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/2212.10405v2
AnnoBERT: Effectively Representing Multiple Annotators' Label Choices to Improve Hate Speech Detection
Wenjie Yin, Vibhor Agarwal, Aiqi Jiang, Arkaitz Zubiaga, Nishanth Sastry
cs.CL, cs.SI
Supervised approaches generally rely on majority-based labels. However, it is hard to achieve high agreement among annotators in subjective tasks such as hate speech detection. Existing neural network models principally regard labels as categorical variables, while ignoring the semantic information in diverse label tex...
http://arxiv.org/pdf/2212.10405v2
2023-01-10T14:48:54
2022-12-20T16:30:11
{ "title": [ 0.020690918, 0.010345459, -0.024139404, -0.043762207, 0.013671875, -0.021392822, 0.0004374981, -0.023910522, -0.0082473755, 0.027832031, -0.013473511, 0.031219482, 0.0068969727, -0.011619568, -0.033355713, 0.0013990402, -0.026657104, ...
http://arxiv.org/abs/2212.10403v2
Towards Reasoning in Large Language Models: A Survey
Jie Huang, Kevin Chen-Chuan Chang
cs.CL, cs.AI
Reasoning is a fundamental aspect of human intelligence that plays a crucial role in activities such as problem solving, decision making, and critical thinking. In recent years, large language models (LLMs) have made significant progress in natural language processing, and there is observation that these models may exh...
http://arxiv.org/pdf/2212.10403v2
2023-05-26T17:59:33
2022-12-20T16:29:03
{ "title": [ -0.03677368, -0.02784729, -0.0368042, -0.103637695, 0.007675171, 0.030273438, -0.08319092, 0.03479004, 0.0158844, 0.022155762, -0.02204895, -0.016433716, 0.006881714, -0.017288208, -0.03918457, 0.0030727386, -0.021072388, 0.036499023...
http://arxiv.org/abs/2212.10400v1
Contrastive Learning Reduces Hallucination in Conversations
Weiwei Sun, Zhengliang Shi, Shen Gao, Pengjie Ren, Maarten de Rijke, Zhaochun Ren
cs.CL, cs.AI
Pre-trained language models (LMs) store knowledge in their parameters and can generate informative responses when used in conversational systems. However, LMs suffer from the problem of "hallucination:" they may generate plausible-looking statements that are irrelevant or factually incorrect. To address this problem, w...
http://arxiv.org/pdf/2212.10400v1
2022-12-20T16:26:18
2022-12-20T16:26:18
{ "title": [ 0.046081543, 0.012931824, -0.0058937073, -0.025314331, -0.021881104, -0.034057617, 0.0046157837, -0.042266846, -0.008415222, 0.03857422, -0.008979797, 0.00035381317, 0.0071525574, -0.043426514, -0.015045166, -0.006088257, -0.007080078, ...
http://arxiv.org/abs/2212.10397v3
Needle in a Haystack: An Analysis of High-Agreement Workers on MTurk for Summarization
Lining Zhang, Simon Mille, Yufang Hou, Daniel Deutsch, Elizabeth Clark, Yixin Liu, Saad Mahamood, Sebastian Gehrmann, Miruna Clinciu, Khyathi Chandu, João Sedoc
cs.CL
To prevent the costly and inefficient use of resources on low-quality annotations, we want a method for creating a pool of dependable annotators who can effectively complete difficult tasks, such as evaluating automatic summarization. Thus, we investigate the recruitment of high-quality Amazon Mechanical Turk workers v...
http://arxiv.org/pdf/2212.10397v3
2023-06-14T01:45:33
2022-12-20T16:25:42
{ "title": [ 0.019577026, 0.0209198, -0.041870117, -0.03829956, -0.009628296, -0.03314209, -0.04208374, -0.020309448, 0.0107040405, 0.051330566, -0.0028133392, 0.022399902, 0.028015137, -0.011268616, -0.025299072, 0.025787354, 0.004547119, 0.0100...
http://arxiv.org/abs/2212.10392v1
Debiasing Stance Detection Models with Counterfactual Reasoning and Adversarial Bias Learning
Jianhua Yuan, Yanyan Zhao, Bing Qin
cs.CL
Stance detection models may tend to rely on dataset bias in the text part as a shortcut and thus fail to sufficiently learn the interaction between the targets and texts. Recent debiasing methods usually treated features learned by small models or big models at earlier steps as bias features and proposed to exclude the...
http://arxiv.org/pdf/2212.10392v1
2022-12-20T16:20:56
2022-12-20T16:20:56
{ "title": [ 0.005847931, 0.027145386, -0.028213501, -0.041900635, 0.013206482, -0.014007568, -0.03866577, -0.05206299, 0.017059326, 0.028121948, -0.038635254, -0.01222229, -0.022369385, -0.012016296, -0.056488037, -0.029418945, -0.0030517578, 0....
http://arxiv.org/abs/2212.10391v2
Empowering Sentence Encoders with Prompting and Label Retrieval for Zero-shot Text Classification
Jimin Hong, Jungsoo Park, Daeyoung Kim, Seongjae Choi, Bokyung Son, Jaewook Kang
cs.CL
With contrastive pre-training, sentence encoders are generally optimized to locate semantically similar samples closer to each other in their embedding spaces. In this work, we focus on the potential of their embedding spaces to be readily adapted to zero-shot text classification, as semantically distinct samples are a...
http://arxiv.org/pdf/2212.10391v2
2023-05-19T08:37:33
2022-12-20T16:18:03
{ "title": [ -0.003921509, -0.003124237, -0.021347046, -0.047943115, -0.00712204, -0.012161255, -0.00970459, 0.0084991455, 0.026168823, 0.03353882, -0.029632568, 0.027740479, 0.00080633163, -0.0036506653, -0.055419922, 0.012573242, -0.009994507, ...
http://arxiv.org/abs/2212.10381v1
To Adapt or to Annotate: Challenges and Interventions for Domain Adaptation in Open-Domain Question Answering
Dheeru Dua, Emma Strubell, Sameer Singh, Pat Verga
cs.CL
Recent advances in open-domain question answering (ODQA) have demonstrated impressive accuracy on standard Wikipedia style benchmarks. However, it is less clear how robust these models are and how well they perform when applied to real-world applications in drastically different domains. While there has been some work ...
http://arxiv.org/pdf/2212.10381v1
2022-12-20T16:06:09
2022-12-20T16:06:09
{ "title": [ -0.0065078735, 0.002204895, -0.031707764, -0.04864502, 0.007358551, -0.008216858, -0.004722595, -0.0038661957, -0.0042495728, 0.03353882, 0.0059890747, 0.011451721, 0.0014286041, -0.009223938, -0.049072266, 0.0034313202, -0.0032901764, ...
http://arxiv.org/abs/2212.10380v2
What Are You Token About? Dense Retrieval as Distributions Over the Vocabulary
Ori Ram, Liat Bezalel, Adi Zicher, Yonatan Belinkov, Jonathan Berant, Amir Globerson
cs.CL, cs.IR
Dual encoders are now the dominant architecture for dense retrieval. Yet, we have little understanding of how they represent text, and why this leads to good performance. In this work, we shed light on this question via distributions over the vocabulary. We propose to interpret the vector representations produced by du...
http://arxiv.org/pdf/2212.10380v2
2023-05-24T12:05:27
2022-12-20T16:03:25
{ "title": [ -0.013999939, 0.0068206787, -0.042663574, -0.01184845, 0.005592346, -0.02468872, 0.00027251244, -0.028396606, 0.023162842, 0.029647827, -0.053955078, 0.0032653809, 0.017593384, -0.005455017, -0.038757324, -0.0012540817, -0.005329132, ...
http://arxiv.org/abs/2212.10378v2
Data Curation Alone Can Stabilize In-context Learning
Ting-Yun Chang, Robin Jia
cs.CL
In-context learning (ICL) enables large language models (LLMs) to perform new tasks by prompting them with a sequence of training examples. However, it is known that ICL is very sensitive to the choice of training examples: randomly sampling examples from a training set leads to high variance in performance. In this pa...
http://arxiv.org/pdf/2212.10378v2
2023-05-24T22:32:56
2022-12-20T15:58:54
{ "title": [ 0.07879639, 0.0030441284, -0.018478394, -0.05657959, -0.0053367615, -0.041259766, 0.003522873, -0.015357971, -0.0128479, 0.04421997, -0.06518555, 0.009513855, 0.0046463013, -0.011047363, -0.006843567, -0.03829956, 0.0012407303, 0.018...
http://arxiv.org/abs/2212.10375v2
Self-Adaptive In-Context Learning: An Information Compression Perspective for In-Context Example Selection and Ordering
Zhiyong Wu, Yaoxiang Wang, Jiacheng Ye, Lingpeng Kong
cs.CL, cs.AI
Despite the surprising few-shot performance of in-context learning (ICL), it is still a common practice to randomly sample examples to serve as context. This paper advocates a new principle for ICL: self-adaptive in-context learning. The self-adaption mechanism is introduced to help each sample find an in-context examp...
http://arxiv.org/pdf/2212.10375v2
2023-05-03T14:43:50
2022-12-20T15:55:21
{ "title": [ 0.017593384, -0.0004823208, -0.051605225, -0.04586792, 0.02720642, -0.018814087, 0.0035896301, -0.0072250366, -0.0021743774, 0.036499023, -0.011413574, 0.0077209473, -0.025939941, -0.026473999, -0.037139893, -0.0096588135, 0.022949219, ...
http://arxiv.org/abs/2212.10356v2
Dissecting Transformer Length Extrapolation via the Lens of Receptive Field Analysis
Ta-Chung Chi, Ting-Han Fan, Alexander I. Rudnicky, Peter J. Ramadge
cs.CL
Length extrapolation permits training a transformer language model on short sequences that preserves perplexities when tested on substantially longer sequences. A relative positional embedding design, ALiBi, has had the widest usage to date. We dissect ALiBi via the lens of receptive field analysis empowered by a novel...
http://arxiv.org/pdf/2212.10356v2
2023-05-23T21:18:09
2022-12-20T15:40:17
{ "title": [ -0.0015630722, -0.004638672, -0.0027599335, 0.017913818, -0.027145386, -0.029052734, 0.002368927, -0.054107666, -0.044830322, 0.020751953, -0.026672363, 0.021194458, -0.029312134, -0.026016235, -0.015808105, -0.03363037, 0.0045280457, ...
http://arxiv.org/abs/2212.10346v1
Does It Affect You? Social and Learning Implications of Using Cognitive-Affective State Recognition for Proactive Human-Robot Tutoring
Matthias Kraus, Diana Betancourt, Wolfgang Minker
cs.RO, cs.CL
Using robots in educational contexts has already shown to be beneficial for a student's learning and social behaviour. For levitating them to the next level of providing more effective and human-like tutoring, the ability to adapt to the user and to express proactivity is fundamental. By acting proactively, intelligent...
http://arxiv.org/pdf/2212.10346v1
2022-12-20T15:31:58
2022-12-20T15:31:58
{ "title": [ 0.018966675, 0.013252258, -0.0010795593, -0.0149383545, -0.006324768, -0.030303955, 0.0034484863, -0.036102295, 0.003622055, 0.020065308, -0.03704834, -0.008125305, 0.008934021, -0.025314331, -0.02935791, -0.032043457, 0.007896423, 0...
http://arxiv.org/abs/2212.10341v2
CoCo: Coherence-Enhanced Machine-Generated Text Detection Under Data Limitation With Contrastive Learning
Xiaoming Liu, Zhaohan Zhang, Yichen Wang, Hang Pu, Yu Lan, Chao Shen
cs.CL
Machine-Generated Text (MGT) detection, a task that discriminates MGT from Human-Written Text (HWT), plays a crucial role in preventing misuse of text generative models, which excel in mimicking human writing style recently. Latest proposed detectors usually take coarse text sequences as input and fine-tune pretrained ...
http://arxiv.org/pdf/2212.10341v2
2023-10-20T13:21:51
2022-12-20T15:26:19
{ "title": [ -0.028411865, 0.012680054, -0.035491943, -0.02645874, 0.014350891, -0.015312195, -0.013191223, -0.027816772, 0.0008735657, 0.052978516, -0.03137207, 0.021575928, 0.023498535, 0.0010099411, -0.046295166, 0.00819397, -0.03753662, 0.069...
http://arxiv.org/abs/2212.10325v5
SeqDiffuSeq: Text Diffusion with Encoder-Decoder Transformers
Hongyi Yuan, Zheng Yuan, Chuanqi Tan, Fei Huang, Songfang Huang
cs.CL
Diffusion model, a new generative modelling paradigm, has achieved great success in image, audio, and video generation. However, considering the discrete categorical nature of text, it is not trivial to extend continuous diffusion models to natural language, and text diffusion models are less studied. Sequence-to-seque...
http://arxiv.org/pdf/2212.10325v5
2023-05-22T17:31:46
2022-12-20T15:16:24
{ "title": [ -0.027496338, -0.0068092346, -0.048950195, -0.02848816, 0.007106781, -0.005657196, -0.0099105835, -0.009735107, 0.012512207, 0.038269043, -0.03527832, 0.029922485, 0.022537231, -0.01625061, -0.029586792, 0.020202637, -0.016464233, 0....
http://arxiv.org/abs/2212.10315v2
HINT: Hypernetwork Instruction Tuning for Efficient Zero- & Few-Shot Generalisation
Hamish Ivison, Akshita Bhagia, Yizhong Wang, Hannaneh Hajishirzi, Matthew Peters
cs.CL
Recent NLP models have shown the remarkable ability to effectively generalise `zero-shot' to new tasks using only natural language instructions as guidance. However, many of these approaches suffer from high computational costs due to their reliance on concatenating lengthy instructions with every input example, result...
http://arxiv.org/pdf/2212.10315v2
2023-05-24T22:30:30
2022-12-20T15:07:37
{ "title": [ 0.006225586, -0.04031372, -0.010139465, -0.050628662, -0.04196167, 0.010765076, -0.033477783, 0.03186035, 0.018463135, 0.02319336, -0.024856567, 0.0041007996, -0.0024585724, -0.022735596, -0.015457153, -0.01852417, 0.022247314, 0.015...
http://arxiv.org/abs/2212.10313v2
Beyond Triplet: Leveraging the Most Data for Multimodal Machine Translation
Yaoming Zhu, Zewei Sun, Shanbo Cheng, Luyang Huang, Liwei Wu, Mingxuan Wang
cs.CL
Multimodal machine translation (MMT) aims to improve translation quality by incorporating information from other modalities, such as vision. Previous MMT systems mainly focus on better access and use of visual information and tend to validate their methods on image-related datasets. These studies face two challenges. F...
http://arxiv.org/pdf/2212.10313v2
2023-09-03T03:46:05
2022-12-20T15:02:38
{ "title": [ -0.03640747, -0.013130188, -0.030960083, -0.067871094, -0.011795044, 0.010231018, -0.054595947, 0.01499176, 0.007144928, 0.051116943, -0.0109939575, 0.019302368, 0.019241333, -0.0079574585, -0.05154419, 0.021911621, 0.0009813309, 0.0...
http://arxiv.org/abs/2212.10297v2
Extrinsic Evaluation of Machine Translation Metrics
Nikita Moghe, Tom Sherborne, Mark Steedman, Alexandra Birch
cs.CL, cs.AI
Automatic machine translation (MT) metrics are widely used to distinguish the translation qualities of machine translation systems across relatively large test sets (system-level evaluation). However, it is unclear if automatic metrics are reliable at distinguishing good translations from bad translations at the senten...
http://arxiv.org/pdf/2212.10297v2
2023-06-19T03:59:55
2022-12-20T14:39:58
{ "title": [ -0.06298828, -0.016052246, -0.01600647, -0.015556335, -0.021911621, 0.0037345886, -0.031158447, -0.010749817, -0.0010185242, 0.050048828, -0.013122559, 0.009925842, 0.025375366, -0.0115737915, -0.04864502, 0.013252258, -0.012504578, ...
http://arxiv.org/abs/2212.10264v1
ReCode: Robustness Evaluation of Code Generation Models
Shiqi Wang, Zheng Li, Haifeng Qian, Chenghao Yang, Zijian Wang, Mingyue Shang, Varun Kumar, Samson Tan, Baishakhi Ray, Parminder Bhatia, Ramesh Nallapati, Murali Krishna Ramanathan, Dan Roth, Bing Xiang
cs.LG, cs.CL, cs.SE
Code generation models have achieved impressive performance. However, they tend to be brittle as slight edits to a prompt could lead to very different generations; these robustness properties, critical for user experience when deployed in real-life applications, are not well understood. Most existing works on robustnes...
http://arxiv.org/pdf/2212.10264v1
2022-12-20T14:11:31
2022-12-20T14:11:31
{ "title": [ -0.049621582, -0.009857178, -0.039123535, -0.057525635, 0.008552551, 0.00756073, -0.06359863, -0.010429382, 0.0066490173, 0.019683838, -0.011077881, 0.011100769, -0.0070610046, -0.026550293, -0.03338623, 0.010948181, -0.02407837, 0.0...
http://arxiv.org/abs/2212.10258v2
In and Out-of-Domain Text Adversarial Robustness via Label Smoothing
Yahan Yang, Soham Dan, Dan Roth, Insup Lee
cs.CL, cs.CR, cs.LG
Recently it has been shown that state-of-the-art NLP models are vulnerable to adversarial attacks, where the predictions of a model can be drastically altered by slight modifications to the input (such as synonym substitutions). While several defense techniques have been proposed, and adapted, to the discrete nature of...
http://arxiv.org/pdf/2212.10258v2
2023-07-11T19:33:44
2022-12-20T14:06:50
{ "title": [ -0.0069122314, 0.012237549, -0.02217102, -0.014892578, 0.030975342, -0.01826477, 0.014411926, -0.020950317, -0.02961731, 0.0126571655, 0.0025367737, 0.03338623, -0.009300232, -0.020843506, -0.015296936, -0.015777588, -0.023162842, 0....
http://arxiv.org/abs/2212.10257v1
Original or Translated? On the Use of Parallel Data for Translation Quality Estimation
Baopu Qiu, Liang Ding, Di Wu, Lin Shang, Yibing Zhan, Dacheng Tao
cs.CL
Machine Translation Quality Estimation (QE) is the task of evaluating translation output in the absence of human-written references. Due to the scarcity of human-labeled QE data, previous works attempted to utilize the abundant unlabeled parallel corpora to produce additional training data with pseudo labels. In this p...
http://arxiv.org/pdf/2212.10257v1
2022-12-20T14:06:45
2022-12-20T14:06:45
{ "title": [ -0.06506348, 0.0044898987, -0.014755249, -0.018630981, -0.032714844, 0.012031555, -0.04660034, -0.02722168, -0.008605957, 0.046722412, -0.006839752, 0.012245178, 0.001543045, -0.021011353, -0.06112671, 0.008865356, -0.0014019012, 0.0...
http://arxiv.org/abs/2212.10240v1
Diff-Glat: Diffusion Glancing Transformer for Parallel Sequence to Sequence Learning
Lihua Qian, Mingxuan Wang, Yang Liu, Hao Zhou
cs.CL
For sequence generation, both autoregressive models and non-autoregressive models have been developed in recent years. Autoregressive models can achieve high generation quality, but the sequential decoding scheme causes slow decoding speed. Non-autoregressive models accelerate the inference speed with parallel decoding...
http://arxiv.org/pdf/2212.10240v1
2022-12-20T13:36:25
2022-12-20T13:36:25
{ "title": [ -0.021255493, -0.0022716522, -0.04446411, -0.07476807, 0.008575439, -0.005405426, -0.028442383, 0.024414062, 0.010322571, 0.043060303, -0.033447266, 0.013320923, 0.02468872, -0.01259613, -0.033569336, 0.009590149, 0.011260986, 0.0412...
http://arxiv.org/abs/2212.10233v1
Pre-trained Language Models for Keyphrase Generation: A Thorough Empirical Study
Di Wu, Wasi Uddin Ahmad, Kai-Wei Chang
cs.CL
Neural models that do not rely on pre-training have excelled in the keyphrase generation task with large annotated datasets. Meanwhile, new approaches have incorporated pre-trained language models (PLMs) for their data efficiency. However, there lacks a systematic study of how the two types of approaches compare and ho...
http://arxiv.org/pdf/2212.10233v1
2022-12-20T13:20:21
2022-12-20T13:20:21
{ "title": [ -0.037109375, -0.021224976, -0.021713257, -0.037109375, -0.018569946, 0.011421204, -0.057037354, -0.002866745, -0.0044441223, 0.05722046, 0.00026917458, 0.017944336, 0.02027893, -0.0066375732, -0.052215576, 0.019088745, -0.027374268, ...
http://arxiv.org/abs/2212.10218v2
GanLM: Encoder-Decoder Pre-training with an Auxiliary Discriminator
Jian Yang, Shuming Ma, Li Dong, Shaohan Huang, Haoyang Huang, Yuwei Yin, Dongdong Zhang, Liqun Yang, Furu Wei, Zhoujun Li
cs.CL
Pre-trained models have achieved remarkable success in natural language processing (NLP). However, existing pre-training methods underutilize the benefits of language understanding for generation. Inspired by the idea of Generative Adversarial Networks (GANs), we propose a GAN-style model for encoder-decoder pre-traini...
http://arxiv.org/pdf/2212.10218v2
2023-05-09T09:08:33
2022-12-20T12:51:11
{ "title": [ -0.023208618, 0.01210022, -0.06100464, -0.050079346, -0.009429932, -0.016281128, -0.03479004, -0.015716553, -0.015098572, 0.04284668, 0.019226074, -0.0045547485, 0.03363037, -0.048339844, -0.017623901, 0.014274597, -0.0115356445, 0.0...
http://arxiv.org/abs/2212.10197v1
EIT: Enhanced Interactive Transformer
Tong Zheng, Bei Li, Huiwen Bao, Tong Xiao, Jingbo Zhu
cs.CL
In this paper, we propose a novel architecture, the Enhanced Interactive Transformer (EIT), to address the issue of head degradation in self-attention mechanisms. Our approach replaces the traditional multi-head self-attention mechanism with the Enhanced Multi-Head Attention (EMHA) mechanism, which relaxes the one-to-o...
http://arxiv.org/pdf/2212.10197v1
2022-12-20T12:16:46
2022-12-20T12:16:46
{ "title": [ 0.014373779, 0.016555786, -0.032196045, -0.03640747, 0.0053138733, -0.019973755, -0.019760132, -0.056854248, -0.005542755, 0.028778076, -0.0022125244, -0.015052795, 0.050201416, -0.057250977, 0.00033855438, 0.024551392, 0.016677856, ...
http://arxiv.org/abs/2212.10192v1
Adam: Dense Retrieval Distillation with Adaptive Dark Examples
Chang Liu, Chongyang Tao, Xiubo Geng, Tao Shen, Dongyan Zhao, Can Xu, Binxing Jiao, Daxin Jiang
cs.CL
To improve the performance of the dual-encoder retriever, one effective approach is knowledge distillation from the cross-encoder ranker. Existing works construct the candidate passages following the supervised learning setting where a query is paired with a positive passage and a batch of negatives. However, through e...
http://arxiv.org/pdf/2212.10192v1
2022-12-20T12:03:19
2022-12-20T12:03:19
{ "title": [ 0.026763916, 0.012931824, -0.043884277, -0.08544922, 0.036590576, -0.008010864, -0.033355713, -0.02444458, 0.05493164, 0.030014038, -0.026107788, 0.013313293, 0.007282257, -0.010131836, -0.03845215, 0.030197144, 0.0044937134, 0.00897...
http://arxiv.org/abs/2212.10191v1
Emotion Selectable End-to-End Text-based Speech Editing
Tao Wang, Jiangyan Yi, Ruibo Fu, Jianhua Tao, Zhengqi Wen, Chu Yuan Zhang
cs.SD, cs.CL, cs.LG, eess.AS
Text-based speech editing allows users to edit speech by intuitively cutting, copying, and pasting text to speed up the process of editing speech. In the previous work, CampNet (context-aware mask prediction network) is proposed to realize text-based speech editing, significantly improving the quality of edited speech....
http://arxiv.org/pdf/2212.10191v1
2022-12-20T12:02:40
2022-12-20T12:02:40
{ "title": [ 0.007522583, 0.0032577515, -0.028381348, -0.04336548, -0.015991211, -0.015930176, -0.029815674, 0.004436493, 0.011695862, 0.051116943, -0.0009784698, -0.011978149, 0.05142212, -0.050689697, -0.02520752, 0.017990112, -0.0036296844, 0....
http://arxiv.org/abs/2212.10190v1
Pay Attention to Your Tone: Introducing a New Dataset for Polite Language Rewrite
Xun Wang, Tao Ge, Allen Mao, Yuki Li, Furu Wei, Si-Qing Chen
cs.CL
We introduce \textsc{PoliteRewrite} -- a dataset for polite language rewrite which is a novel sentence rewrite task. Compared with previous text style transfer tasks that can be mostly addressed by slight token- or phrase-level edits, polite language rewrite requires deep understanding and extensive sentence-level edit...
http://arxiv.org/pdf/2212.10190v1
2022-12-20T12:02:34
2022-12-20T12:02:34
{ "title": [ -0.015464783, -0.01360321, -0.018615723, -0.0715332, -0.009338379, 0.0030345917, -0.0637207, 0.0026550293, -0.020736694, 0.022872925, -0.015586853, 0.0023212433, 0.018112183, -0.023635864, -0.027038574, 0.0026569366, -0.010391235, 0....
http://arxiv.org/abs/2212.10189v2
Do I have the Knowledge to Answer? Investigating Answerability of Knowledge Base Questions
Mayur Patidar, Prayushi Faldu, Avinash Singh, Lovekesh Vig, Indrajit Bhattacharya, Mausam
cs.CL, cs.AI
When answering natural language questions over knowledge bases, missing facts, incomplete schema and limited scope naturally lead to many questions being unanswerable. While answerability has been explored in other QA settings, it has not been studied for QA over knowledge bases (KBQA). We create GrailQAbility, a new b...
http://arxiv.org/pdf/2212.10189v2
2023-06-24T11:06:55
2022-12-20T12:00:26
{ "title": [ -0.016555786, 0.010169983, -0.04284668, -0.041259766, 0.018096924, -0.011299133, -0.0044517517, -0.027709961, 0.03378296, -0.013923645, -0.0496521, 0.008010864, -0.045135498, -0.0000042915344, -0.05441284, -0.021347046, 0.0073165894, ...
http://arxiv.org/abs/2301.05150v1
Unsupervised Question Duplicate and Related Questions Detection in e-learning platforms
Maksimjeet Chowdhary, Sanyam Goyal, Venktesh V, Mukesh Mohania, Vikram Goyal
cs.CL
Online learning platforms provide diverse questions to gauge the learners' understanding of different concepts. The repository of questions has to be constantly updated to ensure a diverse pool of questions to conduct assessments for learners. However, it is impossible for the academician to manually skim through the l...
http://arxiv.org/pdf/2301.05150v1
2022-12-20T11:52:52
2022-12-20T11:52:52
{ "title": [ 0.013343811, 0.033355713, -0.029556274, -0.028182983, 0.029678345, -0.018417358, 0.012878418, -0.03793335, -0.005634308, 0.04019165, -0.038970947, 0.02859497, 0.011497498, -0.002538681, -0.02571106, -0.011734009, -0.020187378, 0.0408...
http://arxiv.org/abs/2212.10180v2
IndicMT Eval: A Dataset to Meta-Evaluate Machine Translation metrics for Indian Languages
Ananya B. Sai, Vignesh Nagarajan, Tanay Dixit, Raj Dabre, Anoop Kunchukuttan, Pratyush Kumar, Mitesh M. Khapra
cs.CL
The rapid growth of machine translation (MT) systems has necessitated comprehensive studies to meta-evaluate evaluation metrics being used, which enables a better selection of metrics that best reflect MT quality. Unfortunately, most of the research focuses on high-resource languages, mainly English, the observations f...
http://arxiv.org/pdf/2212.10180v2
2023-07-03T14:26:38
2022-12-20T11:37:22
{ "title": [ -0.060394287, -0.006137848, -0.02960205, -0.029251099, -0.020065308, 0.01210022, -0.064697266, -0.013168335, -0.036315918, 0.0413208, -0.0031642914, 0.015777588, 0.019439697, -0.029205322, -0.027389526, 0.0022697449, -0.0112838745, 0...
http://arxiv.org/abs/2212.10179v1
Toward Human-Like Evaluation for Natural Language Generation with Error Analysis
Qingyu Lu, Liang Ding, Liping Xie, Kanjian Zhang, Derek F. Wong, Dacheng Tao
cs.CL
The state-of-the-art language model-based automatic metrics, e.g. BARTScore, benefiting from large-scale contextualized pre-training, have been successfully used in a wide range of natural language generation (NLG) tasks, including machine translation, text summarization, and data-to-text. Recent studies show that cons...
http://arxiv.org/pdf/2212.10179v1
2022-12-20T11:36:22
2022-12-20T11:36:22
{ "title": [ -0.037475586, -0.01209259, -0.043182373, -0.042236328, 0.000045120716, 0.0027675629, -0.0368042, 0.0034160614, -0.0013771057, 0.027069092, -0.033447266, 0.009628296, -0.0018415451, 0.0013198853, -0.05831909, -0.005050659, -0.034820557, ...
http://arxiv.org/abs/2212.10173v1
On the Role of Parallel Data in Cross-lingual Transfer Learning
Machel Reid, Mikel Artetxe
cs.CL
While prior work has established that the use of parallel data is conducive for cross-lingual learning, it is unclear if the improvements come from the data itself, or if it is the modeling of parallel interactions that matters. Exploring this, we examine the usage of unsupervised machine translation to generate synthe...
http://arxiv.org/pdf/2212.10173v1
2022-12-20T11:23:04
2022-12-20T11:23:04
{ "title": [ -0.0211792, -0.00010287762, -0.045288086, -0.04699707, 0.0017223358, -0.0077285767, -0.029586792, -0.011917114, -0.010757446, 0.04144287, -0.013160706, 0.021987915, 0.006603241, -0.014923096, -0.031280518, 0.010986328, -0.0141067505, ...
http://arxiv.org/abs/2212.10171v1
Document-level Relation Extraction with Relation Correlations
Ridong Han, Tao Peng, Benyou Wang, Lu Liu, Xiang Wan
cs.CL
Document-level relation extraction faces two overlooked challenges: long-tail problem and multi-label problem. Previous work focuses mainly on obtaining better contextual representations for entity pairs, hardly address the above challenges. In this paper, we analyze the co-occurrence correlation of relations, and intr...
http://arxiv.org/pdf/2212.10171v1
2022-12-20T11:17:52
2022-12-20T11:17:52
{ "title": [ -0.011672974, 0.000006020069, -0.052368164, -0.083618164, 0.020751953, 0.005531311, -0.050598145, 0.0043678284, 0.03164673, 0.01474762, -0.01876831, 0.004096985, -0.010368347, -0.0014152527, -0.04928589, -0.000108897686, -0.013839722, ...
http://arxiv.org/abs/2212.10168v2
Naamapadam: A Large-Scale Named Entity Annotated Data for Indic Languages
Arnav Mhaske, Harshit Kedia, Sumanth Doddapaneni, Mitesh M. Khapra, Pratyush Kumar, Rudra Murthy V, Anoop Kunchukuttan
cs.CL
We present, Naamapadam, the largest publicly available Named Entity Recognition (NER) dataset for the 11 major Indian languages from two language families. The dataset contains more than 400k sentences annotated with a total of at least 100k entities from three standard entity categories (Person, Location, and, Organiz...
http://arxiv.org/pdf/2212.10168v2
2023-05-28T06:26:45
2022-12-20T11:15:24
{ "title": [ -0.018753052, -0.014640808, -0.04928589, -0.0791626, 0.01550293, -0.0014724731, -0.03579712, 0.030883789, -0.005470276, 0.020645142, -0.018554688, 0.029708862, 0.014343262, -0.015533447, -0.012367249, 0.014015198, -0.00340271, 0.0301...
http://arxiv.org/abs/2212.10154v2
Human-Guided Fair Classification for Natural Language Processing
Florian E. Dorner, Momchil Peychev, Nikola Konstantinov, Naman Goel, Elliott Ash, Martin Vechev
cs.CL, cs.AI, cs.CY, cs.LG
Text classifiers have promising applications in high-stake tasks such as resume screening and content moderation. These classifiers must be fair and avoid discriminatory decisions by being invariant to perturbations of sensitive attributes such as gender or ethnicity. However, there is a gap between human intuition abo...
http://arxiv.org/pdf/2212.10154v2
2023-03-16T08:57:14
2022-12-20T10:46:40
{ "title": [ 0.000037252903, 0.00038981438, -0.029891968, -0.057739258, 0.011260986, -0.010505676, -0.02798462, -0.006149292, 0.014778137, 0.030822754, -0.04437256, -0.00018560886, 0.012580872, -0.00459671, -0.055725098, -0.004673004, -0.027053833, ...
http://arxiv.org/abs/2212.10152v1
Quirk or Palmer: A Comparative Study of Modal Verb Frameworks with Annotated Datasets
Risako Owan, Maria Gini, Dongyeop Kang
cs.CL
Modal verbs, such as "can", "may", and "must", are commonly used in daily communication to convey the speaker's perspective related to the likelihood and/or mode of the proposition. They can differ greatly in meaning depending on how they're used and the context of a sentence (e.g. "They 'must' help each other out." vs...
http://arxiv.org/pdf/2212.10152v1
2022-12-20T10:44:18
2022-12-20T10:44:18
{ "title": [ -0.014564514, 0.00031638145, -0.037261963, -0.052947998, 0.0037002563, -0.008026123, -0.055725098, -0.00687027, -0.0026931763, 0.06036377, -0.0021400452, 0.022506714, 0.031982422, -0.019012451, -0.04586792, 0.028411865, -0.0138168335, ...
http://arxiv.org/abs/2212.10140v2
Tackling Ambiguity with Images: Improved Multimodal Machine Translation and Contrastive Evaluation
Matthieu Futeral, Cordelia Schmid, Ivan Laptev, Benoît Sagot, Rachel Bawden
cs.CL, cs.CV
One of the major challenges of machine translation (MT) is ambiguity, which can in some cases be resolved by accompanying context such as images. However, recent work in multimodal MT (MMT) has shown that obtaining improvements from images is challenging, limited not only by the difficulty of building effective cross-m...
http://arxiv.org/pdf/2212.10140v2
2023-05-26T10:52:39
2022-12-20T10:18:18
{ "title": [ -0.014328003, 0.013122559, -0.003358841, -0.027313232, 0.0008330345, -0.0021190643, -0.019622803, -0.039276123, 0.021652222, 0.03652954, -0.00522995, 0.0029182434, 0.0022411346, -0.025024414, -0.05783081, 0.009384155, -0.0067100525, ...
http://arxiv.org/abs/2212.10114v2
True Detective: A Deep Abductive Reasoning Benchmark Undoable for GPT-3 and Challenging for GPT-4
Maksym Del, Mark Fishel
cs.CL
Large language models (LLMs) have demonstrated solid zero-shot reasoning capabilities, which is reflected in their performance on the current test tasks. This calls for a more challenging benchmark requiring highly advanced reasoning ability to be solved. In this paper, we introduce such a benchmark, consisting of 191 ...
http://arxiv.org/pdf/2212.10114v2
2023-06-01T18:50:21
2022-12-20T09:34:43
{ "title": [ 0.030654907, 0.01436615, 0.014579773, 0.029144287, -0.022109985, -0.024490356, 0.022827148, -0.055114746, 0.0070610046, 0.015686035, -0.03201294, -0.013084412, 0.022781372, -0.038970947, -0.0068511963, -0.012527466, -0.0022602081, 0....
http://arxiv.org/abs/2212.10097v1
Toward a Unified Framework for Unsupervised Complex Tabular Reasoning
Zhenyu Li, Xiuxing Li, Zhichao Duan, Bowen Dong, Ning Liu, Jianyong Wang
cs.CL, cs.AI, cs.DB
Structured tabular data exist across nearly all fields. Reasoning task over these data aims to answer questions or determine the truthiness of hypothesis sentences by understanding the semantic meaning of a table. While previous works have devoted significant efforts to the tabular reasoning task, they always assume th...
http://arxiv.org/pdf/2212.10097v1
2022-12-20T09:15:03
2022-12-20T09:15:03
{ "title": [ -0.009468079, -0.00472641, -0.037963867, -0.08068848, 0.010879517, 0.0154953, -0.06329346, 0.005935669, 0.01586914, 0.030578613, -0.04071045, 0.013160706, 0.0040016174, 0.0016231537, -0.033050537, 0.0031776428, -0.0007004738, 0.01115...
http://arxiv.org/abs/2212.10087v1
Hybrid Rule-Neural Coreference Resolution System based on Actor-Critic Learning
Yu Wang, Hongxia Jin
cs.CL
A coreference resolution system is to cluster all mentions that refer to the same entity in a given context. All coreference resolution systems need to tackle two main tasks: one task is to detect all of the potential mentions, and the other is to learn the linking of an antecedent for each possible mention. In this pa...
http://arxiv.org/pdf/2212.10087v1
2022-12-20T08:55:47
2022-12-20T08:55:47
{ "title": [ 0.0039978027, -0.004131317, -0.021026611, -0.05130005, -0.0037384033, -0.008041382, -0.023544312, 0.017105103, -0.044891357, 0.044128418, 0.00018882751, 0.0063095093, -0.0010595322, -0.022125244, -0.029510498, -0.006362915, -0.0038852692...
http://arxiv.org/abs/2212.10080v1
Rumour detection using graph neural network and oversampling in benchmark Twitter dataset
Shaswat Patel, Prince Bansal, Preeti Kaur
cs.CL, cs.LG
Recently, online social media has become a primary source for new information and misinformation or rumours. In the absence of an automatic rumour detection system the propagation of rumours has increased manifold leading to serious societal damages. In this work, we propose a novel method for building automatic rumour...
http://arxiv.org/pdf/2212.10080v1
2022-12-20T08:43:10
2022-12-20T08:43:10
{ "title": [ 0.0073890686, 0.0020389557, -0.01727295, -0.015029907, -0.012390137, -0.012298584, -0.013641357, -0.035186768, -0.00844574, 0.031982422, -0.032226562, 0.0036468506, 0.00060510635, -0.016693115, -0.03918457, -0.016555786, -0.025024414, ...
http://arxiv.org/abs/2212.10079v1
A Survey on Pretrained Language Models for Neural Code Intelligence
Yichen Xu, Yanqiao Zhu
cs.SE, cs.CL, cs.LG
As the complexity of modern software continues to escalate, software engineering has become an increasingly daunting and error-prone endeavor. In recent years, the field of Neural Code Intelligence (NCI) has emerged as a promising solution, leveraging the power of deep learning techniques to tackle analytical tasks on ...
http://arxiv.org/pdf/2212.10079v1
2022-12-20T08:34:56
2022-12-20T08:34:56
{ "title": [ -0.028121948, -0.022842407, -0.015472412, -0.04058838, -0.015136719, 0.0033779144, -0.04055786, 0.0057525635, -0.0025291443, 0.045776367, -0.0015296936, -0.0076789856, 0.019104004, -0.021484375, -0.05038452, 0.018218994, -0.006298065, ...
http://arxiv.org/abs/2212.10077v3
DOC: Improving Long Story Coherence With Detailed Outline Control
Kevin Yang, Dan Klein, Nanyun Peng, Yuandong Tian
cs.CL, cs.AI
We propose the Detailed Outline Control (DOC) framework for improving long-range plot coherence when automatically generating several-thousand-word-long stories. DOC consists of two complementary components: a detailed outliner and a detailed controller. The detailed outliner creates a more detailed, hierarchically str...
http://arxiv.org/pdf/2212.10077v3
2023-06-14T22:05:11
2022-12-20T08:30:58
{ "title": [ -0.007369995, -0.011444092, -0.039916992, -0.047943115, -0.023910522, -0.01335144, -0.04937744, -0.0104522705, -0.0513916, 0.028625488, 0.011047363, 0.011161804, 0.026687622, -0.04724121, 0.0181427, -0.007068634, -0.020843506, 0.0210...
http://arxiv.org/abs/2212.10071v2
Large Language Models Are Reasoning Teachers
Namgyu Ho, Laura Schmid, Se-Young Yun
cs.CL, cs.AI, cs.LG
Recent works have shown that chain-of-thought (CoT) prompting can elicit language models to solve complex reasoning tasks, step-by-step. However, prompt-based CoT methods are dependent on very large models such as GPT-3 175B which are prohibitive to deploy at scale. In this paper, we use these large models as reasoning...
http://arxiv.org/pdf/2212.10071v2
2023-06-13T10:55:29
2022-12-20T08:24:45
{ "title": [ 0.010101318, -0.026519775, -0.017501831, -0.07940674, -0.05859375, 0.00040960312, -0.07281494, 0.024154663, -0.02406311, 0.020492554, -0.036743164, -0.037261963, 0.022827148, -0.059265137, 0.032714844, -0.027557373, 0.012245178, 0.03...
http://arxiv.org/abs/2212.10060v2
I Cast Detect Thoughts: Learning to Converse and Guide with Intents and Theory-of-Mind in Dungeons and Dragons
Pei Zhou, Andrew Zhu, Jennifer Hu, Jay Pujara, Xiang Ren, Chris Callison-Burch, Yejin Choi, Prithviraj Ammanabrolu
cs.CL, cs.AI
We propose a novel task, G4C, to study teacher-student natural language interactions in a goal-driven and grounded environment. Dungeons and Dragons (D&D), a role-playing game, provides an ideal setting to investigate such interactions. Here, the Dungeon Master (DM), i.e., the teacher, guides the actions of several pla...
http://arxiv.org/pdf/2212.10060v2
2023-05-30T23:09:37
2022-12-20T08:06:55
{ "title": [ 0.013786316, -0.019348145, -0.030929565, 0.03753662, -0.030807495, -0.03213501, 0.0368042, -0.029846191, 0.01512146, 0.010345459, -0.0053863525, -0.0056648254, 0.047668457, -0.047668457, -0.00025558472, -0.014892578, -0.0057678223, -...
http://arxiv.org/abs/2212.10057v2
WeCheck: Strong Factual Consistency Checker via Weakly Supervised Learning
Wenhao Wu, Wei Li, Xinyan Xiao, Jiachen Liu, Sujian Li, Yajuan Lv
cs.CL
A crucial issue of current text generation models is that they often uncontrollably generate factually inconsistent text with respective of their inputs. Limited by the lack of annotated data, existing works in evaluating factual consistency directly transfer the reasoning ability of models trained on other data-rich u...
http://arxiv.org/pdf/2212.10057v2
2023-05-27T13:42:23
2022-12-20T08:04:36
{ "title": [ -0.014930725, 0.013572693, -0.0680542, -0.07476807, 0.027313232, -0.003534317, -0.047973633, -0.0077819824, -0.013542175, 0.030975342, -0.008491516, -0.000049948692, 0.02192688, -0.016677856, -0.036071777, -0.0032138824, -0.01890564, ...
http://arxiv.org/abs/2212.10051v1
A Framework of Customer Review Analysis Using the Aspect-Based Opinion Mining Approach
Subhasis Dasgupta, Jaydip Sen
cs.CL, cs.LG
Opinion mining is the branch of computation that deals with opinions, appraisals, attitudes, and emotions of people and their different aspects. This field has attracted substantial research interest in recent years. Aspect-level (called aspect-based opinion mining) is often desired in practical applications as it prov...
http://arxiv.org/pdf/2212.10051v1
2022-12-20T07:54:58
2022-12-20T07:54:58
{ "title": [ 0.0070495605, -0.021270752, -0.010322571, -0.036834717, -0.0024204254, 0.0028877258, -0.010734558, -0.008621216, -0.0022335052, 0.04928589, -0.046569824, -0.0045433044, -0.005290985, 0.010414124, -0.08654785, -0.024520874, -0.009559631, ...
http://arxiv.org/abs/2212.10047v2
An Augmentation Strategy for Visually Rich Documents
Jing Xie, James B. Wendt, Yichao Zhou, Seth Ebner, Sandeep Tata
cs.CL
Many business workflows require extracting important fields from form-like documents (e.g. bank statements, bills of lading, purchase orders, etc.). Recent techniques for automating this task work well only when trained with large datasets. In this work we propose a novel data augmentation technique to improve performa...
http://arxiv.org/pdf/2212.10047v2
2022-12-22T09:13:04
2022-12-20T07:44:25
{ "title": [ 0.0021190643, -0.008110046, -0.018035889, -0.053527832, 0.004047394, -0.007068634, -0.01826477, -0.00041031837, 0.014625549, 0.0423584, -0.034942627, 0.018539429, 0.011505127, 0.00041508675, -0.039886475, 0.012336731, 0.011314392, 0....
http://arxiv.org/abs/2212.10039v1
A Twitter BERT Approach for Offensive Language Detection in Marathi
Tanmay Chavan, Shantanu Patankar, Aditya Kane, Omkar Gokhale, Raviraj Joshi
cs.CL
Automated offensive language detection is essential in combating the spread of hate speech, particularly in social media. This paper describes our work on Offensive Language Identification in low resource Indic language Marathi. The problem is formulated as a text classification task to identify a tweet as offensive or...
http://arxiv.org/pdf/2212.10039v1
2022-12-20T07:22:45
2022-12-20T07:22:45
{ "title": [ 0.013183594, 0.006629944, -0.020050049, -0.06896973, 0.01663208, 0.0016889572, -0.029388428, 0.005306244, 0.0054130554, 0.03552246, -0.050354004, -0.0006375313, 0.0317688, -0.0028266907, -0.058288574, -0.014480591, -0.022705078, 0.05...
http://arxiv.org/abs/2212.10029v3
Do language models have coherent mental models of everyday things?
Yuling Gu, Bhavana Dalvi Mishra, Peter Clark
cs.CL, cs.AI
When people think of everyday things like an egg, they typically have a mental image associated with it. This allows them to correctly judge, for example, that "the yolk surrounds the shell" is a false statement. Do language models similarly have a coherent picture of such everyday things? To investigate this, we propo...
http://arxiv.org/pdf/2212.10029v3
2023-06-08T17:27:44
2022-12-20T06:54:04
{ "title": [ -0.031921387, 0.0060653687, -0.052215576, -0.036010742, -0.031921387, 0.010566711, -0.08746338, -0.0013504028, 0.0048446655, 0.033599854, -0.015670776, -0.03201294, 0.048675537, -0.04849243, -0.011230469, 0.0012340546, 0.005168915, 0...
http://arxiv.org/abs/2212.10025v2
When Federated Learning Meets Pre-trained Language Models' Parameter-Efficient Tuning Methods
Zhuo Zhang, Yuanhang Yang, Yong Dai, Lizhen Qu, Zenglin Xu
cs.LG, cs.CL
With increasing privacy concerns on data, recent studies have made significant progress using federated learning (FL) on privacy-sensitive natural language processing (NLP) tasks. Much literature suggests fully fine-tuning pre-trained language models (PLMs) in the FL paradigm can mitigate the data heterogeneity problem...
http://arxiv.org/pdf/2212.10025v2
2023-06-02T12:58:07
2022-12-20T06:44:32
{ "title": [ -0.008865356, 0.011657715, -0.03173828, -0.093444824, 0.007774353, 0.004371643, -0.055877686, 0.010643005, -0.03475952, 0.04095459, -0.025131226, 0.016143799, -0.010261536, -0.0030097961, -0.025100708, -0.014198303, -0.0015077591, 0....
http://arxiv.org/abs/2212.10020v3
On the Blind Spots of Model-Based Evaluation Metrics for Text Generation
Tianxing He, Jingyu Zhang, Tianle Wang, Sachin Kumar, Kyunghyun Cho, James Glass, Yulia Tsvetkov
cs.CL
In this work, we explore a useful but often neglected methodology for robustness analysis of text generation evaluation metrics: stress tests with synthetic data. Basically, we design and synthesize a wide range of potential errors and check whether they result in a commensurate drop in the metric scores. We examine a ...
http://arxiv.org/pdf/2212.10020v3
2023-05-18T20:46:57
2022-12-20T06:24:25
{ "title": [ -0.046295166, -0.013038635, -0.022857666, -0.011932373, -0.01436615, 0.008514404, -0.04055786, -0.035827637, 0.0051879883, 0.03378296, -0.016174316, 0.007293701, -0.003917694, -0.007801056, -0.06222534, -0.0054740906, -0.03479004, 0....
http://arxiv.org/abs/2212.10019v1
When Do Decompositions Help for Machine Reading?
Kangda Wei, Dawn Lawrie, Benjamin Van Durme, Yunmo Chen, Orion Weller
cs.CL
Answering complex questions often requires multi-step reasoning in order to obtain the final answer. Most research into decompositions of complex questions involves open-domain systems, which have shown success in using these decompositions for improved retrieval. In the machine reading setting, however, work to unders...
http://arxiv.org/pdf/2212.10019v1
2022-12-20T06:23:02
2022-12-20T06:23:02
{ "title": [ -0.03567505, -0.013587952, -0.008476257, -0.020233154, -0.04736328, 0.019927979, -0.0769043, -0.018051147, 0.012313843, 0.081726074, -0.0040397644, -0.015853882, 0.076049805, -0.019821167, -0.031951904, 0.028915405, -0.0046463013, 0....
http://arxiv.org/abs/2212.10018v2
DIONYSUS: A Pre-trained Model for Low-Resource Dialogue Summarization
Yu Li, Baolin Peng, Pengcheng He, Michel Galley, Zhou Yu, Jianfeng Gao
cs.CL
Dialogue summarization has recently garnered significant attention due to its wide range of applications. However, existing methods for summarizing dialogues have limitations because they do not take into account the inherent structure of dialogue and rely heavily on labeled data, which can lead to poor performance in ...
http://arxiv.org/pdf/2212.10018v2
2023-05-26T17:29:01
2022-12-20T06:21:21
{ "title": [ 0.010108948, -0.009857178, -0.024932861, -0.09625244, -0.006477356, 0.006252289, -0.070739746, 0.012237549, 0.006298065, 0.051086426, 0.0024433136, 0.0032463074, -0.004962921, -0.007335663, -0.075927734, 0.009262085, -0.008270264, 0....
http://arxiv.org/abs/2212.10015v3
Benchmarking Spatial Relationships in Text-to-Image Generation
Tejas Gokhale, Hamid Palangi, Besmira Nushi, Vibhav Vineet, Eric Horvitz, Ece Kamar, Chitta Baral, Yezhou Yang
cs.CV, cs.AI, cs.CL
Spatial understanding is a fundamental aspect of computer vision and integral for human-level reasoning about images, making it an important component for grounded language understanding. While recent text-to-image synthesis (T2I) models have shown unprecedented improvements in photorealism, it is unclear whether they ...
http://arxiv.org/pdf/2212.10015v3
2023-10-27T17:24:04
2022-12-20T06:03:51
{ "title": [ -0.0054855347, -0.0032691956, -0.0066070557, -0.04272461, -0.010414124, 0.007369995, -0.03527832, -0.034820557, 0.024551392, 0.055664062, -0.020599365, 0.011451721, -0.00079107285, -0.002866745, -0.06591797, 0.0024375916, 0.006668091, ...
http://arxiv.org/abs/2212.10013v1
DocAsRef: A Pilot Empirical Study on Repurposing Reference-Based Summary Quality Metrics Reference-Freely
Forrest Sheng Bao, Ruixuan Tu, Ge Luo
cs.AI, cs.CL
Summary quality assessment metrics have two categories: reference-based and reference-free. Reference-based metrics are theoretically more accurate but are limited by the availability and quality of the human-written references, which are both difficulty to ensure. This inspires the development of reference-free metric...
http://arxiv.org/pdf/2212.10013v1
2022-12-20T06:01:13
2022-12-20T06:01:13
{ "title": [ -0.02760315, 0.0026073456, -0.04348755, -0.034942627, -0.027770996, -0.014785767, -0.061187744, -0.02772522, -0.008300781, 0.06524658, -0.027069092, 0.013267517, 0.025222778, -0.015205383, -0.042114258, 0.025238037, 0.0029144287, 0.0...
http://arxiv.org/abs/2212.10012v1
Language Modeling with Latent Situations
Belinda Z. Li, Maxwell Nye, Jacob Andreas
cs.CL
Language models (LMs) often generate incoherent outputs: they refer to events and entity states that are incompatible with the state of the world described in their inputs. We introduce SituationSupervision, a family of approaches for improving coherence in LMs by training them to construct and condition on explicit re...
http://arxiv.org/pdf/2212.10012v1
2022-12-20T05:59:42
2022-12-20T05:59:42
{ "title": [ -0.013931274, 0.00081539154, -0.030807495, -0.06048584, 0.0059890747, 0.004283905, -0.043304443, -0.003604889, 0.022232056, 0.025787354, -0.027801514, 0.008728027, 0.0020599365, -0.012901306, -0.040405273, 0.017807007, -0.017913818, ...
http://arxiv.org/abs/2212.10011v2
PLUE: Language Understanding Evaluation Benchmark for Privacy Policies in English
Jianfeng Chi, Wasi Uddin Ahmad, Yuan Tian, Kai-Wei Chang
cs.CL
Privacy policies provide individuals with information about their rights and how their personal information is handled. Natural language understanding (NLU) technologies can support individuals and practitioners to understand better privacy practices described in lengthy and complex documents. However, existing efforts...
http://arxiv.org/pdf/2212.10011v2
2023-05-12T07:38:29
2022-12-20T05:58:32
{ "title": [ -0.014915466, -0.021133423, -0.029403687, -0.049438477, -0.010147095, -0.012550354, -0.03201294, 0.0008778572, -0.0368042, 0.0062179565, -0.015777588, 0.013687134, -0.0016441345, -0.034332275, 0.0028705597, -0.013763428, -0.02128601, ...
http://arxiv.org/abs/2212.10008v2
Enhancing Task Bot Engagement with Synthesized Open-Domain Dialog
Miaoran Li, Baolin Peng, Michel Galley, Jianfeng Gao, Zhu Zhang
cs.CL, cs.AI
Many efforts have been made to construct dialog systems for different types of conversations, such as task-oriented dialog (TOD) and open-domain dialog (ODD). To better mimic human-level conversations that usually fuse various dialog modes, it is essential to build a system that can effectively handle both TOD and ODD ...
http://arxiv.org/pdf/2212.10008v2
2023-07-30T16:04:23
2022-12-20T05:51:47
{ "title": [ 0.0309906, 0.010070801, -0.028961182, -0.064941406, 0.013168335, -0.017425537, -0.010955811, -0.010139465, -0.023895264, 0.048461914, 0.01864624, 0.012664795, -0.00466156, -0.019454956, -0.042144775, -0.0022621155, -0.018600464, 0.02...
http://arxiv.org/abs/2212.10007v2
CoCoMIC: Code Completion By Jointly Modeling In-file and Cross-file Context
Yangruibo Ding, Zijian Wang, Wasi Uddin Ahmad, Murali Krishna Ramanathan, Ramesh Nallapati, Parminder Bhatia, Dan Roth, Bing Xiang
cs.CL, cs.SE
While pre-trained language models (LM) for code have achieved great success in code completion, they generate code conditioned only on the contents within the file, i.e., in-file context, but ignore the rich semantics in other files within the same project, i.e., cross-file context, a critical source of information tha...
http://arxiv.org/pdf/2212.10007v2
2023-05-24T06:56:45
2022-12-20T05:48:09
{ "title": [ -0.024520874, -0.01965332, -0.056427002, -0.091918945, -0.004760742, 0.009391785, -0.06500244, 0.022369385, -0.000013589859, 0.046295166, -0.05795288, 0.019683838, 0.031829834, -0.006134033, -0.026763916, -0.0040779114, -0.027130127, ...
http://arxiv.org/abs/2212.10003v2
(QA)$^2$: Question Answering with Questionable Assumptions
Najoung Kim, Phu Mon Htut, Samuel R. Bowman, Jackson Petty
cs.CL
Naturally occurring information-seeking questions often contain questionable assumptions -- assumptions that are false or unverifiable. Questions containing questionable assumptions are challenging because they require a distinct answer strategy that deviates from typical answers for information-seeking questions. For ...
http://arxiv.org/pdf/2212.10003v2
2023-08-29T19:36:32
2022-12-20T05:25:12
{ "title": [ -0.035064697, -0.007713318, -0.04135132, -0.04055786, 0.0034294128, -0.0034542084, -0.013336182, -0.015014648, 0.0049209595, 0.00049972534, -0.02017212, -0.011604309, -0.028762817, -0.029922485, -0.038757324, -0.010765076, -0.007381439, ...
http://arxiv.org/abs/2212.10002v2
Defending Against Misinformation Attacks in Open-Domain Question Answering
Orion Weller, Aleem Khan, Nathaniel Weir, Dawn Lawrie, Benjamin Van Durme
cs.CL, cs.IR
Recent work in open-domain question answering (ODQA) has shown that adversarial poisoning of the search collection can cause large drops in accuracy for production systems. However, little to no work has proposed methods to defend against these attacks. To do so, we rely on the intuition that redundant information ofte...
http://arxiv.org/pdf/2212.10002v2
2023-05-15T17:46:20
2022-12-20T05:25:01
{ "title": [ 0.023345947, 0.007881165, -0.033691406, -0.059539795, 0.026992798, -0.015701294, 0.009132385, -0.0036563873, -0.04244995, 0.025360107, 0.013519287, 0.0129852295, -0.00881958, -0.022720337, -0.033081055, -0.015602112, -0.023529053, 0....
http://arxiv.org/abs/2212.10001v2
Towards Understanding Chain-of-Thought Prompting: An Empirical Study of What Matters
Boshi Wang, Sewon Min, Xiang Deng, Jiaming Shen, You Wu, Luke Zettlemoyer, Huan Sun
cs.CL
Chain-of-Thought (CoT) prompting can dramatically improve the multi-step reasoning abilities of large language models (LLMs). CoT explicitly encourages the LLM to generate intermediate rationales for solving a problem, by providing a series of reasoning steps in the demonstrations. Despite its success, there is still l...
http://arxiv.org/pdf/2212.10001v2
2023-06-01T05:38:00
2022-12-20T05:20:54
{ "title": [ 0.018127441, -0.015083313, -0.005519867, -0.030258179, -0.0012197495, -0.022994995, -0.0008869171, -0.04272461, -0.024307251, 0.0039100647, -0.034362793, 0.0146865845, -0.016723633, -0.027023315, -0.026184082, -0.0178833, -0.006286621, ...
http://arxiv.org/abs/2212.09994v1
Towards Robustness of Text-to-SQL Models Against Natural and Realistic Adversarial Table Perturbation
Xinyu Pi, Bing Wang, Yan Gao, Jiaqi Guo, Zhoujun Li, Jian-Guang Lou
cs.CL
The robustness of Text-to-SQL parsers against adversarial perturbations plays a crucial role in delivering highly reliable applications. Previous studies along this line primarily focused on perturbations in the natural language question side, neglecting the variability of tables. Motivated by this, we propose the Adve...
http://arxiv.org/pdf/2212.09994v1
2022-12-20T04:38:23
2022-12-20T04:38:23
{ "title": [ 0.00086545944, 0.015792847, -0.02720642, -0.054016113, 0.027053833, 0.0013856888, -0.024246216, -0.007701874, 0.005924225, 0.053375244, -0.017913818, 0.027572632, 0.0070610046, 0.004917145, -0.064208984, -0.0096206665, 0.0047569275, ...
http://arxiv.org/abs/2212.09982v1
Joint Speech Transcription and Translation: Pseudo-Labeling with Out-of-Distribution Data
Mozhdeh Gheini, Tatiana Likhomanenko, Matthias Sperber, Hendra Setiawan
cs.CL, cs.SD, eess.AS
Self-training has been shown to be helpful in addressing data scarcity for many domains, including vision, speech, and language. Specifically, self-training, or pseudo-labeling, labels unsupervised data and adds that to the training pool. In this work, we investigate and use pseudo-labeling for a recently proposed nove...
http://arxiv.org/pdf/2212.09982v1
2022-12-20T03:54:44
2022-12-20T03:54:44
{ "title": [ -0.039123535, -0.00472641, -0.03451538, -0.059295654, -0.010223389, 0.004055023, -0.04107666, 0.01725769, -0.016204834, 0.049468994, -0.006629944, -0.00024962425, 0.006389618, -0.038970947, -0.03717041, -0.0015478134, 0.0014944077, 0...
http://arxiv.org/abs/2212.09968v2
On Improving Summarization Factual Consistency from Natural Language Feedback
Yixin Liu, Budhaditya Deb, Milagro Teruel, Aaron Halfaker, Dragomir Radev, Ahmed H. Awadallah
cs.CL
Despite the recent progress in language generation models, their outputs may not always meet user expectations. In this work, we study whether informational feedback in natural language can be leveraged to improve generation quality and user preference alignment. To this end, we consider factual consistency in summariz...
http://arxiv.org/pdf/2212.09968v2
2023-10-16T04:31:22
2022-12-20T02:47:37
{ "title": [ -0.008712769, 0.00034832954, -0.021530151, -0.08062744, -0.01889038, 0.013053894, -0.068115234, -0.015327454, 0.0012788773, 0.045654297, -0.030960083, 0.011268616, 0.00983429, 0.012832642, -0.06616211, -0.0027580261, -0.011222839, 0....
http://arxiv.org/abs/2212.09955v2
BUMP: A Benchmark of Unfaithful Minimal Pairs for Meta-Evaluation of Faithfulness Metrics
Liang Ma, Shuyang Cao, Robert L. Logan IV, Di Lu, Shihao Ran, Ke Zhang, Joel Tetreault, Alejandro Jaimes
cs.CL
The proliferation of automatic faithfulness metrics for summarization has produced a need for benchmarks to evaluate them. While existing benchmarks measure the correlation with human judgements of faithfulness on model-generated summaries, they are insufficient for diagnosing whether metrics are: 1) consistent, i.e., ...
http://arxiv.org/pdf/2212.09955v2
2023-06-05T01:29:40
2022-12-20T02:17:30
{ "title": [ -0.049041748, 0.0045928955, -0.03387451, -0.05105591, -0.0056266785, 0.0085372925, -0.06591797, -0.027236938, 0.0005850792, 0.061340332, 0.027877808, 0.002210617, 0.031829834, -0.027923584, -0.042022705, 0.025985718, -0.004501343, 0....
http://arxiv.org/abs/2212.09947v1
Future Sight: Dynamic Story Generation with Large Pretrained Language Models
Brian D. Zimmerman, Gaurav Sahu, Olga Vechtomova
cs.CL, cs.AI, cs.LG
Recent advances in deep learning research, such as transformers, have bolstered the ability for automated agents to generate creative texts similar to those that a human would write. By default, transformer decoders can only generate new text with respect to previously generated text. The output distribution of candida...
http://arxiv.org/pdf/2212.09947v1
2022-12-20T01:53:26
2022-12-20T01:53:26
{ "title": [ -0.03152466, -0.025375366, -0.039398193, -0.07446289, -0.010513306, 0.015136719, -0.05255127, 0.009605408, -0.015167236, 0.03036499, 0.0024929047, 0.00308609, -0.007865906, -0.019470215, -0.036346436, 0.00217247, -0.019180298, 0.0429...
http://arxiv.org/abs/2212.09946v1
Dialog2API: Task-Oriented Dialogue with API Description and Example Programs
Raphael Shu, Elman Mansimov, Tamer Alkhouli, Nikolaos Pappas, Salvatore Romeo, Arshit Gupta, Saab Mansour, Yi Zhang, Dan Roth
cs.CL
Functionality and dialogue experience are two important factors of task-oriented dialogue systems. Conventional approaches with closed schema (e.g., conversational semantic parsing) often fail as both the functionality and dialogue experience are strongly constrained by the underlying schema. We introduce a new paradig...
http://arxiv.org/pdf/2212.09946v1
2022-12-20T01:52:46
2022-12-20T01:52:46
{ "title": [ -0.034606934, -0.019973755, -0.07043457, -0.044525146, -0.005622864, -0.022949219, -0.039886475, 0.0104904175, -0.021392822, 0.051574707, -0.0056152344, -0.011650085, 0.024215698, -0.031555176, -0.04547119, 0.0010967255, 0.003868103, ...
http://arxiv.org/abs/2212.09939v2
AnyTOD: A Programmable Task-Oriented Dialog System
Jeffrey Zhao, Yuan Cao, Raghav Gupta, Harrison Lee, Abhinav Rastogi, Mingqiu Wang, Hagen Soltau, Izhak Shafran, Yonghui Wu
cs.CL
We propose AnyTOD, an end-to-end, zero-shot task-oriented dialog (TOD) system capable of handling unseen tasks without task-specific training. We view TOD as a program executed by a language model (LM), where program logic and ontology is provided by a designer as a schema. To enable generalization to unseen schemas an...
http://arxiv.org/pdf/2212.09939v2
2023-02-13T18:26:37
2022-12-20T01:23:01
{ "title": [ -0.020324707, -0.0067634583, -0.04373169, -0.08648682, 0.020553589, 0.015640259, -0.052856445, 0.004852295, 0.017196655, 0.04309082, 0.005458832, -0.01776123, -0.008117676, -0.017425537, -0.06555176, -0.002254486, -0.00233078, 0.0093...
http://arxiv.org/abs/2212.09928v1
Improving the Robustness of Summarization Models by Detecting and Removing Input Noise
Kundan Krishna, Yao Zhao, Jie Ren, Balaji Lakshminarayanan, Jiaming Luo, Mohammad Saleh, Peter J. Liu
cs.CL, cs.LG
The evaluation of abstractive summarization models typically uses test data that is identically distributed as training data. In real-world practice, documents to be summarized may contain input noise caused by text extraction artifacts or data pipeline bugs. The robustness of model performance under distribution shift...
http://arxiv.org/pdf/2212.09928v1
2022-12-20T00:33:11
2022-12-20T00:33:11
{ "title": [ 0.0055351257, 0.025619507, -0.031799316, -0.072021484, 0.006717682, 0.0007123947, -0.0791626, -0.04421997, 0.008392334, 0.04144287, -0.004169464, 0.021621704, -0.014694214, 0.00018084049, -0.064208984, 0.021057129, 0.00071811676, 0.0...
http://arxiv.org/abs/2212.09917v1
Inverse Reinforcement Learning for Text Summarization
Yu Fu, Deyi Xiong, Yue Dong
cs.CL
Current state-of-the-art summarization models are trained with either maximum likelihood estimation (MLE) or reinforcement learning (RL). In this study, we investigate the third training paradigm and argue that inverse reinforcement learning (IRL) may be more suitable for text summarization. IRL focuses on estimating t...
http://arxiv.org/pdf/2212.09917v1
2022-12-19T23:45:05
2022-12-19T23:45:05
{ "title": [ -0.0028209686, 0.00034308434, 0.0054626465, -0.035186768, -0.013954163, 0.004535675, -0.014022827, -0.027923584, -0.027664185, 0.03845215, -0.028961182, 0.035064697, -0.03137207, 0.005329132, -0.03744507, -0.024642944, -0.020462036, ...
http://arxiv.org/abs/2212.09912v2
Tokenization Consistency Matters for Generative Models on Extractive NLP Tasks
Kaiser Sun, Peng Qi, Yuhao Zhang, Lan Liu, William Yang Wang, Zhiheng Huang
cs.CL
Generative models have been widely applied to solve extractive tasks, where parts of the input is extracted to form the desired output, and achieved significant success. For example, in extractive question answering (QA), generative models have constantly yielded state-of-the-art results. In this work, we identify the ...
http://arxiv.org/pdf/2212.09912v2
2023-10-24T20:59:33
2022-12-19T23:33:21
{ "title": [ -0.033966064, -0.0024433136, -0.044677734, -0.047821045, -0.016418457, 0.0008249283, -0.047973633, -0.0011835098, -0.004398346, 0.019622803, -0.036224365, 0.009819031, -0.018356323, -0.007751465, -0.053527832, -0.007774353, -0.021118164,...
http://arxiv.org/abs/2212.09897v1
Inducing Character-level Structure in Subword-based Language Models with Type-level Interchange Intervention Training
Jing Huang, Zhengxuan Wu, Kyle Mahowald, Christopher Potts
cs.CL
Language tasks involving character-level manipulations (e.g., spelling correction, many word games) are challenging for models based in subword tokenization. To address this, we adapt the interchange intervention training method of Geiger et al. (2021) to operate on type-level variables over characters. This allows us ...
http://arxiv.org/pdf/2212.09897v1
2022-12-19T22:37:46
2022-12-19T22:37:46
{ "title": [ -0.041931152, -0.015975952, -0.022369385, -0.026794434, -0.031021118, 0.012260437, -0.03439331, 0.02293396, -0.0028457642, 0.0317688, 0.0033454895, 0.016601562, 0.014595032, -0.032196045, -0.018859863, 0.005050659, -0.029876709, 0.01...
http://arxiv.org/abs/2212.09895v1
Improved Long-Form Spoken Language Translation with Large Language Models
Arya D. McCarthy, Hao Zhang, Shankar Kumar, Felix Stahlberg, Axel H. Ng
cs.CL
A challenge in spoken language translation is that plenty of spoken content is long-form, but short units are necessary for obtaining high-quality translations. To address this mismatch, we fine-tune a general-purpose, large language model to split long ASR transcripts into segments that can be independently translated...
http://arxiv.org/pdf/2212.09895v1
2022-12-19T22:36:53
2022-12-19T22:36:53
{ "title": [ -0.027023315, 0.0012865067, -0.02558899, -0.07183838, 0.004940033, 0.012306213, -0.06451416, 0.010665894, 0.0022697449, 0.04107666, -0.012619019, -0.0021457672, 0.014328003, -0.018508911, -0.041046143, 0.013076782, -0.0004541874, 0.0...
http://arxiv.org/abs/2212.09885v2
Python Code Generation by Asking Clarification Questions
Haau-Sing Li, Mohsen Mesgar, André F. T. Martins, Iryna Gurevych
cs.CL
Code generation from text requires understanding the user's intent from a natural language description and generating an executable code snippet that satisfies this intent. While recent pretrained language models demonstrate remarkable performance for this task, these models fail when the given natural language descrip...
http://arxiv.org/pdf/2212.09885v2
2023-05-26T16:03:08
2022-12-19T22:08:36
{ "title": [ -0.041931152, -0.013221741, -0.04815674, -0.017959595, -0.03189087, -0.0009608269, -0.029418945, 0.012390137, -0.0052833557, 0.0028076172, -0.024978638, -0.025939941, -0.040039062, -0.04852295, -0.0061416626, -0.031585693, 0.0064468384, ...
http://arxiv.org/abs/2212.09879v1
Unsigned Play by Milan Kundera? An Authorship Attribution Study
Lenka Jungmannová, Petr Plecháč
cs.CL
In addition to being a widely recognised novelist, Milan Kundera has also authored three pieces for theatre: The Owners of the Keys (Majitel\'e kl\'i\v{c}\r{u}, 1961), The Blunder (Pt\'akovina, 1967), and Jacques and his Master (Jakub a jeho p\'an, 1971). In recent years, however, the hypothesis has been raised that Ku...
http://arxiv.org/pdf/2212.09879v1
2022-12-19T21:59:22
2022-12-19T21:59:22
{ "title": [ -0.019805908, 0.013626099, -0.058776855, 0.011520386, -0.00843811, -0.045043945, -0.025375366, -0.05404663, -0.048980713, 0.049072266, -0.00053071976, 0.013153076, -0.016189575, -0.027572632, -0.00995636, -0.02796936, -0.0067214966, ...
http://arxiv.org/abs/2212.09873v2
A Comparative Study on Textual Saliency of Styles from Eye Tracking, Annotations, and Language Models
Karin de Langis, Dongyeop Kang
cs.CL
There is growing interest in incorporating eye-tracking data and other implicit measures of human language processing into natural language processing (NLP) pipelines. The data from human language processing contain unique insight into human linguistic understanding that could be exploited by language models. However, ...
http://arxiv.org/pdf/2212.09873v2
2023-10-22T20:28:18
2022-12-19T21:50:36
{ "title": [ -0.030380249, -0.017868042, -0.029296875, 0.009757996, -0.034301758, -0.022583008, -0.01461792, -0.017669678, 0.03286743, 0.039215088, -0.023712158, 0.004085541, 0.036071777, 0.00010293722, -0.047851562, 0.0042304993, 0.005672455, 0....
http://arxiv.org/abs/2212.09867v1
Detecting Contradictory COVID-19 Drug Efficacy Claims from Biomedical Literature
Daniel N. Sosa, Malavika Suresh, Christopher Potts, Russ B. Altman
cs.CL
The COVID-19 pandemic created a deluge of questionable and contradictory scientific claims about drug efficacy -- an "infodemic" with lasting consequences for science and society. In this work, we argue that NLP models can help domain experts distill and understand the literature in this complex, high-stakes area. Our ...
http://arxiv.org/pdf/2212.09867v1
2022-12-19T21:37:36
2022-12-19T21:37:36
{ "title": [ -0.020111084, 0.011703491, -0.06896973, -0.04144287, -0.009475708, -0.024887085, -0.014228821, -0.0073547363, -0.045166016, 0.01474762, -0.015274048, 0.019927979, -0.013931274, -0.036010742, -0.0053634644, -0.01852417, -0.0132369995, ...
http://arxiv.org/abs/2212.09865v2
Z-ICL: Zero-Shot In-Context Learning with Pseudo-Demonstrations
Xinxi Lyu, Sewon Min, Iz Beltagy, Luke Zettlemoyer, Hannaneh Hajishirzi
cs.CL, cs.AI
Although large language models can be prompted for both zero- and few-shot learning, performance drops significantly when no demonstrations are available. In this paper, we introduce Z-ICL, a new zero-shot method that closes the gap by constructing pseudo-demonstrations for a given test input using a raw text corpus. C...
http://arxiv.org/pdf/2212.09865v2
2023-06-03T22:51:39
2022-12-19T21:34:26
{ "title": [ -0.0005455017, -0.006511688, -0.033050537, -0.061584473, 0.011108398, -0.0062065125, -0.0074806213, -0.0075531006, 0.0151901245, 0.02243042, -0.025100708, 0.0033874512, -0.024261475, -0.01776123, -0.027511597, -0.010437012, -0.0072364807...
http://arxiv.org/abs/2212.09864v2
Synthetic Pre-Training Tasks for Neural Machine Translation
Zexue He, Graeme Blackwood, Rameswar Panda, Julian McAuley, Rogerio Feris
cs.CL, cs.AI
Pre-training models with large crawled corpora can lead to issues such as toxicity and bias, as well as copyright and privacy concerns. A promising way of alleviating such concerns is to conduct pre-training with synthetic tasks and data, since no real-world information is ingested by the model. Our goal in this paper ...
http://arxiv.org/pdf/2212.09864v2
2023-05-31T01:34:54
2022-12-19T21:34:00
{ "title": [ -0.022781372, -0.0015935898, -0.011146545, -0.06390381, -0.004585266, 0.011192322, -0.03765869, 0.0031795502, -0.020202637, 0.051086426, 0.0005698204, 0.009170532, 0.008216858, -0.01386261, -0.031677246, 0.020812988, -0.0035877228, 0...
http://arxiv.org/abs/2212.09858v1
Continuous Semi-Supervised Nonnegative Matrix Factorization
Michael R. Lindstrom, Xiaofu Ding, Feng Liu, Anand Somayajula, Deanna Needell
cs.CL, cs.LG
Nonnegative matrix factorization can be used to automatically detect topics within a corpus in an unsupervised fashion. The technique amounts to an approximation of a nonnegative matrix as the product of two nonnegative matrices of lower rank. In this paper, we show this factorization can be combined with regression on...
http://arxiv.org/pdf/2212.09858v1
2022-12-19T21:07:27
2022-12-19T21:07:27
{ "title": [ -0.003435135, 0.0074043274, -0.05734253, -0.07208252, 0.01979065, -0.01222229, -0.036865234, 0.0074920654, -0.013092041, 0.01285553, -0.017120361, -0.0015287399, -0.0079422, -0.03781128, -0.019989014, -0.023498535, -0.014633179, 0.02...
http://arxiv.org/abs/2212.09855v1
MANTIS at TSAR-2022 Shared Task: Improved Unsupervised Lexical Simplification with Pretrained Encoders
Xiaofei Li, Daniel Wiechmann, Yu Qiao, Elma Kerz
cs.CL
In this paper we present our contribution to the TSAR-2022 Shared Task on Lexical Simplification of the EMNLP 2022 Workshop on Text Simplification, Accessibility, and Readability. Our approach builds on and extends the unsupervised lexical simplification system with pretrained encoders (LSBert) system in the following ...
http://arxiv.org/pdf/2212.09855v1
2022-12-19T20:57:45
2022-12-19T20:57:45
{ "title": [ -0.022369385, 0.009887695, -0.042144775, -0.046905518, -0.011077881, 0.0015907288, -0.079711914, -0.008377075, 0.016479492, 0.046325684, -0.004196167, 0.00749588, 0.044921875, -0.027664185, -0.0003592968, 0.036376953, 0.008010864, 0....
http://arxiv.org/abs/2212.09849v5
Dataless Knowledge Fusion by Merging Weights of Language Models
Xisen Jin, Xiang Ren, Daniel Preotiuc-Pietro, Pengxiang Cheng
cs.CL, cs.LG
Fine-tuning pre-trained language models has become the prevalent paradigm for building downstream NLP models. Oftentimes fine-tuned models are readily available but their training data is not, due to data privacy or intellectual property concerns. This creates a barrier to fusing knowledge across individual models to y...
http://arxiv.org/pdf/2212.09849v5
2023-10-12T19:02:38
2022-12-19T20:46:43
{ "title": [ -0.03463745, -0.016845703, -0.060028076, -0.087402344, 0.0021076202, 0.0038928986, -0.060821533, 0.023071289, 0.0049362183, 0.03262329, -0.02609253, -0.008583069, 0.012207031, -0.0073127747, -0.036102295, -0.008934021, -0.003993988, ...
http://arxiv.org/abs/2212.09848v1
(Psycho-)Linguistic Features Meet Transformer Models for Improved Explainable and Controllable Text Simplification
Yu Qiao, Xiaofei Li, Daniel Wiechmann, Elma Kerz
cs.CL
State-of-the-art text simplification (TS) systems adopt end-to-end neural network models to directly generate the simplified version of the input text, and usually function as a blackbox. Moreover, TS is usually treated as an all-purpose generic task under the assumption of homogeneity, where the same simplification is...
http://arxiv.org/pdf/2212.09848v1
2022-12-19T20:46:21
2022-12-19T20:46:21
{ "title": [ -0.02532959, -0.008995056, -0.023742676, -0.050811768, -0.018585205, 0.008888245, -0.067993164, -0.0121536255, -0.0006070137, 0.04260254, -0.00053310394, 0.02279663, 0.043884277, -0.01335144, -0.013015747, 0.010368347, -0.00028562546, ...
http://arxiv.org/abs/2212.09839v1
Exploring Hybrid and Ensemble Models for Multiclass Prediction of Mental Health Status on Social Media
Sourabh Zanwar, Daniel Wiechmann, Yu Qiao, Elma Kerz
cs.CL
In recent years, there has been a surge of interest in research on automatic mental health detection (MHD) from social media data leveraging advances in natural language processing and machine learning techniques. While significant progress has been achieved in this interdisciplinary research area, the vast majority of...
http://arxiv.org/pdf/2212.09839v1
2022-12-19T20:31:47
2022-12-19T20:31:47
{ "title": [ 0.028442383, 0.0066184998, -0.02949524, -0.06341553, 0.03466797, -0.020080566, -0.008522034, 0.009498596, -0.010597229, 0.03201294, -0.033325195, 0.021957397, 0.0010795593, -0.014709473, -0.038269043, -0.0116119385, -0.012107849, 0.0...
http://arxiv.org/abs/2212.09825v2
What to Read in a Contract? Party-Specific Summarization of Legal Obligations, Entitlements, and Prohibitions
Abhilasha Sancheti, Aparna Garimella, Balaji Vasan Srinivasan, Rachel Rudinger
cs.CL
Reviewing and comprehending key obligations, entitlements, and prohibitions in legal contracts can be a tedious task due to their length and domain-specificity. Furthermore, the key rights and duties requiring review vary for each contracting party. In this work, we propose a new task of party-specific extractive summa...
http://arxiv.org/pdf/2212.09825v2
2023-10-24T20:43:40
2022-12-19T19:53:14
{ "title": [ -0.028457642, -0.0018663406, -0.049987793, -0.052093506, -0.026016235, -0.011238098, -0.079711914, 0.01374054, -0.03253174, 0.03793335, -0.024353027, 0.028411865, -0.0014810562, -0.013336182, -0.003616333, -0.030181885, 0.027374268, ...
http://arxiv.org/abs/2212.09811v3
Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model
Yeskendir Koishekenov, Alexandre Berard, Vassilina Nikoulina
cs.CL, cs.AI, cs.LG
The recently released NLLB-200 is a set of multilingual Neural Machine Translation models that cover 202 languages. The largest model is based on a Mixture of Experts architecture and achieves SoTA results across many language pairs. It contains 54.5B parameters and requires at least four 32GB GPUs just for inference. ...
http://arxiv.org/pdf/2212.09811v3
2023-07-07T09:53:20
2022-12-19T19:29:40
{ "title": [ -0.023117065, -0.020584106, -0.0059814453, -0.075805664, -0.009223938, 0.027359009, -0.053100586, 0.032409668, 0.0005660057, 0.053771973, 0.0023422241, 0.0052108765, 0.046142578, -0.020126343, -0.04623413, 0.025558472, -0.002576828, ...
http://arxiv.org/abs/2212.09803v3
Training Trajectories of Language Models Across Scales
Mengzhou Xia, Mikel Artetxe, Chunting Zhou, Xi Victoria Lin, Ramakanth Pasunuru, Danqi Chen, Luke Zettlemoyer, Ves Stoyanov
cs.CL, cs.AI, cs.LG
Scaling up language models has led to unprecedented performance gains, but little is understood about how the training dynamics change as models get larger. How do language models of different sizes learn during pre-training? Why do larger language models demonstrate more desirable behaviors? In this paper, we analyze ...
http://arxiv.org/pdf/2212.09803v3
2023-05-30T03:33:27
2022-12-19T19:16:29
{ "title": [ -0.03378296, -0.0079574585, -0.03778076, -0.06222534, -0.010345459, 0.004306793, -0.06286621, 0.0049362183, -0.00011712313, 0.024627686, -0.0020275116, 0.005012512, 0.0030899048, -0.03414917, -0.026321411, 0.018249512, -0.013130188, ...
http://arxiv.org/abs/2212.09750v1
Human-in-the-loop Abstractive Dialogue Summarization
Jiaao Chen, Mohan Dodda, Diyi Yang
cs.CL, cs.AI
Abstractive dialogue summarization has received increasing attention recently. Despite the fact that most of the current dialogue summarization systems are trained to maximize the likelihood of human-written summaries and have achieved significant results, there is still a huge gap in generating high-quality summaries ...
http://arxiv.org/pdf/2212.09750v1
2022-12-19T19:11:27
2022-12-19T19:11:27
{ "title": [ 0.017196655, -0.0075035095, -0.0317688, -0.07147217, -0.024353027, -0.014305115, -0.056518555, 0.0026016235, 0.019424438, 0.05984497, -0.011856079, 0.00024080276, 0.0073623657, -0.01890564, -0.05078125, 0.010749817, -0.0040779114, 0....
http://arxiv.org/abs/2212.09747v2
Do CoNLL-2003 Named Entity Taggers Still Work Well in 2023?
Shuheng Liu, Alan Ritter
cs.CL
The CoNLL-2003 English named entity recognition (NER) dataset has been widely used to train and evaluate NER models for almost 20 years. However, it is unclear how well models that are trained on this 20-year-old data and developed over a period of decades using the same test set will perform when applied on modern dat...
http://arxiv.org/pdf/2212.09747v2
2023-07-12T02:41:46
2022-12-19T18:59:56
{ "title": [ 0.0013580322, -0.027069092, -0.022628784, -0.040985107, -0.043640137, -0.013999939, -0.028457642, 0.014167786, 0.013160706, 0.02130127, 0.01927185, -0.028839111, 0.0256958, -0.049194336, 0.005432129, -0.006454468, 0.017593384, 0.0072...
http://arxiv.org/abs/2212.09746v4
Evaluating Human-Language Model Interaction
Mina Lee, Megha Srivastava, Amelia Hardy, John Thickstun, Esin Durmus, Ashwin Paranjape, Ines Gerard-Ursin, Xiang Lisa Li, Faisal Ladhak, Frieda Rong, Rose E. Wang, Minae Kwon, Joon Sung Park, Hancheng Cao, Tony Lee, Rishi Bommasani, Michael Bernstein, Percy Liang
cs.CL
Many real-world applications of language models (LMs), such as writing assistance and code autocomplete, involve human-LM interaction. However, most benchmarks are non-interactive in that a model produces output without human involvement. To evaluate human-LM interaction, we develop a new framework, Human-AI Language-b...
http://arxiv.org/pdf/2212.09746v4
2023-09-10T13:31:08
2022-12-19T18:59:45
{ "title": [ -0.018341064, -0.008621216, -0.02520752, -0.0395813, 0.0008597374, -0.0071525574, -0.043884277, -0.018630981, -0.0017194748, 0.03668213, -0.012870789, -0.012580872, -0.0059661865, -0.02041626, -0.06719971, 0.00031208992, -0.015731812, ...
http://arxiv.org/abs/2212.09744v1
DSI++: Updating Transformer Memory with New Documents
Sanket Vaibhav Mehta, Jai Gupta, Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Jinfeng Rao, Marc Najork, Emma Strubell, Donald Metzler
cs.CL, cs.AI, cs.IR, cs.LG
Differentiable Search Indices (DSIs) encode a corpus of documents in the parameters of a model and use the same model to map queries directly to relevant document identifiers. Despite the strong performance of DSI models, deploying them in situations where the corpus changes over time is computationally expensive becau...
http://arxiv.org/pdf/2212.09744v1
2022-12-19T18:59:34
2022-12-19T18:59:34
{ "title": [ 0.0013046265, -0.01159668, -0.016571045, 0.0010814667, -0.030517578, 0.009506226, -0.03149414, -0.033233643, -0.012641907, 0.037322998, -0.054748535, -0.03237915, 0.04574585, -0.02532959, -0.011253357, -0.059020996, 0.02432251, -0.00...
http://arxiv.org/abs/2212.09741v3
One Embedder, Any Task: Instruction-Finetuned Text Embeddings
Hongjin Su, Weijia Shi, Jungo Kasai, Yizhong Wang, Yushi Hu, Mari Ostendorf, Wen-tau Yih, Noah A. Smith, Luke Zettlemoyer, Tao Yu
cs.CL
We introduce INSTRUCTOR, a new method for computing text embeddings given task instructions: every text input is embedded together with instructions explaining the use case (e.g., task and domain descriptions). Unlike encoders from prior work that are more specialized, INSTRUCTOR is a single embedder that can generate ...
http://arxiv.org/pdf/2212.09741v3
2023-05-30T15:22:50
2022-12-19T18:57:05
{ "title": [ 0.046173096, -0.02027893, -0.021606445, -0.06347656, -0.02798462, -0.016723633, -0.018508911, 0.02645874, -0.003446579, 0.02999878, -0.02885437, -0.009361267, 0.022125244, -0.030410767, 0.011856079, -0.03463745, 0.00983429, 0.0256958...
http://arxiv.org/abs/2212.09739v4
LENS: A Learnable Evaluation Metric for Text Simplification
Mounica Maddela, Yao Dou, David Heineman, Wei Xu
cs.CL
Training learnable metrics using modern language models has recently emerged as a promising method for the automatic evaluation of machine translation. However, existing human evaluation datasets for text simplification have limited annotations that are based on unitary or outdated models, making them unsuitable for th...
http://arxiv.org/pdf/2212.09739v4
2023-07-07T20:41:23
2022-12-19T18:56:52
{ "title": [ -0.048797607, -0.011398315, -0.04763794, -0.055023193, -0.0013189316, 0.012252808, -0.064575195, -0.011444092, 0.0011444092, 0.053771973, 0.0045280457, 0.032318115, -0.00024402142, -0.0031490326, -0.062561035, 0.00043940544, 0.0003459453...
http://arxiv.org/abs/2212.09736v2
Don't Generate, Discriminate: A Proposal for Grounding Language Models to Real-World Environments
Yu Gu, Xiang Deng, Yu Su
cs.CL, cs.AI, I.2.7
A key missing capacity of current language models (LMs) is grounding to real-world environments. Most existing work for grounded language understanding uses LMs to directly generate plans that can be executed in the environment to achieve the desired effects. It thereby casts the burden of ensuring grammaticality, fait...
http://arxiv.org/pdf/2212.09736v2
2023-05-03T04:32:35
2022-12-19T18:55:21
{ "title": [ -0.015960693, -0.010536194, -0.03253174, -0.05758667, 0.010192871, -0.0039749146, -0.053649902, -0.003314972, -0.0006079674, 0.032104492, -0.011268616, -0.00062322617, 0.004283905, -0.014389038, -0.05493164, 0.0050354004, -0.020812988, ...
http://arxiv.org/abs/2212.09730v2
Speaking Style Conversion in the Waveform Domain Using Discrete Self-Supervised Units
Gallil Maimon, Yossi Adi
cs.SD, cs.CL, cs.LG, eess.AS
We introduce DISSC, a novel, lightweight method that converts the rhythm, pitch contour and timbre of a recording to a target speaker in a textless manner. Unlike DISSC, most voice conversion (VC) methods focus primarily on timbre, and ignore people's unique speaking style (prosody). The proposed approach uses a pretra...
http://arxiv.org/pdf/2212.09730v2
2023-10-18T19:23:27
2022-12-19T18:53:04
{ "title": [ -0.000033020973, 0.003326416, -0.07952881, -0.04937744, 0.0028533936, -0.04660034, -0.029251099, 0.021484375, -0.013748169, 0.0597229, 0.002937317, -0.003211975, 0.039642334, -0.041625977, -0.031707764, 0.012413025, -0.00223732, 0.03...
http://arxiv.org/abs/2212.09726v1
Improving Faithfulness of Abstractive Summarization by Controlling Confounding Effect of Irrelevant Sentences
Asish Ghoshal, Arash Einolghozati, Ankit Arun, Haoran Li, Lili Yu, Yashar Mehdad, Scott Wen-tau Yih, Asli Celikyilmaz
cs.CL, cs.AI, cs.LG
Lack of factual correctness is an issue that still plagues state-of-the-art summarization systems despite their impressive progress on generating seemingly fluent summaries. In this paper, we show that factual inconsistency can be caused by irrelevant parts of the input text, which act as confounders. To that end, we l...
http://arxiv.org/pdf/2212.09726v1
2022-12-19T18:51:06
2022-12-19T18:51:06
{ "title": [ 0.0103302, -0.0006117821, -0.024612427, -0.06921387, -0.01586914, -0.007369995, -0.052337646, -0.020614624, 0.006916046, 0.06768799, -0.023010254, 0.0259552, 0.016647339, 0.0046081543, -0.06707764, 0.009033203, -0.020187378, 0.050659...
http://arxiv.org/abs/2212.09724v3
A Retrieve-and-Read Framework for Knowledge Graph Link Prediction
Vardaan Pahuja, Boshi Wang, Hugo Latapie, Jayanth Srinivasa, Yu Su
cs.CL, cs.AI
Knowledge graph (KG) link prediction aims to infer new facts based on existing facts in the KG. Recent studies have shown that using the graph neighborhood of a node via graph neural networks (GNNs) provides more useful information compared to just using the query information. Conventional GNNs for KG link prediction f...
http://arxiv.org/pdf/2212.09724v3
2023-10-22T17:57:24
2022-12-19T18:50:54
{ "title": [ -0.021697998, -0.019714355, -0.053253174, -0.08679199, -0.013320923, 0.0048942566, -0.06488037, 0.019699097, -0.009468079, 0.04510498, -0.002368927, 0.0013771057, 0.018081665, -0.0110321045, -0.038116455, 0.005138397, 0.002494812, 0....
http://arxiv.org/abs/2212.09723v1
MANER: Mask Augmented Named Entity Recognition for Extreme Low-Resource Languages
Shashank Sonkar, Zichao Wang, Richard G. Baraniuk
cs.CL
This paper investigates the problem of Named Entity Recognition (NER) for extreme low-resource languages with only a few hundred tagged data samples. NER is a fundamental task in Natural Language Processing (NLP). A critical driver accelerating NER systems' progress is the existence of large-scale language corpora that...
http://arxiv.org/pdf/2212.09723v1
2022-12-19T18:49:50
2022-12-19T18:49:50
{ "title": [ -0.031066895, 0.005695343, -0.062561035, -0.07299805, 0.022506714, 0.0026798248, -0.047607422, 0.015365601, -0.0029125214, 0.028274536, -0.021987915, 0.011062622, 0.01638794, -0.015731812, -0.028182983, 0.011543274, -0.0124053955, 0....