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/2310.17737v1
ArchBERT: Bi-Modal Understanding of Neural Architectures and Natural Languages
Mohammad Akbari, Saeed Ranjbar Alvar, Behnam Kamranian, Amin Banitalebi-Dehkordi, Yong Zhang
cs.CL
Building multi-modal language models has been a trend in the recent years, where additional modalities such as image, video, speech, etc. are jointly learned along with natural languages (i.e., textual information). Despite the success of these multi-modal language models with different modalities, there is no existing...
http://arxiv.org/pdf/2310.17737v1
2023-10-26T18:58:52
2023-10-26T18:58:52
{ "title": [ -0.024597168, -0.008651733, -0.022583008, -0.042633057, -0.004837036, 0.0031032562, -0.03982544, -0.03503418, -0.011665344, 0.022583008, -0.0042381287, -0.011306763, 0.017822266, -0.041412354, -0.029647827, 0.0065193176, -0.01285553, ...
http://arxiv.org/abs/2310.17734v1
Investigating Multilingual Coreference Resolution by Universal Annotations
Haixia Chai, Michael Strube
cs.CL
Multilingual coreference resolution (MCR) has been a long-standing and challenging task. With the newly proposed multilingual coreference dataset, CorefUD (Nedoluzhko et al., 2022), we conduct an investigation into the task by using its harmonized universal morphosyntactic and coreference annotations. First, we study c...
http://arxiv.org/pdf/2310.17734v1
2023-10-26T18:50:04
2023-10-26T18:50:04
{ "title": [ 0.01259613, 0.0049705505, -0.01878357, -0.08679199, 0.00793457, 0.0031490326, -0.056610107, 0.0035324097, -0.0046806335, 0.07281494, 0.005367279, 0.024612427, 0.034179688, -0.008651733, -0.04385376, 0.043121338, -0.01159668, 0.046203...
http://arxiv.org/abs/2310.17723v1
ZeroQuant-HERO: Hardware-Enhanced Robust Optimized Post-Training Quantization Framework for W8A8 Transformers
Zhewei Yao, Reza Yazdani Aminabadi, Stephen Youn, Xiaoxia Wu, Elton Zheng, Yuxiong He
cs.LG, cs.CL
Quantization techniques are pivotal in reducing the memory and computational demands of deep neural network inference. Existing solutions, such as ZeroQuant, offer dynamic quantization for models like BERT and GPT but overlook crucial memory-bounded operators and the complexities of per-token quantization. Addressing t...
http://arxiv.org/pdf/2310.17723v1
2023-10-26T18:34:41
2023-10-26T18:34:41
{ "title": [ 0.0037994385, 0.018844604, -0.028686523, -0.033447266, 0.0014457703, -0.022994995, -0.018066406, -0.03881836, -0.01928711, 0.075683594, -0.0127334595, 0.012634277, 0.061920166, -0.01864624, -0.028869629, 0.014984131, 0.010116577, 0.0...
http://arxiv.org/abs/2310.17722v1
Large Language Models as Generalizable Policies for Embodied Tasks
Andrew Szot, Max Schwarzer, Harsh Agrawal, Bogdan Mazoure, Walter Talbott, Katherine Metcalf, Natalie Mackraz, Devon Hjelm, Alexander Toshev
cs.LG, cs.AI, cs.CL
We show that large language models (LLMs) can be adapted to be generalizable policies for embodied visual tasks. Our approach, called Large LAnguage model Reinforcement Learning Policy (LLaRP), adapts a pre-trained frozen LLM to take as input text instructions and visual egocentric observations and output actions direc...
http://arxiv.org/pdf/2310.17722v1
2023-10-26T18:32:05
2023-10-26T18:32:05
{ "title": [ -0.014785767, -0.013763428, -0.010971069, -0.061553955, -0.019165039, 0.010902405, -0.058532715, -0.000954628, -0.019882202, 0.029296875, -0.009979248, -0.0028324127, 0.0018720627, -0.015571594, -0.030441284, 0.0036907196, -0.010025024, ...
http://arxiv.org/abs/2310.17721v1
From Transcripts to Insights: Uncovering Corporate Risks Using Generative AI
Alex Kim, Maximilian Muhn, Valeri Nikolaev
econ.GN, cs.AI, cs.CL, q-fin.EC
We explore the value of generative AI tools, such as ChatGPT, in helping investors uncover dimensions of corporate risk. We develop and validate firm-level measures of risk exposure to political, climate, and AI-related risks. Using the GPT 3.5 model to generate risk summaries and assessments from the context provided ...
http://arxiv.org/pdf/2310.17721v1
2023-10-26T18:30:37
2023-10-26T18:30:37
{ "title": [ -0.018875122, -0.0116119385, -0.05847168, -0.05343628, -0.005176544, -0.019256592, -0.03878784, -0.0023593903, -0.012611389, 0.028564453, -0.051483154, 0.027786255, -0.028198242, -0.0045547485, -0.03781128, -0.031341553, -0.0088272095, ...
http://arxiv.org/abs/2310.17715v1
Outlier Dimensions Encode Task-Specific Knowledge
William Rudman, Catherine Chen, Carsten Eickhoff
cs.CL, cs.AI
Representations from large language models (LLMs) are known to be dominated by a small subset of dimensions with exceedingly high variance. Previous works have argued that although ablating these outlier dimensions in LLM representations hurts downstream performance, outlier dimensions are detrimental to the representa...
http://arxiv.org/pdf/2310.17715v1
2023-10-26T18:22:13
2023-10-26T18:22:13
{ "title": [ 0.0060043335, -0.01838684, -0.029388428, -0.04611206, -0.060668945, -0.023117065, -0.036865234, 0.007396698, -0.019332886, 0.02911377, -0.0010900497, -0.0032081604, -0.017150879, -0.053955078, -0.0016031265, -0.015083313, 0.035705566, ...
http://arxiv.org/abs/2310.17714v1
Nearest Neighbor Search over Vectorized Lexico-Syntactic Patterns for Relation Extraction from Financial Documents
Pawan Kumar Rajpoot, Ankur Parikh
cs.CL, cs.CE
Relation extraction (RE) has achieved remarkable progress with the help of pre-trained language models. However, existing RE models are usually incapable of handling two situations: implicit expressions and long-tail relation classes, caused by language complexity and data sparsity. Further, these approaches and models...
http://arxiv.org/pdf/2310.17714v1
2023-10-26T18:19:56
2023-10-26T18:19:56
{ "title": [ -0.019363403, -0.018676758, -0.03643799, -0.06201172, -0.012077332, 0.0049324036, -0.054626465, -0.0012807846, 0.02142334, 0.033721924, -0.030136108, -0.011070251, 0.0045280457, 0.007827759, -0.07763672, -0.023590088, -0.010940552, 0...
http://arxiv.org/abs/2310.17711v1
Is Explanation the Cure? Misinformation Mitigation in the Short Term and Long Term
Yi-Li Hsu, Shih-Chieh Dai, Aiping Xiong, Lun-Wei Ku
cs.CL
With advancements in natural language processing (NLP) models, automatic explanation generation has been proposed to mitigate misinformation on social media platforms in addition to adding warning labels to identified fake news. While many researchers have focused on generating good explanations, how these explanations...
http://arxiv.org/pdf/2310.17711v1
2023-10-26T18:12:02
2023-10-26T18:12:02
{ "title": [ 0.013870239, 0.006084442, -0.036499023, -0.06640625, 0.020614624, -0.02722168, 0.001750946, -0.0022182465, -0.06555176, 0.02619934, 0.007572174, 0.004119873, -0.009376526, -0.049072266, -0.0057792664, -0.020889282, 0.0019741058, 0.04...
http://arxiv.org/abs/2310.17703v1
The impact of using an AI chatbot to respond to patient messages
Shan Chen, Marco Guevara, Shalini Moningi, Frank Hoebers, Hesham Elhalawani, Benjamin H. Kann, Fallon E. Chipidza, Jonathan Leeman, Hugo J. W. L. Aerts, Timothy Miller, Guergana K. Savova, Raymond H. Mak, Maryam Lustberg, Majid Afshar, Danielle S. Bitterman
cs.CL
Documentation burden is a major contributor to clinician burnout, which is rising nationally and is an urgent threat to our ability to care for patients. Artificial intelligence (AI) chatbots, such as ChatGPT, could reduce clinician burden by assisting with documentation. Although many hospitals are actively integratin...
http://arxiv.org/pdf/2310.17703v1
2023-10-26T18:03:46
2023-10-26T18:03:46
{ "title": [ 0.024719238, -0.010375977, -0.011444092, -0.04385376, -0.017456055, -0.014160156, 0.0054512024, 0.0032539368, -0.032562256, 0.06124878, -0.0211792, -0.016921997, 0.01109314, -0.016693115, -0.014640808, -0.042419434, 0.023849487, 0.02...
http://arxiv.org/abs/2310.17690v1
Non-contrastive sentence representations via self-supervision
Marco Farina, Duccio Pappadopulo
cs.CL
Sample contrastive methods, typically referred to simply as contrastive are the foundation of most unsupervised methods to learn text and sentence embeddings. On the other hand, a different class of self-supervised loss functions and methods have been considered in the computer vision community and referred to as dimen...
http://arxiv.org/pdf/2310.17690v1
2023-10-26T18:00:00
2023-10-26T18:00:00
{ "title": [ 0.017318726, 0.011024475, -0.013618469, -0.06713867, 0.0018978119, -0.0008444786, -0.048980713, -0.006626129, 0.012229919, 0.028259277, -0.024032593, 0.006072998, -0.0031013489, -0.004055023, -0.04711914, 0.0018777847, -0.012771606, ...
http://arxiv.org/abs/2310.17688v2
Managing AI Risks in an Era of Rapid Progress
Yoshua Bengio, Geoffrey Hinton, Andrew Yao, Dawn Song, Pieter Abbeel, Yuval Noah Harari, Ya-Qin Zhang, Lan Xue, Shai Shalev-Shwartz, Gillian Hadfield, Jeff Clune, Tegan Maharaj, Frank Hutter, Atılım Güneş Baydin, Sheila McIlraith, Qiqi Gao, Ashwin Acharya, David Krueger, Anca Dragan, Philip Torr, Stuart Russell, Daniel...
cs.CY, cs.AI, cs.CL, cs.LG
In this short consensus paper, we outline risks from upcoming, advanced AI systems. We examine large-scale social harms and malicious uses, as well as an irreversible loss of human control over autonomous AI systems. In light of rapid and continuing AI progress, we propose urgent priorities for AI R&D and governance.
http://arxiv.org/pdf/2310.17688v2
2023-11-12T14:33:20
2023-10-26T17:59:06
{ "title": [ 0.020523071, 0.0020179749, -0.033599854, -0.04840088, -0.005584717, -0.023422241, 0.012435913, 0.008117676, -0.036468506, 0.026016235, -0.045684814, 0.033599854, -0.02494812, -0.017471313, -0.006149292, -0.015609741, 0.024047852, 0.0...
http://arxiv.org/abs/2310.17644v1
torchdistill Meets Hugging Face Libraries for Reproducible, Coding-Free Deep Learning Studies: A Case Study on NLP
Yoshitomo Matsubara
cs.CL, cs.CV, cs.LG
Reproducibility in scientific work has been becoming increasingly important in research communities such as machine learning, natural language processing, and computer vision communities due to the rapid development of the research domains supported by recent advances in deep learning. In this work, we present a signif...
http://arxiv.org/pdf/2310.17644v1
2023-10-26T17:57:15
2023-10-26T17:57:15
{ "title": [ -0.022567749, -0.0135269165, -0.05038452, -0.015808105, -0.031555176, -0.034576416, -0.025024414, -0.020889282, 0.0073394775, 0.026687622, -0.044189453, 0.018814087, 0.012710571, -0.0079956055, -0.02330017, 0.0065345764, 0.001036644, ...
http://arxiv.org/abs/2310.17639v1
In-Context Learning Dynamics with Random Binary Sequences
Eric J. Bigelow, Ekdeep Singh Lubana, Robert P. Dick, Hidenori Tanaka, Tomer D. Ullman
cs.AI, cs.CL, cs.LG
Large language models (LLMs) trained on huge corpora of text datasets demonstrate complex, emergent capabilities, achieving state-of-the-art performance on tasks they were not explicitly trained for. The precise nature of LLM capabilities is often mysterious, and different prompts can elicit different capabilities thro...
http://arxiv.org/pdf/2310.17639v1
2023-10-26T17:54:52
2023-10-26T17:54:52
{ "title": [ -0.0021877289, -0.00087976456, -0.035949707, -0.03326416, 0.013267517, -0.03111267, 0.0234375, -0.0041007996, -0.0005235672, 0.03817749, -0.033203125, 0.02420044, -0.015266418, 0.0027275085, -0.04019165, -0.001783371, 0.0062026978, 0...
http://arxiv.org/abs/2310.17631v1
JudgeLM: Fine-tuned Large Language Models are Scalable Judges
Lianghui Zhu, Xinggang Wang, Xinlong Wang
cs.CL, cs.AI
Evaluating Large Language Models (LLMs) in open-ended scenarios is challenging because existing benchmarks and metrics can not measure them comprehensively. To address this problem, we propose to fine-tune LLMs as scalable judges (JudgeLM) to evaluate LLMs efficiently and effectively in open-ended benchmarks. We first ...
http://arxiv.org/pdf/2310.17631v1
2023-10-26T17:48:58
2023-10-26T17:48:58
{ "title": [ -0.006866455, -0.02947998, -0.038146973, -0.103637695, -0.005722046, 0.0037708282, -0.060150146, 0.031951904, -0.0031661987, 0.024749756, -0.02229309, -0.009277344, 0.01576233, -0.019729614, -0.039855957, 0.0037269592, -0.01576233, 0...
http://arxiv.org/abs/2310.17630v1
InstOptima: Evolutionary Multi-objective Instruction Optimization via Large Language Model-based Instruction Operators
Heng Yang, Ke Li
cs.CL
Instruction-based language modeling has received significant attention in pretrained language models. However, the efficiency of instruction engineering remains low and hinders the development of instruction studies. Recent studies have focused on automating instruction generation, but they primarily aim to improve per...
http://arxiv.org/pdf/2310.17630v1
2023-10-26T17:48:45
2023-10-26T17:48:45
{ "title": [ -0.06008911, -0.019424438, -0.0029964447, -0.017822266, -0.02067566, 0.0138168335, -0.034820557, -0.020950317, 0.0046920776, 0.029632568, 0.000035643578, 0.0016975403, 0.0005373955, -0.025680542, -0.028579712, 0.012329102, -0.021759033, ...
http://arxiv.org/abs/2310.17623v1
Proving Test Set Contamination in Black Box Language Models
Yonatan Oren, Nicole Meister, Niladri Chatterji, Faisal Ladhak, Tatsunori B. Hashimoto
cs.CL, cs.LG
Large language models are trained on vast amounts of internet data, prompting concerns and speculation that they have memorized public benchmarks. Going from speculation to proof of contamination is challenging, as the pretraining data used by proprietary models are often not publicly accessible. We show that it is pos...
http://arxiv.org/pdf/2310.17623v1
2023-10-26T17:43:13
2023-10-26T17:43:13
{ "title": [ -0.008926392, 0.00084781647, -0.04345703, -0.10839844, 0.0029087067, 0.015571594, -0.088256836, 0.011741638, -0.016189575, 0.029830933, -0.035858154, 0.011398315, -0.021194458, 0.0050086975, -0.05569458, -0.023406982, -0.038146973, 0...
http://arxiv.org/abs/2310.17611v1
Uncovering Meanings of Embeddings via Partial Orthogonality
Yibo Jiang, Bryon Aragam, Victor Veitch
cs.LG, cs.CL, stat.ML
Machine learning tools often rely on embedding text as vectors of real numbers. In this paper, we study how the semantic structure of language is encoded in the algebraic structure of such embeddings. Specifically, we look at a notion of ``semantic independence'' capturing the idea that, e.g., ``eggplant'' and ``tomato...
http://arxiv.org/pdf/2310.17611v1
2023-10-26T17:34:32
2023-10-26T17:34:32
{ "title": [ -0.024093628, 0.007247925, -0.04498291, -0.028533936, 0.0074691772, -0.012863159, -0.023529053, -0.03036499, 0.0023860931, 0.0055351257, 0.011154175, 0.013145447, -0.0052108765, -0.040252686, -0.0033683777, 0.009902954, 0.006996155, ...
http://arxiv.org/abs/2310.17609v1
LeCaRDv2: A Large-Scale Chinese Legal Case Retrieval Dataset
Haitao Li, Yunqiu Shao, Yueyue Wu, Qingyao Ai, Yixiao Ma, Yiqun Liu
cs.CL, cs.IR
As an important component of intelligent legal systems, legal case retrieval plays a critical role in ensuring judicial justice and fairness. However, the development of legal case retrieval technologies in the Chinese legal system is restricted by three problems in existing datasets: limited data size, narrow definiti...
http://arxiv.org/pdf/2310.17609v1
2023-10-26T17:32:55
2023-10-26T17:32:55
{ "title": [ -0.01776123, -0.011390686, -0.038482666, -0.048614502, 0.012077332, 0.0010633469, -0.031951904, 0.003944397, 0.027236938, 0.035064697, -0.03933716, -0.0064926147, 0.038208008, -0.0045928955, -0.053985596, 0.014022827, -0.017456055, 0...
http://arxiv.org/abs/2310.17606v1
Using State-of-the-Art Speech Models to Evaluate Oral Reading Fluency in Ghana
Owen Henkel, Hannah Horne-Robinson, Libby Hills, Bill Roberts, Joshua McGrane
cs.CL, cs.AI
This paper reports on a set of three recent experiments utilizing large-scale speech models to evaluate the oral reading fluency (ORF) of students in Ghana. While ORF is a well-established measure of foundational literacy, assessing it typically requires one-on-one sessions between a student and a trained evaluator, a ...
http://arxiv.org/pdf/2310.17606v1
2023-10-26T17:30:13
2023-10-26T17:30:13
{ "title": [ -0.0025482178, -0.0026073456, -0.015197754, -0.030380249, -0.03326416, -0.00024461746, -0.039916992, -0.013847351, -0.020736694, 0.03265381, -0.0010738373, 0.008255005, 0.009017944, -0.03265381, -0.04675293, 0.005126953, -0.024993896, ...
http://arxiv.org/abs/2310.17591v1
Lil-Bevo: Explorations of Strategies for Training Language Models in More Humanlike Ways
Venkata S Govindarajan, Juan Diego Rodriguez, Kaj Bostrom, Kyle Mahowald
cs.CL
We present Lil-Bevo, our submission to the BabyLM Challenge. We pretrained our masked language models with three ingredients: an initial pretraining with music data, training on shorter sequences before training on longer ones, and masking specific tokens to target some of the BLiMP subtasks. Overall, our baseline mode...
http://arxiv.org/pdf/2310.17591v1
2023-10-26T17:13:07
2023-10-26T17:13:07
{ "title": [ -0.022750854, -0.021194458, -0.03488159, -0.041931152, -0.0056152344, -0.010345459, -0.040527344, -0.005996704, -0.0037822723, 0.03829956, -0.019241333, 0.02935791, 0.013801575, -0.015266418, -0.04309082, 0.015029907, -0.022125244, 0...
http://arxiv.org/abs/2310.17589v1
An Open Source Data Contamination Report for Llama Series Models
Yucheng Li
cs.CL, cs.AI
Data contamination in language model evaluation is increasingly prevalent as the popularity of large language models. It allows models to "cheat" via memorisation instead of displaying true capabilities. Therefore, contamination analysis has became an crucial part of reliable model evaluation to validate results. Howev...
http://arxiv.org/pdf/2310.17589v1
2023-10-26T17:11:42
2023-10-26T17:11:42
{ "title": [ 0.016708374, 0.036376953, -0.05038452, -0.078186035, -0.00073575974, -0.020004272, -0.07019043, -0.009429932, -0.04852295, 0.049682617, -0.0042419434, -0.0038375854, 0.010498047, -0.0513916, 0.015029907, -0.0037403107, 0.0058059692, ...
http://arxiv.org/abs/2310.17588v1
PAC-tuning:Fine-tuning Pretrained Language Models with PAC-driven Perturbed Gradient Descent
Guangliang Liu, Zhiyu Xue, Xitong Zhang, Kristen Marie Johnson, Rongrong Wang
cs.LG, cs.CL
Fine-tuning pretrained language models (PLMs) for downstream tasks is a large-scale optimization problem, in which the choice of the training algorithm critically determines how well the trained model can generalize to unseen test data, especially in the context of few-shot learning. To achieve good generalization perf...
http://arxiv.org/pdf/2310.17588v1
2023-10-26T17:09:13
2023-10-26T17:09:13
{ "title": [ -0.02330017, -0.0178833, -0.024963379, -0.07409668, -0.01940918, 0.004878998, -0.07183838, 0.0019350052, -0.010284424, 0.021514893, -0.015838623, 0.013267517, -0.0046958923, -0.024475098, -0.03201294, -0.006225586, -0.0067596436, 0.0...
http://arxiv.org/abs/2310.17586v1
Global Voices, Local Biases: Socio-Cultural Prejudices across Languages
Anjishnu Mukherjee, Chahat Raj, Ziwei Zhu, Antonios Anastasopoulos
cs.CL
Human biases are ubiquitous but not uniform: disparities exist across linguistic, cultural, and societal borders. As large amounts of recent literature suggest, language models (LMs) trained on human data can reflect and often amplify the effects of these social biases. However, the vast majority of existing studies on...
http://arxiv.org/pdf/2310.17586v1
2023-10-26T17:07:50
2023-10-26T17:07:50
{ "title": [ -0.015541077, -0.005592346, -0.030975342, -0.045410156, 0.0096588135, -0.019088745, -0.014732361, -0.0049362183, -0.0053253174, 0.03100586, -0.015525818, -0.0036697388, 0.009742737, -0.027542114, -0.033569336, -0.00034165382, -0.01101684...
http://arxiv.org/abs/2310.18373v1
Can LLMs Grade Short-answer Reading Comprehension Questions : Foundational Literacy Assessment in LMICs
Owen Henkel, Libby Hills, Bill Roberts, Joshua McGrane
cs.CL, cs.AI
This paper presents emerging evidence of using generative large language models (i.e., GPT-4) to reliably evaluate short-answer reading comprehension questions. Specifically, we explore how various configurations of generative (LLMs) are able to evaluate student responses from a new dataset, drawn from a battery of rea...
http://arxiv.org/pdf/2310.18373v1
2023-10-26T17:05:40
2023-10-26T17:05:40
{ "title": [ -0.062042236, 0.0044898987, -0.021560669, 0.02279663, -0.05407715, 0.005672455, -0.059448242, -0.031707764, -0.011894226, 0.004184723, -0.048675537, -0.006549835, -0.020751953, -0.038848877, -0.0063972473, -0.05810547, -0.01108551, 0...
http://arxiv.org/abs/2310.17576v1
1D-Touch: NLP-Assisted Coarse Text Selection via a Semi-Direct Gesture
Peiling Jiang, Li Feng, Fuling Sun, Parakrant Sarkar, Haijun Xia, Can Liu
cs.HC, cs.CL
Existing text selection techniques on touchscreen focus on improving the control for moving the carets. Coarse-grained text selection on word and phrase levels has not received much support beyond word-snapping and entity recognition. We introduce 1D-Touch, a novel text selection method that complements the carets-base...
http://arxiv.org/pdf/2310.17576v1
2023-10-26T17:01:22
2023-10-26T17:01:22
{ "title": [ -0.03253174, -0.012886047, -0.0368042, -0.032043457, -0.012397766, 0.0002708435, -0.024337769, -0.0034065247, 0.015609741, 0.024337769, -0.055267334, 0.023040771, 0.009063721, -0.0019817352, -0.025131226, -0.01651001, -0.032470703, 0...
http://arxiv.org/abs/2310.17570v1
DiffS2UT: A Semantic Preserving Diffusion Model for Textless Direct Speech-to-Speech Translation
Yongxin Zhu, Zhujin Gao, Xinyuan Zhou, Zhongyi Ye, Linli Xu
cs.CL
While Diffusion Generative Models have achieved great success on image generation tasks, how to efficiently and effectively incorporate them into speech generation especially translation tasks remains a non-trivial problem. Specifically, due to the low information density of speech data, the transformed discrete speech...
http://arxiv.org/pdf/2310.17570v1
2023-10-26T16:58:14
2023-10-26T16:58:14
{ "title": [ -0.020812988, -0.010482788, -0.032958984, -0.059906006, 0.0055885315, 0.00970459, -0.027740479, 0.02357483, 0.0138549805, 0.046539307, -0.008972168, 0.0034236908, 0.033813477, -0.03692627, -0.04574585, 0.0146484375, -0.006454468, 0.0...
http://arxiv.org/abs/2310.17568v1
Navigating to Success in Multi-Modal Human-Robot Collaboration: Analysis and Corpus Release
Stephanie M. Lukin, Kimberly A. Pollard, Claire Bonial, Taylor Hudson, Ron Arstein, Clare Voss, David Traum
cs.HC, cs.CL, cs.RO
Human-guided robotic exploration is a useful approach to gathering information at remote locations, especially those that might be too risky, inhospitable, or inaccessible for humans. Maintaining common ground between the remotely-located partners is a challenge, one that can be facilitated by multi-modal communication...
http://arxiv.org/pdf/2310.17568v1
2023-10-26T16:56:01
2023-10-26T16:56:01
{ "title": [ -0.02418518, -0.002090454, -0.027023315, -0.012199402, -0.0011463165, -0.020553589, -0.022857666, -0.027633667, -0.035491943, 0.048431396, -0.013290405, 0.013183594, -0.0019340515, 0.0040779114, -0.038116455, -0.022598267, -0.012664795, ...
http://arxiv.org/abs/2310.17567v1
Skill-Mix: a Flexible and Expandable Family of Evaluations for AI models
Dingli Yu, Simran Kaur, Arushi Gupta, Jonah Brown-Cohen, Anirudh Goyal, Sanjeev Arora
cs.CL, cs.AI, cs.LG, cs.NE
With LLMs shifting their role from statistical modeling of language to serving as general-purpose AI agents, how should LLM evaluations change? Arguably, a key ability of an AI agent is to flexibly combine, as needed, the basic skills it has learned. The capability to combine skills plays an important role in (human) p...
http://arxiv.org/pdf/2310.17567v1
2023-10-26T16:55:05
2023-10-26T16:55:05
{ "title": [ -0.027786255, 0.0045433044, -0.03567505, -0.01625061, 0.016738892, -0.018569946, -0.002456665, -0.04522705, 0.00006610155, 0.0115356445, -0.020812988, 0.009857178, -0.0178833, -0.028076172, -0.035217285, 0.0041999817, -0.014190674, 0...
http://arxiv.org/abs/2310.17558v1
Towards Matching Phones and Speech Representations
Gene-Ping Yang, Hao Tang
cs.CL, cs.LG, cs.SD, eess.AS
Learning phone types from phone instances has been a long-standing problem, while still being open. In this work, we revisit this problem in the context of self-supervised learning, and pose it as the problem of matching cluster centroids to phone embeddings. We study two key properties that enable matching, namely, wh...
http://arxiv.org/pdf/2310.17558v1
2023-10-26T16:47:52
2023-10-26T16:47:52
{ "title": [ 0.0013618469, 0.0099487305, -0.029129028, -0.06677246, 0.010787964, 0.0039405823, -0.04849243, 0.0029010773, 0.0069236755, 0.04650879, -0.014518738, -0.016998291, 0.012214661, -0.021774292, -0.04232788, 0.011726379, -0.009681702, 0.0...
http://arxiv.org/abs/2310.17551v1
Unpacking the Ethical Value Alignment in Big Models
Xiaoyuan Yi, Jing Yao, Xiting Wang, Xing Xie
cs.CY, cs.AI, cs.CL
Big models have greatly advanced AI's ability to understand, generate, and manipulate information and content, enabling numerous applications. However, as these models become increasingly integrated into everyday life, their inherent ethical values and potential biases pose unforeseen risks to society. This paper provi...
http://arxiv.org/pdf/2310.17551v1
2023-10-26T16:45:40
2023-10-26T16:45:40
{ "title": [ -0.020339966, 0.0034599304, -0.061950684, -0.039123535, 0.0030498505, -0.025543213, -0.0395813, -0.008865356, -0.013191223, 0.027130127, -0.022079468, -0.0107040405, -0.024520874, -0.025909424, -0.031173706, -0.015151978, 0.00067806244, ...
http://arxiv.org/abs/2310.17530v1
Evaluating Bias and Fairness in Gender-Neutral Pretrained Vision-and-Language Models
Laura Cabello, Emanuele Bugliarello, Stephanie Brandl, Desmond Elliott
cs.CV, cs.CL, cs.LG
Pretrained machine learning models are known to perpetuate and even amplify existing biases in data, which can result in unfair outcomes that ultimately impact user experience. Therefore, it is crucial to understand the mechanisms behind those prejudicial biases to ensure that model performance does not result in discr...
http://arxiv.org/pdf/2310.17530v1
2023-10-26T16:19:19
2023-10-26T16:19:19
{ "title": [ -0.017196655, -0.008018494, -0.047424316, -0.020874023, -0.007858276, -0.031036377, -0.018447876, -0.013221741, 0.0032806396, 0.04373169, -0.040649414, -0.010284424, 0.0309906, -0.0036830902, -0.05596924, -0.021606445, -0.016952515, ...
http://arxiv.org/abs/2310.17526v2
Can large language models replace humans in the systematic review process? Evaluating GPT-4's efficacy in screening and extracting data from peer-reviewed and grey literature in multiple languages
Qusai Khraisha, Sophie Put, Johanna Kappenberg, Azza Warraitch, Kristin Hadfield
cs.CL, cs.AI, cs.LG
Systematic reviews are vital for guiding practice, research, and policy, yet they are often slow and labour-intensive. Large language models (LLMs) could offer a way to speed up and automate systematic reviews, but their performance in such tasks has not been comprehensively evaluated against humans, and no study has t...
http://arxiv.org/pdf/2310.17526v2
2023-10-27T12:14:27
2023-10-26T16:18:30
{ "title": [ -0.02130127, 0.0137786865, 0.011672974, 0.0062713623, -0.048797607, 0.015670776, -0.055511475, -0.052001953, -0.017501831, 0.018432617, -0.0020942688, 0.010124207, -0.03894043, -0.02468872, -0.0036640167, -0.028259277, -0.010887146, ...
http://arxiv.org/abs/2310.17514v1
The Validity of Evaluation Results: Assessing Concurrence Across Compositionality Benchmarks
Kaiser Sun, Adina Williams, Dieuwke Hupkes
cs.CL
NLP models have progressed drastically in recent years, according to numerous datasets proposed to evaluate performance. Questions remain, however, about how particular dataset design choices may impact the conclusions we draw about model capabilities. In this work, we investigate this question in the domain of composi...
http://arxiv.org/pdf/2310.17514v1
2023-10-26T16:11:04
2023-10-26T16:11:04
{ "title": [ -0.010261536, -0.022781372, -0.036834717, -0.07232666, -0.003709793, 0.009185791, -0.05947876, -0.003396988, -0.000166893, 0.026550293, -0.0044784546, -0.004058838, -0.032928467, -0.031707764, -0.068725586, -0.009490967, -0.037322998, ...
http://arxiv.org/abs/2310.17513v2
The Expressive Power of Low-Rank Adaptation
Yuchen Zeng, Kangwook Lee
cs.LG, cs.AI, cs.CL, stat.ML
Low-Rank Adaptation (LoRA), a parameter-efficient fine-tuning method that leverages low-rank adaptation of weight matrices, has emerged as a prevalent technique for fine-tuning pre-trained models such as large language models and diffusion models. Despite its huge success in practice, the theoretical underpinnings of L...
http://arxiv.org/pdf/2310.17513v2
2023-10-27T02:36:44
2023-10-26T16:08:33
{ "title": [ -0.015945435, -0.011726379, -0.021499634, -0.015670776, -0.0053138733, -0.016784668, 0.0007748604, -0.037322998, 0.0072364807, 0.024520874, -0.0059013367, -0.01663208, -0.008300781, -0.03137207, -0.038757324, 0.0075187683, 0.024597168, ...
http://arxiv.org/abs/2310.17512v1
CompeteAI: Understanding the Competition Behaviors in Large Language Model-based Agents
Qinlin Zhao, Jindong Wang, Yixuan Zhang, Yiqiao Jin, Kaijie Zhu, Hao Chen, Xing Xie
cs.AI, cs.CL, cs.HC, cs.MA
Large language models (LLMs) have been widely used as agents to complete different tasks, such as personal assistance or event planning. While most work has focused on cooperation and collaboration between agents, little work explores competition, another important mechanism that fosters the development of society and ...
http://arxiv.org/pdf/2310.17512v1
2023-10-26T16:06:20
2023-10-26T16:06:20
{ "title": [ -0.028152466, -0.013946533, -0.021850586, -0.037902832, 0.006587982, -0.004722595, -0.016342163, 0.00041913986, -0.017196655, 0.03427124, -0.020462036, 0.012535095, -0.0011472702, -0.008628845, -0.039031982, -0.004119873, -0.021652222, ...
http://arxiv.org/abs/2310.17499v1
The IMS Toucan System for the Blizzard Challenge 2023
Florian Lux, Julia Koch, Sarina Meyer, Thomas Bott, Nadja Schauffler, Pavel Denisov, Antje Schweitzer, Ngoc Thang Vu
cs.CL, cs.LG, eess.AS
For our contribution to the Blizzard Challenge 2023, we improved on the system we submitted to the Blizzard Challenge 2021. Our approach entails a rule-based text-to-phoneme processing system that includes rule-based disambiguation of homographs in the French language. It then transforms the phonemes to spectrograms as...
http://arxiv.org/pdf/2310.17499v1
2023-10-26T15:53:29
2023-10-26T15:53:29
{ "title": [ -0.03213501, -0.036621094, -0.01574707, -0.023773193, -0.044647217, 0.018707275, -0.033691406, 0.018478394, -0.005039215, 0.04260254, -0.01737976, 0.010566711, 0.02116394, -0.03704834, 0.035949707, -0.021697998, 0.031280518, -0.02061...
http://arxiv.org/abs/2310.17490v3
Improving Zero-shot Reader by Reducing Distractions from Irrelevant Documents in Open-Domain Question Answering
Sukmin Cho, Jeongyeon Seo, Soyeong Jeong, Jong C. Park
cs.CL, cs.AI
Large language models (LLMs) enable zero-shot approaches in open-domain question answering (ODQA), yet with limited advancements as the reader is compared to the retriever. This study aims at the feasibility of a zero-shot reader that addresses the challenges of computational cost and the need for labeled data. We find...
http://arxiv.org/pdf/2310.17490v3
2023-11-14T06:49:33
2023-10-26T15:45:12
{ "title": [ 0.023239136, -0.0056877136, -0.013114929, -0.047668457, -0.0107803345, -0.026397705, -0.010749817, -0.00818634, 0.02053833, 0.05407715, -0.010536194, -0.0030002594, 0.03262329, -0.0026721954, -0.06317139, 0.009155273, 0.006023407, 0....
http://arxiv.org/abs/2310.17488v2
LightLM: A Lightweight Deep and Narrow Language Model for Generative Recommendation
Kai Mei, Yongfeng Zhang
cs.IR, cs.CL
This paper presents LightLM, a lightweight Transformer-based language model for generative recommendation. While Transformer-based generative modeling has gained importance in various AI sub-fields such as NLP and vision, generative recommendation is still in its infancy due to its unique demand on personalized generat...
http://arxiv.org/pdf/2310.17488v2
2023-10-30T02:50:17
2023-10-26T15:44:57
{ "title": [ -0.0154418945, -0.02456665, -0.045837402, -0.059265137, -0.00793457, -0.0051879883, -0.045898438, 0.009994507, 0.0043640137, 0.026138306, -0.014442444, 0.02067566, -0.00048661232, -0.017044067, -0.031402588, 0.009544373, -0.014350891, ...
http://arxiv.org/abs/2310.17448v1
Dialect Adaptation and Data Augmentation for Low-Resource ASR: TalTech Systems for the MADASR 2023 Challenge
Tanel Alumäe, Jiaming Kong, Daniil Robnikov
cs.CL, eess.AS
This paper describes Tallinn University of Technology (TalTech) systems developed for the ASRU MADASR 2023 Challenge. The challenge focuses on automatic speech recognition of dialect-rich Indian languages with limited training audio and text data. TalTech participated in two tracks of the challenge: Track 1 that allowe...
http://arxiv.org/pdf/2310.17448v1
2023-10-26T14:57:08
2023-10-26T14:57:08
{ "title": [ -0.013153076, -0.018463135, -0.068115234, -0.06878662, -0.005580902, -0.010498047, -0.061767578, 0.029449463, -0.028030396, 0.06567383, 0.011131287, -0.0055122375, 0.0287323, -0.021530151, -0.04244995, 0.020751953, 0.018508911, 0.023...
http://arxiv.org/abs/2310.17428v1
''Fifty Shades of Bias'': Normative Ratings of Gender Bias in GPT Generated English Text
Rishav Hada, Agrima Seth, Harshita Diddee, Kalika Bali
cs.CL
Language serves as a powerful tool for the manifestation of societal belief systems. In doing so, it also perpetuates the prevalent biases in our society. Gender bias is one of the most pervasive biases in our society and is seen in online and offline discourses. With LLMs increasingly gaining human-like fluency in tex...
http://arxiv.org/pdf/2310.17428v1
2023-10-26T14:34:06
2023-10-26T14:34:06
{ "title": [ -0.012023926, 0.023651123, -0.027938843, 0.0046539307, 0.0029830933, -0.0345459, 0.02078247, -0.023605347, -0.01878357, 0.007160187, -0.05810547, 0.012336731, 0.028320312, -0.013710022, 0.0104522705, -0.02558899, -0.052368164, 0.0642...
http://arxiv.org/abs/2310.17415v1
PETA: Evaluating the Impact of Protein Transfer Learning with Sub-word Tokenization on Downstream Applications
Yang Tan, Mingchen Li, Pan Tan, Ziyi Zhou, Huiqun Yu, Guisheng Fan, Liang Hong
cs.CL, cs.AI, q-bio.BM
Large protein language models are adept at capturing the underlying evolutionary information in primary structures, offering significant practical value for protein engineering. Compared to natural language models, protein amino acid sequences have a smaller data volume and a limited combinatorial space. Choosing an ap...
http://arxiv.org/pdf/2310.17415v1
2023-10-26T14:20:44
2023-10-26T14:20:44
{ "title": [ -0.032165527, -0.015388489, -0.05114746, -0.035369873, -0.006160736, -0.016433716, -0.027755737, 0.006450653, -0.028335571, 0.031311035, -0.013664246, 0.019332886, -0.0025901794, -0.022705078, -0.014465332, 0.0010194778, -0.014854431, ...
http://arxiv.org/abs/2310.17413v1
Harnessing GPT-3.5-turbo for Rhetorical Role Prediction in Legal Cases
Anas Belfathi, Nicolas Hernandez, Laura Monceaux
cs.CL
We propose a comprehensive study of one-stage elicitation techniques for querying a large pre-trained generative transformer (GPT-3.5-turbo) in the rhetorical role prediction task of legal cases. This task is known as requiring textual context to be addressed. Our study explores strategies such as zero-few shots, task ...
http://arxiv.org/pdf/2310.17413v1
2023-10-26T14:19:48
2023-10-26T14:19:48
{ "title": [ 0.014633179, -0.0011672974, -0.02243042, -0.052703857, -0.016494751, 0.008522034, -0.056915283, -0.010986328, -0.018661499, 0.025222778, -0.021560669, 0.007858276, 0.007118225, -0.004272461, -0.039978027, -0.008583069, -0.0385437, 0....
http://arxiv.org/abs/2310.17408v2
Tackling the Matrix Multiplication Micro-kernel Generation with Exo
Adrián Castelló, Julian Bellavita, Grace Dinh, Yuka Ikarashi, Héctor Martínez
cs.MS, cs.CL, cs.PF
The optimization of the matrix multiplication (or GEMM) has been a need during the last decades. This operation is considered the flagship of current linear algebra libraries such as BLIS, OpenBLAS, or Intel OneAPI because of its widespread use in a large variety of scientific applications. The GEMM is usually implemen...
http://arxiv.org/pdf/2310.17408v2
2023-10-27T08:28:03
2023-10-26T14:09:57
{ "title": [ -0.046844482, -0.010902405, -0.036102295, -0.03741455, -0.015197754, -0.009140015, -0.028930664, 0.0003452301, -0.010032654, 0.0552063, 0.0042533875, -0.018249512, 0.024871826, -0.040802002, -0.040222168, 0.012313843, 0.014465332, 0....
http://arxiv.org/abs/2310.17407v1
Meaning and understanding in large language models
Vladimír Havlík
cs.CL
Can a machine understand the meanings of natural language? Recent developments in the generative large language models (LLMs) of artificial intelligence have led to the belief that traditional philosophical assumptions about machine understanding of language need to be revised. This article critically evaluates the pre...
http://arxiv.org/pdf/2310.17407v1
2023-10-26T14:06:14
2023-10-26T14:06:14
{ "title": [ -0.040222168, -0.013122559, -0.036376953, -0.0680542, 0.003396988, 0.007850647, -0.061157227, 0.0059661865, 0.020843506, 0.021331787, -0.014671326, -0.016845703, 0.011138916, -0.018936157, -0.05026245, 0.013381958, -0.012039185, 0.02...
http://arxiv.org/abs/2310.17389v1
ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversation
Zi Lin, Zihan Wang, Yongqi Tong, Yangkun Wang, Yuxin Guo, Yujia Wang, Jingbo Shang
cs.CL, cs.AI
Despite remarkable advances that large language models have achieved in chatbots, maintaining a non-toxic user-AI interactive environment has become increasingly critical nowadays. However, previous efforts in toxicity detection have been mostly based on benchmarks derived from social media content, leaving the unique ...
http://arxiv.org/pdf/2310.17389v1
2023-10-26T13:35:41
2023-10-26T13:35:41
{ "title": [ 0.021469116, -0.005268097, -0.023422241, -0.0569458, 0.023956299, -0.012817383, -0.0047912598, -0.015327454, 0.005622864, 0.029769897, -0.028671265, 0.028320312, 0.0053749084, -0.005558014, -0.036834717, -0.0047798157, -0.022506714, ...
http://arxiv.org/abs/2310.17372v1
Dialogue-based generation of self-driving simulation scenarios using Large Language Models
Antonio Valerio Miceli-Barone, Alex Lascarides, Craig Innes
cs.AI, cs.CL, cs.RO
Simulation is an invaluable tool for developing and evaluating controllers for self-driving cars. Current simulation frameworks are driven by highly-specialist domain specific languages, and so a natural language interface would greatly enhance usability. But there is often a gap, consisting of tacit assumptions the us...
http://arxiv.org/pdf/2310.17372v1
2023-10-26T13:07:01
2023-10-26T13:07:01
{ "title": [ -0.008033752, -0.018508911, -0.023117065, -0.042053223, 0.0047302246, -0.012786865, -0.017608643, -0.0017585754, -0.026535034, 0.029083252, -0.018005371, 0.020477295, -0.014404297, -0.015403748, -0.017150879, -0.0070228577, -0.0284729, ...
http://arxiv.org/abs/2310.17369v2
Language and Mental Health: Measures of Emotion Dynamics from Text as Linguistic Biosocial Markers
Daniela Teodorescu, Tiffany Cheng, Alona Fyshe, Saif M. Mohammad
cs.CL
Research in psychopathology has shown that, at an aggregate level, the patterns of emotional change over time -- emotion dynamics -- are indicators of one's mental health. One's patterns of emotion change have traditionally been determined through self-reports of emotions; however, there are known issues with accuracy,...
http://arxiv.org/pdf/2310.17369v2
2023-11-04T17:19:29
2023-10-26T13:00:26
{ "title": [ -0.028518677, 0.002023697, -0.027770996, -0.041534424, 0.0049972534, -0.00819397, -0.03363037, -0.016113281, 0.007472992, 0.02748108, -0.0048065186, 0.015380859, 0.0009994507, -0.029281616, -0.040039062, 0.009239197, 0.0058288574, 0....
http://arxiv.org/abs/2310.17353v1
Cultural Adaptation of Recipes
Yong Cao, Yova Kementchedjhieva, Ruixiang Cui, Antonia Karamolegkou, Li Zhou, Megan Dare, Lucia Donatelli, Daniel Hershcovich
cs.CL, cs.AI
Building upon the considerable advances in Large Language Models (LLMs), we are now equipped to address more sophisticated tasks demanding a nuanced understanding of cross-cultural contexts. A key example is recipe adaptation, which goes beyond simple translation to include a grasp of ingredients, culinary techniques, ...
http://arxiv.org/pdf/2310.17353v1
2023-10-26T12:39:20
2023-10-26T12:39:20
{ "title": [ 0.015235901, -0.0064582825, 0.008720398, -0.012207031, -0.05203247, -0.02633667, -0.014091492, -0.00932312, -0.033996582, 0.070739746, 0.012802124, -0.016448975, 0.02293396, -0.036987305, 0.01940918, 0.01525116, 0.050811768, -0.03015...
http://arxiv.org/abs/2310.17342v1
ACT-SQL: In-Context Learning for Text-to-SQL with Automatically-Generated Chain-of-Thought
Hanchong Zhang, Ruisheng Cao, Lu Chen, Hongshen Xu, Kai Yu
cs.CL
Recently Large Language Models (LLMs) have been proven to have strong abilities in various domains and tasks. We study the problem of prompt designing in the text-to-SQL task and attempt to improve the LLMs' reasoning ability when generating SQL queries. Besides the trivial few-shot in-context learning setting, we desi...
http://arxiv.org/pdf/2310.17342v1
2023-10-26T12:16:25
2023-10-26T12:16:25
{ "title": [ -0.0018625259, -0.012084961, -0.038360596, -0.060760498, 0.011940002, 0.0009303093, -0.010047913, 0.021362305, 0.03643799, 0.035217285, -0.061187744, 0.019348145, 0.008422852, -0.0013389587, -0.048736572, -0.018493652, -0.0058250427, ...
http://arxiv.org/abs/2310.17333v1
Arabic Fine-Grained Entity Recognition
Haneen Liqreina, Mustafa Jarrar, Mohammed Khalilia, Ahmed Oumar El-Shangiti, Muhammad AbdulMageed
cs.CL
Traditional NER systems are typically trained to recognize coarse-grained entities, and less attention is given to classifying entities into a hierarchy of fine-grained lower-level subtypes. This article aims to advance Arabic NER with fine-grained entities. We chose to extend Wojood (an open-source Nested Arabic Named...
http://arxiv.org/pdf/2310.17333v1
2023-10-26T11:59:45
2023-10-26T11:59:45
{ "title": [ -0.008460999, -0.0043258667, -0.03805542, -0.07434082, -0.00085783005, -0.001707077, -0.03152466, 0.026550293, -0.012573242, 0.040252686, -0.011871338, 0.0059051514, 0.008255005, -0.032073975, -0.03250122, 0.0044021606, 0.01852417, 0...
http://arxiv.org/abs/2310.17315v1
Nabra: Syrian Arabic Dialects with Morphological Annotations
Amal Nayouf, Tymaa Hammouda, Mustafa Jarrar, Fadi Zaraket, Mohamad-Bassam Kurdy
cs.CL
This paper presents Nabra, a corpora of Syrian Arabic dialects with morphological annotations. A team of Syrian natives collected more than 6K sentences containing about 60K words from several sources including social media posts, scripts of movies and series, lyrics of songs and local proverbs to build Nabra. Nabra co...
http://arxiv.org/pdf/2310.17315v1
2023-10-26T11:23:05
2023-10-26T11:23:05
{ "title": [ -0.020858765, 0.0032444, -0.07141113, -0.055511475, -0.0256958, 0.0041046143, -0.06506348, 0.026443481, -0.047088623, 0.031982422, 0.018463135, -0.00081443787, 0.03050232, -0.07208252, 0.025619507, 0.0007548332, 0.0009536743, 0.04202...
http://arxiv.org/abs/2310.17312v1
An Ensemble Method Based on the Combination of Transformers with Convolutional Neural Networks to Detect Artificially Generated Text
Vijini Liyanage, Davide Buscaldi
cs.CL
Thanks to the state-of-the-art Large Language Models (LLMs), language generation has reached outstanding levels. These models are capable of generating high quality content, thus making it a challenging task to detect generated text from human-written content. Despite the advantages provided by Natural Language Generat...
http://arxiv.org/pdf/2310.17312v1
2023-10-26T11:17:03
2023-10-26T11:17:03
{ "title": [ -0.0011129379, 0.015777588, -0.028747559, -0.039093018, 0.012886047, -0.018478394, -0.032226562, -0.022766113, -0.0104904175, 0.035705566, -0.02079773, 0.005138397, 0.0011825562, -0.016311646, -0.031951904, -0.009094238, -0.013908386, ...
http://arxiv.org/abs/2310.18371v1
In-Context Ability Transfer for Question Decomposition in Complex QA
Venktesh V, Sourangshu Bhattacharya, Avishek Anand
cs.CL, cs.AI
Answering complex questions is a challenging task that requires question decomposition and multistep reasoning for arriving at the solution. While existing supervised and unsupervised approaches are specialized to a certain task and involve training, recently proposed prompt-based approaches offer generalizable solutio...
http://arxiv.org/pdf/2310.18371v1
2023-10-26T11:11:07
2023-10-26T11:11:07
{ "title": [ 0.014343262, -0.0054130554, -0.024261475, -0.044433594, -0.005558014, -0.013679504, 0.019836426, 0.003320694, 0.015434265, 0.037078857, -0.007926941, 0.017868042, 0.02180481, -0.018188477, -0.03274536, 0.011634827, -0.0010738373, 0.0...
http://arxiv.org/abs/2310.17680v3
CodeFusion: A Pre-trained Diffusion Model for Code Generation
Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, Gust Verbruggen
cs.SE, cs.AI, cs.CL, cs.PL
Imagine a developer who can only change their last line of code, how often would they have to start writing a function from scratch before it is correct? Auto-regressive models for code generation from natural language have a similar limitation: they do not easily allow reconsidering earlier tokens generated. We introd...
http://arxiv.org/pdf/2310.17680v3
2023-11-01T17:30:47
2023-10-26T11:06:15
{ "title": [ -0.039855957, -0.03161621, -0.0435791, -0.04623413, 0.0027427673, 0.0021686554, -0.027877808, 0.014099121, 0.013763428, 0.016998291, -0.008682251, 0.013023376, -0.018920898, -0.024551392, -0.028884888, 0.0047836304, -0.016174316, 0.0...
http://arxiv.org/abs/2310.17306v3
FormaT5: Abstention and Examples for Conditional Table Formatting with Natural Language
Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, Elnaz Nouri, Mohammad Raza, Gust Verbruggen
cs.AI, cs.CL, cs.DB, cs.PL
Formatting is an important property in tables for visualization, presentation, and analysis. Spreadsheet software allows users to automatically format their tables by writing data-dependent conditional formatting (CF) rules. Writing such rules is often challenging for users as it requires them to understand and impleme...
http://arxiv.org/pdf/2310.17306v3
2023-11-01T17:31:30
2023-10-26T11:05:15
{ "title": [ 0.010848999, 0.0026988983, -0.023742676, -0.046295166, -0.01259613, -0.0025463104, -0.054992676, -0.010894775, -0.009422302, 0.03643799, -0.050964355, -0.01084137, 0.046844482, -0.033966064, 0.004585266, -0.03253174, 0.010154724, 0.0...
http://arxiv.org/abs/2310.17300v1
Comparing Photorealistic and Animated Embodied Conversational Agents in Serious Games: An Empirical Study on User Experience
Danai Korre
cs.HC, cs.AI, cs.CL, cs.MM, I.2.1; H.5.2; K.3.1
Embodied conversational agents (ECAs) are paradigms of conversational user interfaces in the form of embodied characters. While ECAs offer various manipulable features, this paper focuses on a study conducted to explore two distinct levels of presentation realism. The two agent versions are photorealistic and animated....
http://arxiv.org/pdf/2310.17300v1
2023-10-26T10:45:26
2023-10-26T10:45:26
{ "title": [ 0.014328003, -0.0007586479, -0.0058174133, 0.022735596, -0.014808655, -0.03186035, 0.020263672, -0.042419434, 0.004184723, 0.04058838, -0.017913818, 0.006778717, -0.0011816025, -0.019088745, -0.05529785, -0.01197052, -0.028961182, -0...
http://arxiv.org/abs/2310.17284v1
Learning to Abstract with Nonparametric Variational Information Bottleneck
Melika Behjati, Fabio Fehr, James Henderson
cs.CL
Learned representations at the level of characters, sub-words, words and sentences, have each contributed to advances in understanding different NLP tasks and linguistic phenomena. However, learning textual embeddings is costly as they are tokenization specific and require different models to be trained for each level ...
http://arxiv.org/pdf/2310.17284v1
2023-10-26T10:04:31
2023-10-26T10:04:31
{ "title": [ -0.027526855, 0.02734375, -0.061157227, -0.036895752, 0.0011844635, -0.014808655, -0.04824829, -0.019500732, 0.010505676, 0.021438599, -0.031234741, 0.016067505, -0.036956787, -0.003660202, -0.051116943, -0.01524353, -0.004219055, 0....
http://arxiv.org/abs/2310.17279v1
Automatic Logical Forms improve fidelity in Table-to-Text generation
Iñigo Alonso, Eneko Agirre
cs.CL
Table-to-text systems generate natural language statements from structured data like tables. While end-to-end techniques suffer from low factual correctness (fidelity), a previous study reported gains when using manual logical forms (LF) that represent the selected content and the semantics of the target text. Given th...
http://arxiv.org/pdf/2310.17279v1
2023-10-26T10:00:24
2023-10-26T10:00:24
{ "title": [ -0.014656067, 0.010803223, -0.046203613, -0.047943115, -0.01802063, -0.011489868, -0.06976318, -0.02027893, -0.0029468536, 0.076171875, -0.026794434, 0.02406311, 0.055236816, -0.010635376, -0.031951904, 0.027008057, 0.0046806335, 0.0...
http://arxiv.org/abs/2310.17271v1
Understanding the Role of Input Token Characters in Language Models: How Does Information Loss Affect Performance?
Ahmed Alajrami, Katerina Margatina, Nikolaos Aletras
cs.CL
Understanding how and what pre-trained language models (PLMs) learn about language is an open challenge in natural language processing. Previous work has focused on identifying whether they capture semantic and syntactic information, and how the data or the pre-training objective affects their performance. However, to ...
http://arxiv.org/pdf/2310.17271v1
2023-10-26T09:47:50
2023-10-26T09:47:50
{ "title": [ -0.04244995, -0.012992859, -0.022827148, -0.0115356445, -0.04397583, -0.0079193115, -0.04638672, -0.004348755, -0.008010864, 0.03778076, -0.01197052, -0.016937256, 0.020614624, -0.028625488, -0.039123535, -0.0073165894, -0.018951416, ...
http://arxiv.org/abs/2310.17238v1
Joint Entity and Relation Extraction with Span Pruning and Hypergraph Neural Networks
Zhaohui Yan, Songlin Yang, Wei Liu, Kewei Tu
cs.CL, cs.AI, cs.LG
Entity and Relation Extraction (ERE) is an important task in information extraction. Recent marker-based pipeline models achieve state-of-the-art performance, but still suffer from the error propagation issue. Also, most of current ERE models do not take into account higher-order interactions between multiple entities ...
http://arxiv.org/pdf/2310.17238v1
2023-10-26T08:36:39
2023-10-26T08:36:39
{ "title": [ -0.013534546, 0.002046585, -0.06311035, -0.07897949, 0.027664185, -0.005207062, -0.044677734, 0.010925293, 0.025024414, 0.014465332, -0.0068588257, -0.0008649826, 0.006439209, -0.01878357, -0.040283203, 0.00014543533, -0.012573242, 0...
http://arxiv.org/abs/2310.17233v1
EMMA-X: An EM-like Multilingual Pre-training Algorithm for Cross-lingual Representation Learning
Ping Guo, Xiangpeng Wei, Yue Hu, Baosong Yang, Dayiheng Liu, Fei Huang, Jun Xie
cs.CL
Expressing universal semantics common to all languages is helpful in understanding the meanings of complex and culture-specific sentences. The research theme underlying this scenario focuses on learning universal representations across languages with the usage of massive parallel corpora. However, due to the sparsity a...
http://arxiv.org/pdf/2310.17233v1
2023-10-26T08:31:00
2023-10-26T08:31:00
{ "title": [ 0.00044894218, -0.0052757263, -0.022979736, -0.06329346, 0.007762909, 0.0034046173, -0.05142212, -0.017700195, 0.0058288574, 0.045318604, -0.010185242, 0.020004272, 0.0289917, -0.01739502, -0.017059326, 0.026016235, -0.0039711, 0.057...
http://arxiv.org/abs/2310.17230v1
Codebook Features: Sparse and Discrete Interpretability for Neural Networks
Alex Tamkin, Mohammad Taufeeque, Noah D. Goodman
cs.LG, cs.CL
Understanding neural networks is challenging in part because of the dense, continuous nature of their hidden states. We explore whether we can train neural networks to have hidden states that are sparse, discrete, and more interpretable by quantizing their continuous features into what we call codebook features. Codebo...
http://arxiv.org/pdf/2310.17230v1
2023-10-26T08:28:48
2023-10-26T08:28:48
{ "title": [ -0.045837402, 0.0063438416, -0.036132812, -0.034210205, 0.017120361, 0.009498596, -0.042144775, -0.024627686, 0.029159546, 0.005710602, -0.01348114, -0.004878998, -0.029693604, -0.017242432, -0.062164307, 0.0025577545, -0.0022220612, ...
http://arxiv.org/abs/2310.17228v2
TST$^\mathrm{R}$: Target Similarity Tuning Meets the Real World
Anirudh Khatry, Sumit Gulwani, Priyanshu Gupta, Vu Le, Ananya Singha, Mukul Singh, Gust Verbruggen
cs.AI, cs.CL, cs.SE
Target similarity tuning (TST) is a method of selecting relevant examples in natural language (NL) to code generation through large language models (LLMs) to improve performance. Its goal is to adapt a sentence embedding model to have the similarity between two NL inputs match the similarity between their associated co...
http://arxiv.org/pdf/2310.17228v2
2023-10-28T08:24:06
2023-10-26T08:27:36
{ "title": [ -0.01739502, -0.022918701, -0.03010559, -0.04324341, -0.010673523, -0.005142212, -0.033569336, -0.022201538, -0.025878906, 0.020721436, 0.017410278, 0.02130127, -0.00008672476, -0.049591064, 0.0035209656, 0.032684326, 0.004524231, 0....
http://arxiv.org/abs/2310.17217v1
Beyond MLE: Convex Learning for Text Generation
Chenze Shao, Zhengrui Ma, Min Zhang, Yang Feng
cs.CL, cs.AI, cs.LG
Maximum likelihood estimation (MLE) is a statistical method used to estimate the parameters of a probability distribution that best explain the observed data. In the context of text generation, MLE is often used to train generative language models, which can then be used to generate new text. However, we argue that MLE...
http://arxiv.org/pdf/2310.17217v1
2023-10-26T08:08:43
2023-10-26T08:08:43
{ "title": [ -0.034088135, -0.015594482, -0.028640747, -0.04360962, -0.020095825, 0.014457703, -0.057250977, -0.0053634644, -0.015991211, 0.04611206, -0.006210327, 0.033569336, -0.0016374588, 0.0018081665, -0.045806885, 0.0025997162, -0.013282776, ...
http://arxiv.org/abs/2310.17207v1
Efficient Data Fusion using the Tsetlin Machine
Rupsa Saha, Vladimir I. Zadorozhny, Ole-Christoffer Granmo
cs.AI, cs.CL
We propose a novel way of assessing and fusing noisy dynamic data using a Tsetlin Machine. Our approach consists in monitoring how explanations in form of logical clauses that a TM learns changes with possible noise in dynamic data. This way TM can recognize the noise by lowering weights of previously learned clauses, ...
http://arxiv.org/pdf/2310.17207v1
2023-10-26T07:49:25
2023-10-26T07:49:25
{ "title": [ -0.008049011, 0.02331543, -0.048858643, -0.054138184, 0.043701172, -0.0011978149, -0.007259369, -0.04473877, 0.0041542053, 0.03286743, -0.025115967, -0.007434845, 0.00042271614, -0.015335083, -0.040252686, -0.0034561157, -0.014587402, ...
http://arxiv.org/abs/2310.17191v1
How do Language Models Bind Entities in Context?
Jiahai Feng, Jacob Steinhardt
cs.LG, cs.AI, cs.CL
To correctly use in-context information, language models (LMs) must bind entities to their attributes. For example, given a context describing a "green square" and a "blue circle", LMs must bind the shapes to their respective colors. We analyze LM representations and identify the binding ID mechanism: a general mechani...
http://arxiv.org/pdf/2310.17191v1
2023-10-26T07:10:31
2023-10-26T07:10:31
{ "title": [ 0.0016860962, 0.0067367554, -0.029922485, -0.04348755, -0.01386261, 0.00178051, -0.045440674, 0.006832123, -0.012367249, 0.06707764, -0.011070251, -0.01411438, 0.04647827, -0.023956299, -0.007904053, 0.013122559, 0.009468079, 0.01837...
http://arxiv.org/abs/2310.17166v1
X-SNS: Cross-Lingual Transfer Prediction through Sub-Network Similarity
Taejun Yun, Jinhyeon Kim, Deokyeong Kang, Seong Hoon Lim, Jihoon Kim, Taeuk Kim
cs.CL
Cross-lingual transfer (XLT) is an emergent ability of multilingual language models that preserves their performance on a task to a significant extent when evaluated in languages that were not included in the fine-tuning process. While English, due to its widespread usage, is typically regarded as the primary language ...
http://arxiv.org/pdf/2310.17166v1
2023-10-26T05:39:49
2023-10-26T05:39:49
{ "title": [ -0.026992798, 0.0017271042, -0.029190063, -0.03387451, 0.006286621, 0.0025997162, -0.046875, -0.02859497, -0.0019512177, 0.038604736, 0.015731812, 0.02229309, 0.039031982, -0.039245605, -0.0053596497, 0.031982422, -0.008781433, 0.032...
http://arxiv.org/abs/2310.17143v1
Supercharging academic writing with generative AI: framework, techniques, and caveats
Zhicheng Lin
cs.CY, cs.CL
Academic writing is an indispensable yet laborious part of the research enterprise. This Perspective maps out principles and methods for using generative artificial intelligence (AI), specifically large language models (LLMs), to elevate the quality and efficiency of academic writing. We introduce a human-AI collaborat...
http://arxiv.org/pdf/2310.17143v1
2023-10-26T04:35:00
2023-10-26T04:35:00
{ "title": [ -0.018859863, 0.0042381287, -0.025741577, -0.018173218, -0.0005722046, -0.0034275055, -0.004043579, -0.04324341, -0.0072669983, 0.025283813, -0.013130188, 0.022735596, 0.000101685524, -0.021255493, -0.03250122, -0.0033245087, -0.00953674...
http://arxiv.org/abs/2310.17140v1
Symbolic Planning and Code Generation for Grounded Dialogue
Justin T. Chiu, Wenting Zhao, Derek Chen, Saujas Vaduguru, Alexander M. Rush, Daniel Fried
cs.CL, cs.AI
Large language models (LLMs) excel at processing and generating both text and code. However, LLMs have had limited applicability in grounded task-oriented dialogue as they are difficult to steer toward task objectives and fail to handle novel grounding. We present a modular and interpretable grounded dialogue system th...
http://arxiv.org/pdf/2310.17140v1
2023-10-26T04:22:23
2023-10-26T04:22:23
{ "title": [ -0.015960693, -0.015380859, -0.049468994, -0.06744385, -0.010353088, 0.0025539398, -0.052093506, 0.012504578, -0.020553589, 0.023590088, -0.0014810562, 0.011070251, -0.036834717, -0.025543213, -0.034606934, -0.024642944, -0.004558563, ...
http://arxiv.org/abs/2310.17133v1
Incorporating Probing Signals into Multimodal Machine Translation via Visual Question-Answering Pairs
Yuxin Zuo, Bei Li, Chuanhao Lv, Tong Zheng, Tong Xiao, Jingbo Zhu
cs.CL, cs.AI
This paper presents an in-depth study of multimodal machine translation (MMT), examining the prevailing understanding that MMT systems exhibit decreased sensitivity to visual information when text inputs are complete. Instead, we attribute this phenomenon to insufficient cross-modal interaction, rather than image infor...
http://arxiv.org/pdf/2310.17133v1
2023-10-26T04:13:49
2023-10-26T04:13:49
{ "title": [ -0.01979065, -0.0072402954, 0.01739502, -0.02079773, -0.037017822, 0.0011873245, -0.027862549, -0.019638062, 0.017211914, 0.028259277, -0.013916016, 0.0008406639, 0.00894165, -0.027114868, -0.06072998, -0.0021743774, 0.005756378, -0....
http://arxiv.org/abs/2310.17130v1
M2C: Towards Automatic Multimodal Manga Complement
Hongcheng Guo, Boyang Wang, Jiaqi Bai, Jiaheng Liu, Jian Yang, Zhoujun Li
cs.CL
Multimodal manga analysis focuses on enhancing manga understanding with visual and textual features, which has attracted considerable attention from both natural language processing and computer vision communities. Currently, most comics are hand-drawn and prone to problems such as missing pages, text contamination, an...
http://arxiv.org/pdf/2310.17130v1
2023-10-26T04:10:16
2023-10-26T04:10:16
{ "title": [ -0.06347656, -0.025009155, -0.056152344, -0.054473877, -0.0317688, 0.018722534, -0.091308594, 0.011871338, 0.023391724, 0.06347656, 0.031555176, -0.028579712, 0.06439209, -0.033721924, -0.057434082, 0.03062439, -0.010719299, 0.034149...
http://arxiv.org/abs/2310.17121v1
Test-time Augmentation for Factual Probing
Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui
cs.CL
Factual probing is a method that uses prompts to test if a language model "knows" certain world knowledge facts. A problem in factual probing is that small changes to the prompt can lead to large changes in model output. Previous work aimed to alleviate this problem by optimizing prompts via text mining or fine-tuning....
http://arxiv.org/pdf/2310.17121v1
2023-10-26T03:41:32
2023-10-26T03:41:32
{ "title": [ 0.0063285828, -0.01890564, -0.011749268, -0.041137695, -0.027282715, -0.016967773, -0.0034599304, -0.007156372, -0.0098724365, 0.026367188, -0.017822266, 0.0015802383, -0.008422852, -0.030761719, -0.020401001, 0.004837036, -0.013114929, ...
http://arxiv.org/abs/2310.17120v1
Topic Segmentation of Semi-Structured and Unstructured Conversational Datasets using Language Models
Reshmi Ghosh, Harjeet Singh Kajal, Sharanya Kamath, Dhuri Shrivastava, Samyadeep Basu, Hansi Zeng, Soundararajan Srinivasan
cs.CL, cs.AI, cs.LG
Breaking down a document or a conversation into multiple contiguous segments based on its semantic structure is an important and challenging problem in NLP, which can assist many downstream tasks. However, current works on topic segmentation often focus on segmentation of structured texts. In this paper, we comprehensi...
http://arxiv.org/pdf/2310.17120v1
2023-10-26T03:37:51
2023-10-26T03:37:51
{ "title": [ -0.03652954, 0.01360321, -0.0513916, -0.045166016, -0.0064849854, 0.0043678284, -0.08959961, -0.012680054, 0.0043907166, 0.019332886, 0.0009288788, 0.017425537, -0.010070801, -0.019210815, -0.036621094, 0.0029907227, -0.013168335, -0...
http://arxiv.org/abs/2310.17119v1
FLEEK: Factual Error Detection and Correction with Evidence Retrieved from External Knowledge
Farima Fatahi Bayat, Kun Qian, Benjamin Han, Yisi Sang, Anton Belyi, Samira Khorshidi, Fei Wu, Ihab F. Ilyas, Yunyao Li
cs.CL
Detecting factual errors in textual information, whether generated by large language models (LLM) or curated by humans, is crucial for making informed decisions. LLMs' inability to attribute their claims to external knowledge and their tendency to hallucinate makes it difficult to rely on their responses. Humans, too, ...
http://arxiv.org/pdf/2310.17119v1
2023-10-26T03:28:30
2023-10-26T03:28:30
{ "title": [ -0.024383545, -0.0030593872, -0.04345703, -0.033721924, 0.0034618378, -0.0061950684, -0.020843506, -0.025772095, 0.001953125, 0.02330017, -0.023162842, 0.018859863, 0.008804321, -0.014381409, -0.03024292, 0.000020742416, -0.02357483, ...
http://arxiv.org/abs/2310.17086v1
Transformers Learn Higher-Order Optimization Methods for In-Context Learning: A Study with Linear Models
Deqing Fu, Tian-Qi Chen, Robin Jia, Vatsal Sharan
cs.LG, cs.AI, cs.CL
Transformers are remarkably good at in-context learning (ICL) -- learning from demonstrations without parameter updates -- but how they perform ICL remains a mystery. Recent work suggests that Transformers may learn in-context by internally running Gradient Descent, a first-order optimization method. In this paper, we ...
http://arxiv.org/pdf/2310.17086v1
2023-10-26T01:08:47
2023-10-26T01:08:47
{ "title": [ 0.010261536, 0.02330017, -0.009086609, -0.009391785, 0.020629883, -0.039489746, 0.006790161, -0.04663086, -0.048217773, 0.037902832, -0.057769775, 0.03677368, -0.0050964355, -0.010566711, -0.000323534, -0.022415161, -0.0022449493, 0....
http://arxiv.org/abs/2310.17064v1
math-PVS: A Large Language Model Framework to Map Scientific Publications to PVS Theories
Hassen Saidi, Susmit Jha, Tuhin Sahai
cs.AI, cs.CL, cs.LG, cs.LO
As artificial intelligence (AI) gains greater adoption in a wide variety of applications, it has immense potential to contribute to mathematical discovery, by guiding conjecture generation, constructing counterexamples, assisting in formalizing mathematics, and discovering connections between different mathematical are...
http://arxiv.org/pdf/2310.17064v1
2023-10-25T23:54:04
2023-10-25T23:54:04
{ "title": [ -0.034698486, -0.009178162, -0.046691895, -0.043762207, -0.007843018, 0.005470276, -0.032592773, -0.019714355, 0.0006570816, 0.012306213, 0.036071777, -0.00009030104, -0.010299683, -0.04071045, -0.023361206, 0.023742676, 0.0032405853, ...
http://arxiv.org/abs/2310.17054v1
BOOST: Harnessing Black-Box Control to Boost Commonsense in LMs' Generation
Yufei Tian, Felix Zhang, Nanyun Peng
cs.CL, cs.LG
Large language models (LLMs) such as GPT-3 have demonstrated a strong capability to generate coherent and contextually relevant text. However, amidst their successes, a crucial issue persists: their generated outputs still lack commonsense at times. Moreover, fine-tuning the entire LLM towards more commonsensical outpu...
http://arxiv.org/pdf/2310.17054v1
2023-10-25T23:32:12
2023-10-25T23:32:12
{ "title": [ 0.014533997, -0.021987915, -0.029159546, -0.033111572, -0.016937256, -0.02355957, -0.011787415, -0.033721924, -0.011985779, 0.02420044, -0.034606934, 0.022125244, -0.006137848, -0.017410278, -0.018920898, -0.027694702, -0.024459839, ...
http://arxiv.org/abs/2310.17041v1
On Surgical Fine-tuning for Language Encoders
Abhilasha Lodha, Gayatri Belapurkar, Saloni Chalkapurkar, Yuanming Tao, Reshmi Ghosh, Samyadeep Basu, Dmitrii Petrov, Soundararajan Srinivasan
cs.CL, cs.AI, cs.IR
Fine-tuning all the layers of a pre-trained neural language encoder (either using all the parameters or using parameter-efficient methods) is often the de-facto way of adapting it to a new task. We show evidence that for different downstream language tasks, fine-tuning only a subset of layers is sufficient to obtain pe...
http://arxiv.org/pdf/2310.17041v1
2023-10-25T22:42:30
2023-10-25T22:42:30
{ "title": [ -0.02947998, -0.011230469, -0.004425049, -0.037994385, -0.027801514, 0.0110321045, -0.043548584, 0.00028300285, -0.007850647, 0.05618286, -0.011367798, 0.014877319, -0.003917694, -0.006576538, -0.03656006, 0.02268982, 0.008651733, 0....
http://arxiv.org/abs/2310.18369v1
Apollo: Zero-shot MultiModal Reasoning with Multiple Experts
Daniela Ben-David, Tzuf Paz-Argaman, Reut Tsarfaty
cs.CL, cs.AI, cs.CV, I.2.7; I.5.4
We propose a modular framework that leverages the expertise of different foundation models over different modalities and domains in order to perform a single, complex, multi-modal task, without relying on prompt engineering or otherwise tailor-made multi-modal training. Our approach enables decentralized command execut...
http://arxiv.org/pdf/2310.18369v1
2023-10-25T22:36:40
2023-10-25T22:36:40
{ "title": [ -0.019729614, -0.009170532, -0.05731201, -0.06713867, 0.0046081543, -0.0063552856, -0.03451538, 0.0087509155, 0.024627686, 0.010391235, 0.0056762695, -0.008735657, -0.017059326, -0.025405884, -0.0362854, 0.0063552856, -0.0076179504, ...
http://arxiv.org/abs/2310.17034v1
Follow-on Question Suggestion via Voice Hints for Voice Assistants
Besnik Fetahu, Pedro Faustini, Giuseppe Castellucci, Anjie Fang, Oleg Rokhlenko, Shervin Malmasi
cs.CL
The adoption of voice assistants like Alexa or Siri has grown rapidly, allowing users to instantly access information via voice search. Query suggestion is a standard feature of screen-based search experiences, allowing users to explore additional topics. However, this is not trivial to implement in voice-based setting...
http://arxiv.org/pdf/2310.17034v1
2023-10-25T22:22:18
2023-10-25T22:22:18
{ "title": [ 0.030563354, -0.011802673, -0.026748657, -0.05456543, -0.031707764, -0.022094727, -0.023727417, 0.0022602081, -0.030441284, 0.045532227, -0.033111572, 0.007843018, 0.045715332, -0.025344849, 0.011138916, -0.0009050369, 0.019256592, 0...
http://arxiv.org/abs/2310.17022v1
Controlled Decoding from Language Models
Sidharth Mudgal, Jong Lee, Harish Ganapathy, YaGuang Li, Tao Wang, Yanping Huang, Zhifeng Chen, Heng-Tze Cheng, Michael Collins, Trevor Strohman, Jilin Chen, Alex Beutel, Ahmad Beirami
cs.LG, cs.AI, cs.CL
We propose controlled decoding (CD), a novel off-policy reinforcement learning method to control the autoregressive generation from language models towards high reward outcomes. CD solves an off-policy reinforcement learning problem through a value function for the reward, which we call a prefix scorer. The prefix scor...
http://arxiv.org/pdf/2310.17022v1
2023-10-25T22:00:05
2023-10-25T22:00:05
{ "title": [ -0.054138184, -0.01386261, -0.032043457, -0.0496521, -0.023376465, 0.010551453, -0.06951904, -0.002565384, 0.0013113022, 0.053375244, -0.003780365, -0.018325806, 0.023284912, -0.022155762, -0.051879883, 0.029937744, -0.0011835098, 0....
http://arxiv.org/abs/2310.17019v1
Conditionally Combining Robot Skills using Large Language Models
K. R. Zentner, Ryan Julian, Brian Ichter, Gaurav S. Sukhatme
cs.LG, cs.CL, cs.RO
This paper combines two contributions. First, we introduce an extension of the Meta-World benchmark, which we call "Language-World," which allows a large language model to operate in a simulated robotic environment using semi-structured natural language queries and scripted skills described using natural language. By u...
http://arxiv.org/pdf/2310.17019v1
2023-10-25T21:46:34
2023-10-25T21:46:34
{ "title": [ -0.026138306, 0.012184143, -0.035095215, -0.042297363, 0.0050811768, -0.0062942505, -0.046417236, -0.01586914, -0.004573822, 0.02458191, -0.041625977, -0.0101623535, -0.0074768066, -0.0032348633, -0.02267456, -0.02394104, -0.006515503, ...
http://arxiv.org/abs/2310.17017v1
An Integrative Survey on Mental Health Conversational Agents to Bridge Computer Science and Medical Perspectives
Young Min Cho, Sunny Rai, Lyle Ungar, João Sedoc, Sharath Chandra Guntuku
cs.CL, cs.AI
Mental health conversational agents (a.k.a. chatbots) are widely studied for their potential to offer accessible support to those experiencing mental health challenges. Previous surveys on the topic primarily consider papers published in either computer science or medicine, leading to a divide in understanding and hind...
http://arxiv.org/pdf/2310.17017v1
2023-10-25T21:37:57
2023-10-25T21:37:57
{ "title": [ 0.006877899, -0.0054740906, -0.029373169, -0.05215454, 0.016693115, -0.01991272, -0.012840271, -0.0044898987, 0.0027828217, 0.042907715, -0.012359619, -0.0064811707, 0.03475952, -0.04095459, -0.02583313, 0.006462097, -0.010269165, 0....
http://arxiv.org/abs/2310.17015v3
Data Augmentation for Emotion Detection in Small Imbalanced Text Data
Anna Koufakou, Diego Grisales, Ragy Costa de jesus, Oscar Fox
cs.CL
Emotion recognition in text, the task of identifying emotions such as joy or anger, is a challenging problem in NLP with many applications. One of the challenges is the shortage of available datasets that have been annotated with emotions. Certain existing datasets are small, follow different emotion taxonomies and dis...
http://arxiv.org/pdf/2310.17015v3
2023-10-30T13:33:16
2023-10-25T21:29:36
{ "title": [ -0.0004541874, 0.004951477, -0.04171753, -0.024337769, 0.009239197, -0.019882202, -0.00409317, -0.016067505, 0.041931152, 0.020080566, -0.053619385, 0.0006365776, 0.013893127, -0.008613586, -0.055511475, 0.0031661987, -0.016204834, 0...
http://arxiv.org/abs/2310.17010v1
This Reads Like That: Deep Learning for Interpretable Natural Language Processing
Claudio Fanconi, Moritz Vandenhirtz, Severin Husmann, Julia E. Vogt
cs.CL, cs.AI
Prototype learning, a popular machine learning method designed for inherently interpretable decisions, leverages similarities to learned prototypes for classifying new data. While it is mainly applied in computer vision, in this work, we build upon prior research and further explore the extension of prototypical networ...
http://arxiv.org/pdf/2310.17010v1
2023-10-25T21:18:35
2023-10-25T21:18:35
{ "title": [ -0.029968262, 0.0054359436, -0.05822754, -0.01876831, -0.021469116, -0.017959595, -0.06719971, -0.019866943, -0.02444458, 0.0054244995, -0.04434204, 0.00881958, -0.024765015, -0.01625061, -0.0063209534, -0.03729248, 0.0025672913, 0.0...
http://arxiv.org/abs/2310.16995v1
Quality > Quantity: Synthetic Corpora from Foundation Models for Closed-Domain Extractive Question Answering
Saptarshi Sengupta, Connor Heaton, Shreya Ghosh, Preslav Nakov, Prasenjit Mitra
cs.CL
Domain adaptation, the process of training a model in one domain and applying it to another, has been extensively explored in machine learning. While training a domain-specific foundation model (FM) from scratch is an option, recent methods have focused on adapting pre-trained FMs for domain-specific tasks. However, ou...
http://arxiv.org/pdf/2310.16995v1
2023-10-25T20:48:16
2023-10-25T20:48:16
{ "title": [ -0.034698486, -0.00207901, -0.033081055, -0.053741455, -0.014137268, 0.00447464, -0.046020508, 0.0021820068, 0.007385254, 0.02067566, -0.044189453, 0.022872925, -0.03253174, 0.0058021545, -0.053375244, -0.018692017, -0.018234253, 0.0...
http://arxiv.org/abs/2310.16992v1
How well can machine-generated texts be identified and can language models be trained to avoid identification?
Sinclair Schneider, Florian Steuber, Joao A. G. Schneider, Gabi Dreo Rodosek
cs.CL
With the rise of generative pre-trained transformer models such as GPT-3, GPT-NeoX, or OPT, distinguishing human-generated texts from machine-generated ones has become important. We refined five separate language models to generate synthetic tweets, uncovering that shallow learning classification algorithms, like Naive...
http://arxiv.org/pdf/2310.16992v1
2023-10-25T20:43:07
2023-10-25T20:43:07
{ "title": [ 0.012893677, 0.019210815, 0.005329132, 0.017532349, -0.021316528, -0.002653122, -0.00023579597, -0.050628662, -0.0209198, 0.03302002, -0.009033203, 0.010185242, 0.015777588, -0.028671265, 0.006210327, -0.014099121, -0.029754639, 0.07...
http://arxiv.org/abs/2310.16990v1
STEER: Semantic Turn Extension-Expansion Recognition for Voice Assistants
Leon Liyang Zhang, Jiarui Lu, Joel Ruben Antony Moniz, Aditya Kulkarni, Dhivya Piraviperumal, Tien Dung Tran, Nicholas Tzou, Hong Yu
cs.CL, cs.AI, cs.LG
In the context of a voice assistant system, steering refers to the phenomenon in which a user issues a follow-up command attempting to direct or clarify a previous turn. We propose STEER, a steering detection model that predicts whether a follow-up turn is a user's attempt to steer the previous command. Constructing a ...
http://arxiv.org/pdf/2310.16990v1
2023-10-25T20:41:30
2023-10-25T20:41:30
{ "title": [ -0.020950317, -0.021560669, -0.050323486, -0.062164307, -0.0036678314, -0.00687027, -0.05429077, 0.023727417, -0.007633209, 0.05368042, -0.044769287, 0.018371582, 0.011451721, 0.012016296, -0.054840088, -0.004333496, -0.015045166, 0....
http://arxiv.org/abs/2310.16968v1
Understanding Social Structures from Contemporary Literary Fiction using Character Interaction Graph -- Half Century Chronology of Influential Bengali Writers
Nafis Irtiza Tripto, Mohammed Eunus Ali
cs.CL, cs.CY
Social structures and real-world incidents often influence contemporary literary fiction. Existing research in literary fiction analysis explains these real-world phenomena through the manual critical analysis of stories. Conventional Natural Language Processing (NLP) methodologies, including sentiment analysis, narrat...
http://arxiv.org/pdf/2310.16968v1
2023-10-25T20:09:14
2023-10-25T20:09:14
{ "title": [ 0.016296387, -0.016845703, -0.025878906, -0.043518066, -0.018508911, -0.03201294, 0.002855301, 0.026306152, -0.01979065, 0.020111084, -0.0041236877, 0.011795044, 0.00541687, -0.042236328, 0.024734497, -0.017288208, -0.015380859, 0.02...
http://arxiv.org/abs/2310.16964v1
Critic-Driven Decoding for Mitigating Hallucinations in Data-to-text Generation
Mateusz Lango, Ondřej Dušek
cs.CL, I.2.7
Hallucination of text ungrounded in the input is a well-known problem in neural data-to-text generation. Many methods have been proposed to mitigate it, but they typically require altering model architecture or collecting additional data, and thus cannot be easily applied to an existing model. In this paper, we explore...
http://arxiv.org/pdf/2310.16964v1
2023-10-25T20:05:07
2023-10-25T20:05:07
{ "title": [ -0.04434204, -0.0084991455, -0.029327393, -0.030700684, -0.03475952, 0.01361084, -0.043884277, -0.0068473816, 0.0060310364, 0.05078125, -0.008399963, 0.023040771, 0.014045715, 0.00060510635, -0.05734253, 0.017684937, -0.004550934, 0....
http://arxiv.org/abs/2310.18368v1
Muslim-Violence Bias Persists in Debiased GPT Models
Babak Hemmatian, Razan Baltaji, Lav R. Varshney
cs.CL, I.2.7
Abid et al. (2021) showed a tendency in GPT-3 to generate violent completions when prompted about Muslims, compared with other religions. Two pre-registered replication attempts found few violent completions and only the weakest anti-Muslim bias in the Instruct version, fine-tuned to eliminate biased and toxic outputs....
http://arxiv.org/pdf/2310.18368v1
2023-10-25T19:39:58
2023-10-25T19:39:58
{ "title": [ 0.04949951, 0.01838684, -0.006011963, -0.020690918, -0.005130768, -0.041229248, 0.004005432, -0.032318115, -0.013877869, 0.01461792, -0.025741577, -0.0012969971, 0.00072193146, -0.048950195, -0.0026893616, -0.0068855286, 0.002904892, ...
http://arxiv.org/abs/2310.16944v1
Zephyr: Direct Distillation of LM Alignment
Lewis Tunstall, Edward Beeching, Nathan Lambert, Nazneen Rajani, Kashif Rasul, Younes Belkada, Shengyi Huang, Leandro von Werra, Clémentine Fourrier, Nathan Habib, Nathan Sarrazin, Omar Sanseviero, Alexander M. Rush, Thomas Wolf
cs.LG, cs.CL
We aim to produce a smaller language model that is aligned to user intent. Previous research has shown that applying distilled supervised fine-tuning (dSFT) on larger models significantly improves task accuracy; however, these models are unaligned, i.e. they do not respond well to natural prompts. To distill this prope...
http://arxiv.org/pdf/2310.16944v1
2023-10-25T19:25:16
2023-10-25T19:25:16
{ "title": [ -0.033355713, 0.0041542053, -0.08337402, -0.045013428, -0.045959473, -0.008178711, -0.1104126, -0.017700195, 0.0047836304, 0.021347046, -0.014724731, -0.020706177, 0.058166504, -0.05734253, 0.018249512, 0.0032367706, 0.009735107, 0.0...
http://arxiv.org/abs/2310.16937v1
Learning Transfers over Several Programming Languages
Razan Baltaji, Saurabh Pujar, Louis Mandel, Martin Hirzel, Luca Buratti, Lav Varshney
cs.CL, I.2.7; I.2.5
Large language models (LLMs) have recently become remarkably good at improving developer productivity for high-resource programming languages. These models use two kinds of data: large amounts of unlabeled code samples for pretraining and relatively smaller amounts of labeled code samples for fine-tuning or in-context ...
http://arxiv.org/pdf/2310.16937v1
2023-10-25T19:04:33
2023-10-25T19:04:33
{ "title": [ 0.021865845, -0.017105103, -0.03326416, -0.037902832, -0.042388916, -0.04711914, -0.010643005, 0.029922485, -0.023452759, 0.040039062, -0.02079773, 0.015052795, -0.013900757, -0.036987305, 0.01914978, -0.010185242, 0.032470703, -0.02...
http://arxiv.org/abs/2310.16931v1
CL-MASR: A Continual Learning Benchmark for Multilingual ASR
Luca Della Libera, Pooneh Mousavi, Salah Zaiem, Cem Subakan, Mirco Ravanelli
cs.CL, cs.AI
Modern multilingual automatic speech recognition (ASR) systems like Whisper have made it possible to transcribe audio in multiple languages with a single model. However, current state-of-the-art ASR models are typically evaluated on individual languages or in a multi-task setting, overlooking the challenge of continual...
http://arxiv.org/pdf/2310.16931v1
2023-10-25T18:55:40
2023-10-25T18:55:40
{ "title": [ -0.022567749, -0.0026168823, -0.041503906, -0.08166504, 0.00061130524, 0.008865356, -0.08258057, 0.019378662, -0.008956909, 0.04724121, 0.012573242, 0.0042037964, 0.0023612976, -0.02947998, -0.039916992, 0.016342163, -0.0032749176, 0...
http://arxiv.org/abs/2310.16924v1
Physician Detection of Clinical Harm in Machine Translation: Quality Estimation Aids in Reliance and Backtranslation Identifies Critical Errors
Nikita Mehandru, Sweta Agrawal, Yimin Xiao, Elaine C Khoong, Ge Gao, Marine Carpuat, Niloufar Salehi
cs.CL, cs.HC
A major challenge in the practical use of Machine Translation (MT) is that users lack guidance to make informed decisions about when to rely on outputs. Progress in quality estimation research provides techniques to automatically assess MT quality, but these techniques have primarily been evaluated in vitro by comparis...
http://arxiv.org/pdf/2310.16924v1
2023-10-25T18:44:14
2023-10-25T18:44:14
{ "title": [ -0.015670776, 0.0043907166, 0.006515503, -0.0012016296, -0.041290283, -0.00349617, -0.06561279, -0.06286621, -0.035339355, 0.029769897, 0.006767273, 0.015991211, 0.039978027, -0.0625, 0.024459839, -0.00045990944, -0.014816284, 0.0270...
http://arxiv.org/abs/2310.16897v1
Divide et Impera: Multi-Transformer Architectures for Complex NLP-Tasks
Solveig Helland, Elena Gavagnin, Alexandre de Spindler
cs.CL
The growing capabilities of transformer models pave the way for solving increasingly complex NLP tasks. A key to supporting application-specific requirements is the ability to fine-tune. However, compiling a fine-tuning dataset tailored to complex tasks is tedious and results in large datasets, limiting the ability to ...
http://arxiv.org/pdf/2310.16897v1
2023-10-25T18:00:15
2023-10-25T18:00:15
{ "title": [ -0.014564514, -0.009674072, -0.030731201, -0.06011963, 0.002796173, 0.00831604, -0.03793335, 0.00024914742, 0.009498596, 0.03366089, -0.039245605, 0.008407593, 0.021743774, 0.0023498535, -0.042785645, 0.010650635, -0.018585205, 0.057...
http://arxiv.org/abs/2310.16836v1
LLM-FP4: 4-Bit Floating-Point Quantized Transformers
Shih-yang Liu, Zechun Liu, Xijie Huang, Pingcheng Dong, Kwang-Ting Cheng
cs.CL, cs.AI, cs.AR, cs.CV
We propose LLM-FP4 for quantizing both weights and activations in large language models (LLMs) down to 4-bit floating-point values, in a post-training manner. Existing post-training quantization (PTQ) solutions are primarily integer-based and struggle with bit widths below 8 bits. Compared to integer quantization, floa...
http://arxiv.org/pdf/2310.16836v1
2023-10-25T17:59:32
2023-10-25T17:59:32
{ "title": [ -0.016601562, 0.006511688, -0.026397705, -0.021728516, -0.017990112, -0.010192871, -0.031143188, -0.0072402954, -0.026519775, 0.062286377, -0.017227173, 0.011291504, 0.015129089, -0.030090332, -0.015312195, 0.003604889, 0.011566162, ...
http://arxiv.org/abs/2310.16834v1
Discrete Diffusion Language Modeling by Estimating the Ratios of the Data Distribution
Aaron Lou, Chenlin Meng, Stefano Ermon
stat.ML, cs.CL, cs.LG
Despite their groundbreaking performance for many generative modeling tasks, diffusion models have fallen short on discrete data domains such as natural language. Crucially, standard diffusion models rely on the well-established theory of score matching, but efforts to generalize this to discrete structures have not yi...
http://arxiv.org/pdf/2310.16834v1
2023-10-25T17:59:12
2023-10-25T17:59:12
{ "title": [ -0.038757324, 0.0044174194, -0.07043457, -0.019363403, 0.00050497055, -0.020767212, -0.021560669, -0.008079529, -0.013496399, 0.026779175, -0.03692627, -0.0036773682, 0.010810852, -0.031402588, -0.030685425, -0.0058135986, -0.02243042, ...
http://arxiv.org/abs/2310.16822v1
Prompt Me Up: Unleashing the Power of Alignments for Multimodal Entity and Relation Extraction
Xuming Hu, Junzhe Chen, Aiwei Liu, Shiao Meng, Lijie Wen, Philip S. Yu
cs.CL, cs.AI, cs.MM
How can we better extract entities and relations from text? Using multimodal extraction with images and text obtains more signals for entities and relations, and aligns them through graphs or hierarchical fusion, aiding in extraction. Despite attempts at various fusions, previous works have overlooked many unlabeled im...
http://arxiv.org/pdf/2310.16822v1
2023-10-25T17:51:56
2023-10-25T17:51:56
{ "title": [ -0.019042969, 0.003730774, -0.0670166, -0.066223145, 0.02003479, -0.010063171, -0.047546387, 0.0031547546, 0.025924683, 0.02911377, -0.011131287, 0.012588501, 0.022521973, -0.009933472, -0.041900635, 0.017730713, -0.018508911, 0.0506...
http://arxiv.org/abs/2310.16810v1
Can GPT models Follow Human Summarization Guidelines? Evaluating ChatGPT and GPT-4 for Dialogue Summarization
Yongxin Zhou, Fabien Ringeval, François Portet
cs.CL, cs.AI
This study explores the capabilities of prompt-driven Large Language Models (LLMs) like ChatGPT and GPT-4 in adhering to human guidelines for dialogue summarization. Experiments employed DialogSum (English social conversations) and DECODA (French call center interactions), testing various prompts: including prompts fro...
http://arxiv.org/pdf/2310.16810v1
2023-10-25T17:39:07
2023-10-25T17:39:07
{ "title": [ -0.013175964, -0.0016756058, -0.0061187744, -0.034210205, -0.039001465, 0.0032043457, -0.0574646, -0.023147583, -0.02407837, 0.038482666, -0.0021152496, 0.0154800415, -0.0034160614, -0.013298035, -0.046142578, 0.015296936, -0.029815674, ...