, leveraging multimodal data to improve model performance and address limitations of individual modalities has shown significant effectiveness. In this paper, we harness the complementary strengths of speech and text, which are disparate modalities. We observe three levels of modality gap between t...
http://arxiv.org/pdf/2305.14042v1
2023-05-23T13:13:48
2023-05-23T13:13:48
{ "title": [ -0.0052337646, 0.0018177032, -0.025527954, -0.05218506, -0.022567749, -0.004245758, -0.041107178, 0.009162903, -0.011512756, 0.07116699, -0.020996094, -0.0026798248, 0.032287598, -0.019866943, -0.02796936, 0.009010315, 0.010444641, 0...
http://arxiv.org/abs/2305.14020v2
Does ChatGPT have Theory of Mind?
Bart Holterman, Kees van Deemter
cs.CL, cs.AI
Theory of Mind (ToM) is the ability to understand human thinking and decision-making, an ability that plays a crucial role in social interaction between people, including linguistic communication. This paper investigates to what extent recent Large Language Models in the ChatGPT tradition possess ToM. We posed six well...
http://arxiv.org/pdf/2305.14020v2
2023-09-13T11:22:19
2023-05-23T12:55:21
{ "title": [ -0.0032520294, -0.02420044, 0.03149414, 0.050231934, -0.054901123, 0.005458832, 0.0037212372, -0.038085938, -0.030731201, 0.024841309, 0.008506775, 0.011054993, 0.00894165, -0.02809143, 0.014144897, -0.00033020973, -0.015113831, 0.01...
http://arxiv.org/abs/2305.14016v2
Target-Agnostic Gender-Aware Contrastive Learning for Mitigating Bias in Multilingual Machine Translation
Minwoo Lee, Hyukhun Koh, Kang-il Lee, Dongdong Zhang, Minsung Kim, Kyomin Jung
cs.CL
Gender bias is a significant issue in machine translation, leading to ongoing research efforts in developing bias mitigation techniques. However, most works focus on debiasing bilingual models without much consideration for multilingual systems. In this paper, we specifically target the gender bias issue of multilingua...
http://arxiv.org/pdf/2305.14016v2
2023-11-10T04:35:23
2023-05-23T12:53:39
{ "title": [ -0.026016235, 0.001991272, -0.019622803, -0.037475586, -0.0030822754, -0.0008382797, -0.025131226, -0.0031776428, 0.012374878, 0.056610107, -0.029205322, 0.007369995, 0.039764404, -0.002714157, -0.05065918, 0.0017318726, -0.031173706, ...
http://arxiv.org/abs/2305.14012v1
A Simple Method for Unsupervised Bilingual Lexicon Induction for Data-Imbalanced, Closely Related Language Pairs
Niyati Bafna, Cristina España-Bonet, Josef van Genabith, Benoît Sagot, Rachel Bawden
cs.CL
Existing approaches for unsupervised bilingual lexicon induction (BLI) often depend on good quality static or contextual embeddings trained on large monolingual corpora for both languages. In reality, however, unsupervised BLI is most likely to be useful for dialects and languages that do not have abundant amounts of m...
http://arxiv.org/pdf/2305.14012v1
2023-05-23T12:49:21
2023-05-23T12:49:21
{ "title": [ -0.044799805, -0.00040650368, -0.043884277, -0.03237915, -0.007835388, 0.009048462, -0.0657959, -0.010620117, 0.013694763, 0.014656067, -0.04623413, 0.015823364, -0.008201599, -0.015853882, -0.022781372, -0.018798828, -0.03414917, 0....
http://arxiv.org/abs/2305.16335v1
Robust Representation Learning with Reliable Pseudo-labels Generation via Self-Adaptive Optimal Transport for Short Text Clustering
Xiaolin Zheng, Mengling Hu, Weiming Liu, Chaochao Chen, Xinting Liao
cs.CL, cs.AI, cs.LG
Short text clustering is challenging since it takes imbalanced and noisy data as inputs. Existing approaches cannot solve this problem well, since (1) they are prone to obtain degenerate solutions especially on heavy imbalanced datasets, and (2) they are vulnerable to noises. To tackle the above issues, we propose a Ro...
http://arxiv.org/pdf/2305.16335v1
2023-05-23T12:43:40
2023-05-23T12:43:40
{ "title": [ 0.0032367706, 0.009788513, -0.018005371, -0.020553589, -0.002773285, 0.002067566, -0.011497498, -0.034423828, -0.008613586, 0.04055786, 0.008903503, 0.031204224, -0.010910034, -0.024856567, -0.02658081, 0.009979248, 0.010795593, 0.00...
http://arxiv.org/abs/2305.14010v1
IfQA: A Dataset for Open-domain Question Answering under Counterfactual Presuppositions
Wenhao Yu, Meng Jiang, Peter Clark, Ashish Sabharwal
cs.CL
Although counterfactual reasoning is a fundamental aspect of intelligence, the lack of large-scale counterfactual open-domain question-answering (QA) benchmarks makes it difficult to evaluate and improve models on this ability. To address this void, we introduce the first such dataset, named IfQA, where each question i...
http://arxiv.org/pdf/2305.14010v1
2023-05-23T12:43:19
2023-05-23T12:43:19
{ "title": [ 0.005832672, 0.009208679, -0.036834717, -0.06008911, 0.017471313, -0.0104522705, -0.024505615, -0.02116394, 0.048431396, -0.00806427, -0.04019165, -0.013885498, -0.030639648, -0.014892578, -0.05319214, -0.019042969, -0.003211975, 0.0...
http://arxiv.org/abs/2305.14007v1
When Does Aggregating Multiple Skills with Multi-Task Learning Work? A Case Study in Financial NLP
Jingwei Ni, Zhijing Jin, Qian Wang, Mrinmaya Sachan, Markus Leippold
cs.CL
Multi-task learning (MTL) aims at achieving a better model by leveraging data and knowledge from multiple tasks. However, MTL does not always work -- sometimes negative transfer occurs between tasks, especially when aggregating loosely related skills, leaving it an open question when MTL works. Previous studies show th...
http://arxiv.org/pdf/2305.14007v1
2023-05-23T12:37:14
2023-05-23T12:37:14
{ "title": [ -0.006542206, 0.009963989, -0.06161499, -0.03945923, 0.00491333, -0.034240723, -0.019683838, -0.0056495667, -0.013076782, 0.028839111, -0.03439331, 0.012245178, 0.001332283, 0.0018978119, -0.047210693, -0.027740479, -0.01777649, 0.02...
http://arxiv.org/abs/2305.14006v1
Multi-Granularity Prompts for Topic Shift Detection in Dialogue
Jiangyi Lin, Yaxin Fan, Xiaomin Chu, Peifeng Li, Qiaoming Zhu
cs.CL
The goal of dialogue topic shift detection is to identify whether the current topic in a conversation has changed or needs to change. Previous work focused on detecting topic shifts using pre-trained models to encode the utterance, failing to delve into the various levels of topic granularity in the dialogue and unders...
http://arxiv.org/pdf/2305.14006v1
2023-05-23T12:35:49
2023-05-23T12:35:49
{ "title": [ -0.0022087097, 0.014221191, -0.029693604, -0.046325684, 0.0061035156, -0.005207062, -0.07171631, -0.03930664, 0.022872925, 0.0018005371, -0.011802673, 0.0026111603, 0.0002424717, -0.02909851, -0.023956299, -0.0066566467, -0.015625, 0...
http://arxiv.org/abs/2305.14004v1
Sāmayik: A Benchmark and Dataset for English-Sanskrit Translation
Ayush Maheshwari, Ashim Gupta, Amrith Krishna, Ganesh Ramakrishnan, G. Anil Kumar, Jitin Singla
cs.CL
Sanskrit is a low-resource language with a rich heritage. Digitized Sanskrit corpora reflective of the contemporary usage of Sanskrit, specifically that too in prose, is heavily under-represented at present. Presently, no such English-Sanskrit parallel dataset is publicly available. We release a dataset, S\={a}mayik, o...
http://arxiv.org/pdf/2305.14004v1
2023-05-23T12:32:24
2023-05-23T12:32:24
{ "title": [ -0.04220581, 0.00013899803, -0.0211792, -0.04034424, -0.004486084, 0.015655518, -0.047973633, -0.010147095, 0.0007543564, 0.042999268, -0.0126953125, 0.022125244, 0.007736206, -0.0103302, -0.058929443, 0.010177612, 0.01134491, 0.0137...
http://arxiv.org/abs/2305.14002v1
Improving Language Models via Plug-and-Play Retrieval Feedback
Wenhao Yu, Zhihan Zhang, Zhenwen Liang, Meng Jiang, Ashish Sabharwal
cs.CL
Large language models (LLMs) exhibit remarkable performance across various NLP tasks. However, they often generate incorrect or hallucinated information, which hinders their practical applicability in real-world scenarios. Human feedback has been shown to effectively enhance the factuality and quality of generated cont...
http://arxiv.org/pdf/2305.14002v1
2023-05-23T12:29:44
2023-05-23T12:29:44
{ "title": [ -0.009391785, 0.0012550354, -0.004863739, -0.04940796, -0.0056266785, 0.0129470825, -0.058044434, -0.020751953, 0.004272461, 0.044555664, -0.025665283, 0.023208618, 0.010482788, -0.00039601326, -0.05935669, 0.018218994, -0.0096206665, ...
http://arxiv.org/abs/2305.13999v3
Towards A Unified View of Sparse Feed-Forward Network in Pretraining Large Language Model
Zeyu Leo Liu, Tim Dettmers, Xi Victoria Lin, Veselin Stoyanov, Xian Li
cs.CL, cs.LG
Large and sparse feed-forward layers (S-FFN) such as Mixture-of-Experts (MoE) have proven effective in scaling up Transformers model size for \textit{pretraining} large language models. By only activating part of the FFN parameters conditioning on input, S-FFN improves generalization performance while keeping training ...
http://arxiv.org/pdf/2305.13999v3
2023-10-24T03:41:37
2023-05-23T12:28:37
{ "title": [ -0.0040893555, -0.0116119385, -0.039001465, -0.07128906, -0.01134491, -0.0067329407, -0.058807373, 0.026855469, -0.0054016113, 0.025222778, 0.0041542053, 0.008987427, -0.0056648254, -0.02911377, -0.035583496, 0.0104522705, 0.0053710938, ...
http://arxiv.org/abs/2305.13993v3
Condensing Multilingual Knowledge with Lightweight Language-Specific Modules
Haoran Xu, Weiting Tan, Shuyue Stella Li, Yunmo Chen, Benjamin Van Durme, Philipp Koehn, Kenton Murray
cs.CL
Incorporating language-specific (LS) modules is a proven method to boost performance in multilingual machine translation. This approach bears similarity to Mixture-of-Experts (MoE) because it does not inflate FLOPs. However, the scalability of this approach to hundreds of languages (experts) tends to be unmanageable du...
http://arxiv.org/pdf/2305.13993v3
2023-10-22T17:52:19
2023-05-23T12:21:38
{ "title": [ -0.047821045, -0.044769287, -0.062927246, -0.10656738, -0.016921997, 0.018478394, -0.09075928, 0.043945312, -0.013290405, 0.04348755, 0.0050201416, 0.0041770935, -0.007369995, -0.018737793, -0.03086853, -0.001326561, 0.006416321, 0.0...
http://arxiv.org/abs/2305.13989v1
MasakhaPOS: Part-of-Speech Tagging for Typologically Diverse African Languages
Cheikh M. Bamba Dione, David Adelani, Peter Nabende, Jesujoba Alabi, Thapelo Sindane, Happy Buzaaba, Shamsuddeen Hassan Muhammad, Chris Chinenye Emezue, Perez Ogayo, Anuoluwapo Aremu, Catherine Gitau, Derguene Mbaye, Jonathan Mukiibi, Blessing Sibanda, Bonaventure F. P. Dossou, Andiswa Bukula, Rooweither Mabuya, Allahs...
cs.CL
In this paper, we present MasakhaPOS, the largest part-of-speech (POS) dataset for 20 typologically diverse African languages. We discuss the challenges in annotating POS for these languages using the UD (universal dependencies) guidelines. We conducted extensive POS baseline experiments using conditional random field ...
http://arxiv.org/pdf/2305.13989v1
2023-05-23T12:15:33
2023-05-23T12:15:33
{ "title": [ -0.017440796, -0.022460938, -0.0002603531, -0.05239868, -0.01234436, 0.019744873, -0.05154419, -0.011932373, -0.017486572, 0.036010742, -0.0075263977, 0.05166626, 0.019958496, -0.018447876, -0.016616821, 0.029159546, -0.012619019, 0....
http://arxiv.org/abs/2305.13981v2
Preserving Knowledge Invariance: Rethinking Robustness Evaluation of Open Information Extraction
Ji Qi, Chuchun Zhang, Xiaozhi Wang, Kaisheng Zeng, Jifan Yu, Jinxin Liu, Jiuding Sun, Yuxiang Chen, Lei Hou, Juanzi Li, Bin Xu
cs.CL, cs.AI
The robustness to distribution changes ensures that NLP models can be successfully applied in the realistic world, especially for information extraction tasks. However, most prior evaluation benchmarks have been devoted to validating pairwise matching correctness, ignoring the crucial measurement of robustness. In this...
http://arxiv.org/pdf/2305.13981v2
2023-10-24T06:03:23
2023-05-23T12:05:09
{ "title": [ -0.032348633, 0.017745972, -0.06518555, -0.048583984, 0.017303467, -0.012489319, -0.04397583, -0.024932861, -0.008850098, 0.028640747, -0.037384033, 0.019577026, -0.0112838745, 0.0017881393, -0.042114258, -0.027877808, -0.013244629, ...
http://arxiv.org/abs/2305.13973v1
Effortless Integration of Memory Management into Open-Domain Conversation Systems
Eunbi Choi, Kyoung-Woon On, Gunsoo Han, Sungwoong Kim, Daniel Wontae Nam, Daejin Jo, Seung Eun Rho, Taehwan Kwon, Minjoon Seo
cs.CL
Open-domain conversation systems integrate multiple conversation skills into a single system through a modular approach. One of the limitations of the system, however, is the absence of management capability for external memory. In this paper, we propose a simple method to improve BlenderBot3 by integrating memory mana...
http://arxiv.org/pdf/2305.13973v1
2023-05-23T11:56:17
2023-05-23T11:56:17
{ "title": [ 0.013076782, -0.0013561249, -0.042877197, -0.106933594, 0.010429382, 0.0043258667, -0.056488037, 0.023590088, 0.0071487427, 0.051574707, -0.0021953583, 0.0017728806, 0.016494751, -0.020477295, -0.045837402, 0.010360718, -0.008613586, ...
http://arxiv.org/abs/2305.13972v1
Make a Choice! Knowledge Base Question Answering with In-Context Learning
Chuanyuan Tan, Yuehe Chen, Wenbiao Shao, Wenliang Chen
cs.CL
Question answering over knowledge bases (KBQA) aims to answer factoid questions with a given knowledge base (KB). Due to the large scale of KB, annotated data is impossible to cover all fact schemas in KB, which poses a challenge to the generalization ability of methods that require a sufficient amount of annotated dat...
http://arxiv.org/pdf/2305.13972v1
2023-05-23T11:56:03
2023-05-23T11:56:03
{ "title": [ -0.013038635, -0.0070495605, -0.039154053, -0.06402588, 0.00843811, -0.0015497208, -0.023086548, 0.0038166046, 0.025375366, 0.015388489, -0.042053223, 0.013587952, -0.017211914, -0.0048332214, -0.03945923, -0.00907135, 0.006504059, 0...
http://arxiv.org/abs/2305.13971v5
Grammar-Constrained Decoding for Structured NLP Tasks without Finetuning
Saibo Geng, Martin Josifoski, Maxime Peyrard, Robert West
cs.CL, cs.AI, cs.LG
Despite their impressive performance, large language models (LMs) still struggle with reliably generating complex output structures when not finetuned to follow the required output format exactly. To address this issue, grammar-constrained decoding (GCD) can be used to control the generation of LMs, guaranteeing that t...
http://arxiv.org/pdf/2305.13971v5
2023-11-10T09:37:20
2023-05-23T11:54:37
{ "title": [ -0.04650879, -0.019515991, -0.04547119, -0.06451416, -0.0143585205, 0.010032654, -0.0670166, 0.01235199, 0.014282227, 0.009300232, -0.025802612, 0.004348755, -0.013702393, -0.0128479, -0.040496826, 0.004283905, -0.008453369, 0.041778...
http://arxiv.org/abs/2305.13954v2
Robust Prompt Optimization for Large Language Models Against Distribution Shifts
Moxin Li, Wenjie Wang, Fuli Feng, Yixin Cao, Jizhi Zhang, Tat-Seng Chua
cs.CL, cs.AI
Large Language Model (LLM) has demonstrated significant ability in various Natural Language Processing tasks. However, their effectiveness is highly dependent on the phrasing of the task prompt, leading to research on automatic prompt optimization using labeled task data. We reveal that these prompt optimization techni...
http://arxiv.org/pdf/2305.13954v2
2023-10-16T03:00:12
2023-05-23T11:30:43
{ "title": [ -0.038085938, 0.0041007996, -0.053710938, -0.04156494, -0.0018014908, -0.0018472672, -0.0769043, -0.016021729, 0.0037250519, 0.022018433, -0.009109497, 0.009178162, 0.0014543533, -0.02557373, -0.016799927, 0.0141067505, -0.0034694672, ...
http://arxiv.org/abs/2306.05540v1
DetectLLM: Leveraging Log Rank Information for Zero-Shot Detection of Machine-Generated Text
Jinyan Su, Terry Yue Zhuo, Di Wang, Preslav Nakov
cs.CL, cs.AI, 68T50, F.2.2; I.2.7
With the rapid progress of large language models (LLMs) and the huge amount of text they generated, it becomes more and more impractical to manually distinguish whether a text is machine-generated. Given the growing use of LLMs in social media and education, it prompts us to develop methods to detect machine-generated ...
http://arxiv.org/pdf/2306.05540v1
2023-05-23T11:18:30
2023-05-23T11:18:30
{ "title": [ 0.0051193237, -0.0065345764, -0.016815186, -0.03390503, 0.0032234192, -0.011741638, -0.001004219, -0.020339966, -0.016540527, 0.03692627, -0.0032176971, 0.024398804, -0.0012998581, -0.019577026, -0.046783447, 0.0028438568, -0.023498535, ...
http://arxiv.org/abs/2305.13944v1
Acquiring Frame Element Knowledge with Deep Metric Learning for Semantic Frame Induction
Kosuke Yamada, Ryohei Sasano, Koichi Takeda
cs.CL
The semantic frame induction tasks are defined as a clustering of words into the frames that they evoke, and a clustering of their arguments according to the frame element roles that they should fill. In this paper, we address the latter task of argument clustering, which aims to acquire frame element knowledge, and pr...
http://arxiv.org/pdf/2305.13944v1
2023-05-23T11:02:28
2023-05-23T11:02:28
{ "title": [ -0.021560669, -0.022094727, -0.050354004, -0.048431396, -0.0099105835, -0.012535095, -0.033935547, 0.0075149536, 0.006389618, 0.037719727, -0.0066337585, 0.012672424, 0.025421143, -0.022354126, -0.030731201, 0.016601562, 0.0031833649, ...
http://arxiv.org/abs/2305.13917v1
Generating Data for Symbolic Language with Large Language Models
Jiacheng Ye, Chengzu Li, Lingpeng Kong, Tao Yu
cs.CL, cs.AI
While large language models (LLMs) bring not only performance but also complexity, recent work has started to turn LLMs into data generators rather than task inferencers, where another affordable task model is trained for efficient deployment and inference. However, such an approach has primarily been applied to natura...
http://arxiv.org/pdf/2305.13917v1
2023-05-23T10:44:00
2023-05-23T10:44:00
{ "title": [ -0.03201294, -0.017929077, -0.048553467, -0.074035645, -0.0060424805, 0.01525116, -0.0814209, 0.0036296844, 0.0046539307, 0.030883789, -0.0012989044, -0.005847931, -0.0070114136, -0.016433716, -0.050445557, 0.014328003, -0.012916565, ...
http://arxiv.org/abs/2305.13915v1
DAPR: A Benchmark on Document-Aware Passage Retrieval
Kexin Wang, Nils Reimers, Iryna Gurevych
cs.IR, cs.CL
Recent neural retrieval mainly focuses on ranking short texts and is challenged with long documents. Existing work mainly evaluates either ranking passages or whole documents. However, there are many cases where the users want to find a relevant passage within a long document from a huge corpus, e.g. legal cases, resea...
http://arxiv.org/pdf/2305.13915v1
2023-05-23T10:39:57
2023-05-23T10:39:57
{ "title": [ -0.005821228, -0.010391235, -0.028518677, -0.05609131, 0.0017986298, -0.0052337646, -0.024810791, -0.016082764, 0.023040771, 0.039520264, -0.041229248, 0.02470398, 0.012496948, 0.0060310364, -0.046691895, 0.0077552795, 0.000998497, 0...
http://arxiv.org/abs/2305.13905v1
EfficientSpeech: An On-Device Text to Speech Model
Rowel Atienza
eess.AS, cs.CL, cs.SD
State of the art (SOTA) neural text to speech (TTS) models can generate natural-sounding synthetic voices. These models are characterized by large memory footprints and substantial number of operations due to the long-standing focus on speech quality with cloud inference in mind. Neural TTS models are generally not des...
http://arxiv.org/pdf/2305.13905v1
2023-05-23T10:28:41
2023-05-23T10:28:41
{ "title": [ -0.026000977, -0.011306763, -0.024780273, -0.052886963, 0.0103302, 0.0041236877, -0.033569336, 0.012557983, -0.007598877, 0.050079346, -0.008010864, 0.012939453, 0.02104187, -0.011978149, -0.04647827, 0.010314941, -0.019210815, 0.042...
http://arxiv.org/abs/2305.13903v3
Let's Think Frame by Frame with VIP: A Video Infilling and Prediction Dataset for Evaluating Video Chain-of-Thought
Vaishnavi Himakunthala, Andy Ouyang, Daniel Rose, Ryan He, Alex Mei, Yujie Lu, Chinmay Sonar, Michael Saxon, William Yang Wang
cs.CL, cs.CV
Despite exciting recent results showing vision-language systems' capacity to reason about images using natural language, their capacity for video reasoning remains under-explored. We motivate framing video reasoning as the sequential understanding of a small number of keyframes, thereby leveraging the power and robustn...
http://arxiv.org/pdf/2305.13903v3
2023-11-09T06:50:26
2023-05-23T10:26:42
{ "title": [ 0.02293396, -0.015472412, -0.030044556, -0.049987793, -0.003572464, -0.030578613, 0.0040359497, -0.0067367554, 0.011924744, 0.02545166, -0.05319214, 0.013168335, -0.006511688, -0.019012451, -0.041992188, -0.033813477, -0.008605957, 0...
http://arxiv.org/abs/2305.13899v2
Sequence-Level Knowledge Distillation for Class-Incremental End-to-End Spoken Language Understanding
Umberto Cappellazzo, Muqiao Yang, Daniele Falavigna, Alessio Brutti
eess.AS, cs.CL
The ability to learn new concepts sequentially is a major weakness for modern neural networks, which hinders their use in non-stationary environments. Their propensity to fit the current data distribution to the detriment of the past acquired knowledge leads to the catastrophic forgetting issue. In this work we tackle ...
http://arxiv.org/pdf/2305.13899v2
2023-07-31T19:02:23
2023-05-23T10:24:07
{ "title": [ 0.0043182373, -0.009864807, -0.05593872, -0.12084961, 0.008460999, -0.0026226044, -0.0859375, 0.040863037, 0.022033691, 0.005077362, -0.052001953, -0.012878418, 0.00079631805, -0.00894165, -0.03845215, -0.025741577, 0.0049438477, 0.0...
http://arxiv.org/abs/2305.13888v1
PaD: Program-aided Distillation Specializes Large Models in Reasoning
Xuekai Zhu, Biqing Qi, Kaiyan Zhang, Xingwei Long, Bowen Zhou
cs.CL
While Large Language Models (LLMs) excel in several natural language processing tasks, their size and inaccessibility present challenges for extensive practical application. Previous studies acquire specialized skills through distillation on LLMs, which result in trading generic abilities, called model specialization. ...
http://arxiv.org/pdf/2305.13888v1
2023-05-23T10:11:56
2023-05-23T10:11:56
{ "title": [ 0.017501831, -0.008369446, -0.078308105, -0.068359375, -0.013420105, -0.028839111, -0.0524292, 0.0051231384, 0.005874634, 0.011550903, -0.0013408661, -0.047454834, 0.03756714, -0.068115234, 0.014335632, -0.025909424, -0.004634857, 0....
http://arxiv.org/abs/2305.13877v1
Narrative XL: A Large-scale Dataset For Long-Term Memory Models
Arseny Moskvichev, Ky-Vinh Mai
cs.CL, cs.AI, I.2.7; I.2.6
Despite their tremendous successes, most large language models do not have any long-term memory mechanisms, which restricts their applications. Overcoming this limitation would not only require changes to the typical transformer architectures or training procedures, but also a dataset on which these new models could be...
http://arxiv.org/pdf/2305.13877v1
2023-05-23T09:55:32
2023-05-23T09:55:32
{ "title": [ -0.020263672, 0.002122879, -0.050964355, -0.04724121, 0.013519287, -0.0132751465, -0.022445679, -0.008132935, -0.0010004044, 0.023590088, -0.015716553, 0.008491516, -0.012908936, -0.02267456, -0.03375244, 0.0066452026, -0.01838684, 0...
http://arxiv.org/abs/2305.16334v1
OlaGPT: Empowering LLMs With Human-like Problem-Solving Abilities
Yuanzhen Xie, Tao Xie, Mingxiong Lin, WenTao Wei, Chenglin Li, Beibei Kong, Lei Chen, Chengxiang Zhuo, Bo Hu, Zang Li
cs.CL, cs.AI
In most current research, large language models (LLMs) are able to perform reasoning tasks by generating chains of thought through the guidance of specific prompts. However, there still exists a significant discrepancy between their capability in solving complex reasoning problems and that of humans. At present, most a...
http://arxiv.org/pdf/2305.16334v1
2023-05-23T09:36:51
2023-05-23T09:36:51
{ "title": [ -0.016998291, -0.0022182465, -0.038848877, -0.005962372, -0.04220581, -0.01436615, -0.030715942, -0.022827148, -0.025619507, 0.016540527, -0.06185913, 0.03665161, 0.024124146, -0.011451721, 0.027160645, -0.018478394, -0.005680084, -0...
http://arxiv.org/abs/2305.13863v1
Probing Brain Context-Sensitivity with Masked-Attention Generation
Alexandre Pasquiou, Yair Lakretz, Bertrand Thirion, Christophe Pallier
cs.CL
Two fundamental questions in neurolinguistics concerns the brain regions that integrate information beyond the lexical level, and the size of their window of integration. To address these questions we introduce a new approach named masked-attention generation. It uses GPT-2 transformers to generate word embeddings that...
http://arxiv.org/pdf/2305.13863v1
2023-05-23T09:36:21
2023-05-23T09:36:21
{ "title": [ -0.0023975372, 0.018127441, -0.03768921, -0.00415802, 0.008666992, -0.030532837, 0.007335663, -0.03250122, 0.008926392, 0.051727295, -0.0075950623, -0.015617371, 0.015022278, -0.030334473, -0.06390381, 0.008132935, -0.0016870499, 0.0...
http://arxiv.org/abs/2305.13862v2
A Trip Towards Fairness: Bias and De-Biasing in Large Language Models
Leonardo Ranaldi, Elena Sofia Ruzzetti, Davide Venditti, Dario Onorati, Fabio Massimo Zanzotto
cs.CL
Cheap-to-Build Very Large-Language Models (CtB-LLMs) with affordable training are emerging as the next big revolution in natural language processing and understanding. These CtB-LLMs are democratizing access to trainable Very Large-Language Models (VLLMs) and, thus, may represent the building blocks of many NLP systems...
http://arxiv.org/pdf/2305.13862v2
2023-08-29T13:55:13
2023-05-23T09:35:37
{ "title": [ -0.027908325, -0.0021858215, -0.0579834, -0.0574646, 0.0102005005, -0.0074920654, -0.046783447, -0.0034313202, 0.0126953125, 0.030822754, -0.03527832, -0.00894165, 0.03137207, -0.011367798, -0.04574585, 0.001124382, -0.032287598, 0.0...
http://arxiv.org/abs/2305.13860v1
Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study
Yi Liu, Gelei Deng, Zhengzi Xu, Yuekang Li, Yaowen Zheng, Ying Zhang, Lida Zhao, Tianwei Zhang, Yang Liu
cs.SE, cs.AI, cs.CL
Large Language Models (LLMs), like ChatGPT, have demonstrated vast potential but also introduce challenges related to content constraints and potential misuse. Our study investigates three key research questions: (1) the number of different prompt types that can jailbreak LLMs, (2) the effectiveness of jailbreak prompt...
http://arxiv.org/pdf/2305.13860v1
2023-05-23T09:33:38
2023-05-23T09:33:38
{ "title": [ -0.023635864, -0.02293396, -0.008651733, 0.007820129, -0.02468872, -0.0010099411, -0.0034236908, -0.010017395, -0.05432129, 0.018188477, 0.02104187, 0.009140015, -0.02507019, -0.050476074, -0.014160156, -0.005329132, -0.028533936, 0....
http://arxiv.org/abs/2305.13857v1
Revealing User Familiarity Bias in Task-Oriented Dialogue via Interactive Evaluation
Takyoung Kim, Jamin Shin, Young-Ho Kim, Sanghwan Bae, Sungdong Kim
cs.CL
Most task-oriented dialogue (TOD) benchmarks assume users that know exactly how to use the system by constraining the user behaviors within the system's capabilities via strict user goals, namely "user familiarity" bias. This data bias deepens when it combines with data-driven TOD systems, as it is impossible to fathom...
http://arxiv.org/pdf/2305.13857v1
2023-05-23T09:24:53
2023-05-23T09:24:53
{ "title": [ 0.01007843, 0.0050811768, -0.036712646, -0.043762207, 0.014854431, -0.031555176, -0.0050354004, -0.0124435425, 0.010055542, 0.021499634, -0.021881104, -0.0030021667, -0.022949219, -0.01902771, -0.06689453, -0.034118652, -0.028503418, ...
http://arxiv.org/abs/2305.13850v3
Global Structure Knowledge-Guided Relation Extraction Method for Visually-Rich Document
Xiangnan Chen, Qian Xiao, Juncheng Li, Duo Dong, Jun Lin, Xiaozhong Liu, Siliang Tang
cs.CL, cs.AI
Visual Relation Extraction (VRE) is a powerful means of discovering relationships between entities within visually-rich documents. Existing methods often focus on manipulating entity features to find pairwise relations, yet neglect the more fundamental structural information that links disparate entity pairs together. ...
http://arxiv.org/pdf/2305.13850v3
2023-10-27T04:42:12
2023-05-23T09:18:47
{ "title": [ -0.020812988, -0.017089844, -0.063964844, -0.08239746, -0.010810852, -0.0044822693, -0.052703857, 0.017730713, 0.032196045, 0.014884949, -0.02015686, 0.006248474, 0.010772705, -0.020843506, -0.035491943, 0.002670288, 0.014785767, 0.0...
http://arxiv.org/abs/2305.13844v1
Arukikata Travelogue Dataset with Geographic Entity Mention, Coreference, and Link Annotation
Shohei Higashiyama, Hiroki Ouchi, Hiroki Teranishi, Hiroyuki Otomo, Yusuke Ide, Aitaro Yamamoto, Hiroyuki Shindo, Yuki Matsuda, Shoko Wakamiya, Naoya Inoue, Ikuya Yamada, Taro Watanabe
cs.CL
Geoparsing is a fundamental technique for analyzing geo-entity information in text. We focus on document-level geoparsing, which considers geographic relatedness among geo-entity mentions, and presents a Japanese travelogue dataset designed for evaluating document-level geoparsing systems. Our dataset comprises 200 tra...
http://arxiv.org/pdf/2305.13844v1
2023-05-23T09:07:42
2023-05-23T09:07:42
{ "title": [ 0.019058228, -0.017440796, -0.042633057, -0.03643799, -0.0023040771, -0.031066895, -0.006591797, 0.027374268, 0.0042266846, 0.05709839, -0.023040771, -0.0020828247, 0.06933594, -0.015350342, -0.0039367676, -0.005748749, 0.019805908, ...
http://arxiv.org/abs/2305.13833v2
Reducing Sensitivity on Speaker Names for Text Generation from Dialogues
Qi Jia, Haifeng Tang, Kenny Q. Zhu
cs.CL
Changing speaker names consistently throughout a dialogue should not affect its meaning and corresponding outputs for text generation from dialogues. However, pre-trained language models, serving as the backbone for dialogue-processing tasks, have shown to be sensitive to nuances. This may result in unfairness in real-...
http://arxiv.org/pdf/2305.13833v2
2023-08-20T08:42:07
2023-05-23T08:53:33
{ "title": [ -0.034362793, 0.00064468384, -0.043792725, -0.0513916, -0.003604889, 0.006160736, -0.06604004, 0.0075035095, 0.011199951, 0.03552246, 0.0065727234, 0.012992859, 0.003271103, -0.014877319, -0.05517578, 0.008934021, -0.021148682, 0.042...
http://arxiv.org/abs/2305.13831v1
ZET-Speech: Zero-shot adaptive Emotion-controllable Text-to-Speech Synthesis with Diffusion and Style-based Models
Minki Kang, Wooseok Han, Sung Ju Hwang, Eunho Yang
cs.SD, cs.CL, eess.AS
Emotional Text-To-Speech (TTS) is an important task in the development of systems (e.g., human-like dialogue agents) that require natural and emotional speech. Existing approaches, however, only aim to produce emotional TTS for seen speakers during training, without consideration of the generalization to unseen speaker...
http://arxiv.org/pdf/2305.13831v1
2023-05-23T08:52:00
2023-05-23T08:52:00
{ "title": [ -0.020599365, -0.010848999, -0.046295166, -0.040374756, -0.009918213, -0.009407043, -0.034820557, 0.018173218, 0.021072388, 0.04119873, -0.028244019, 0.014015198, 0.029373169, -0.01727295, -0.038726807, 0.017669678, -0.0126571655, 0....
http://arxiv.org/abs/2305.13829v3
Learning from Mistakes via Cooperative Study Assistant for Large Language Models
Danqing Wang, Lei Li
cs.CL
Large language models (LLMs) have demonstrated their potential to refine their generation based on their own feedback. However, the feedback from LLM itself is often inaccurate, thereby limiting its benefits. In this paper, we propose Study Assistant for Large LAnguage Model (SALAM), a novel framework with an auxiliary...
http://arxiv.org/pdf/2305.13829v3
2023-10-24T16:55:19
2023-05-23T08:51:08
{ "title": [ -0.046905518, -0.014823914, -0.029678345, -0.0463562, -0.03050232, 0.005126953, -0.0569458, 0.004940033, 0.034698486, 0.04067993, -0.05078125, 0.0028305054, 0.030899048, 0.0012111664, -0.053100586, 0.0017051697, -0.024795532, 0.02964...
http://arxiv.org/abs/2305.13826v1
"Is the Pope Catholic?" Applying Chain-of-Thought Reasoning to Understanding Conversational Implicatures
Zae Myung Kim, David E. Taylor, Dongyeop Kang
cs.CL
Conversational implicatures are pragmatic inferences that require listeners to deduce the intended meaning conveyed by a speaker from their explicit utterances. Although such inferential reasoning is fundamental to human communication, recent research indicates that large language models struggle to comprehend these im...
http://arxiv.org/pdf/2305.13826v1
2023-05-23T08:49:50
2023-05-23T08:49:50
{ "title": [ 0.030563354, -0.0067634583, -0.0034427643, -0.025543213, -0.002696991, -0.0357666, 0.004047394, -0.04336548, -0.01689148, 0.021392822, -0.026519775, -0.008735657, 0.012680054, -0.025650024, -0.044830322, -0.035125732, -0.029968262, 0...
http://arxiv.org/abs/2305.13820v1
An Open Dataset and Model for Language Identification
Laurie Burchell, Alexandra Birch, Nikolay Bogoychev, Kenneth Heafield
cs.CL
Language identification (LID) is a fundamental step in many natural language processing pipelines. However, current LID systems are far from perfect, particularly on lower-resource languages. We present a LID model which achieves a macro-average F1 score of 0.93 and a false positive rate of 0.033 across 201 languages, ...
http://arxiv.org/pdf/2305.13820v1
2023-05-23T08:43:42
2023-05-23T08:43:42
{ "title": [ -0.02607727, 0.0076141357, -0.048339844, -0.04397583, 0.012191772, -0.005874634, -0.050048828, -0.009109497, -0.017852783, 0.032318115, -0.0032691956, -0.0020065308, -0.0045776367, -0.023101807, -0.044128418, 0.00048327446, 0.0028266907,...
http://arxiv.org/abs/2305.13817v1
Detecting automatically the layout of clinical documents to enhance the performances of downstream natural language processing
Christel Gérardin, Perceval Wajsbürt, Basile Dura, Alice Calliger, Alexandre Moucher, Xavier Tannier, Romain Bey
cs.CL
Objective:Develop and validate an algorithm for analyzing the layout of PDF clinical documents to improve the performance of downstream natural language processing tasks. Materials and Methods: We designed an algorithm to process clinical PDF documents and extract only clinically relevant text. The algorithm consists o...
http://arxiv.org/pdf/2305.13817v1
2023-05-23T08:38:33
2023-05-23T08:38:33
{ "title": [ -0.027038574, -0.009674072, -0.049713135, -0.057403564, -0.022918701, 0.0019893646, -0.07928467, -0.010635376, -0.000024020672, 0.0335083, -0.033721924, 0.008201599, 0.0231781, -0.014274597, -0.010887146, 0.007003784, 0.00026988983, ...
http://arxiv.org/abs/2305.13812v3
Coarse-to-Fine Contrastive Learning in Image-Text-Graph Space for Improved Vision-Language Compositionality
Harman Singh, Pengchuan Zhang, Qifan Wang, Mengjiao Wang, Wenhan Xiong, Jingfei Du, Yu Chen
cs.CL, cs.CV
Contrastively trained vision-language models have achieved remarkable progress in vision and language representation learning, leading to state-of-the-art models for various downstream multimodal tasks. However, recent research has highlighted severe limitations of these models in their ability to perform compositional...
http://arxiv.org/pdf/2305.13812v3
2023-10-24T21:21:00
2023-05-23T08:28:38
{ "title": [ 0.005870819, 0.001074791, -0.019348145, -0.039398193, -0.0027770996, -0.020126343, -0.015533447, -0.019805908, 0.03048706, 0.027297974, -0.03894043, 0.0206604, -0.016555786, -0.009880066, -0.051452637, -0.019699097, 0.0027275085, 0.0...