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/2310.01846v1
Benchmarking and Improving Generator-Validator Consistency of Language Models
Xiang Lisa Li, Vaishnavi Shrivastava, Siyan Li, Tatsunori Hashimoto, Percy Liang
cs.CL, cs.LG
As of September 2023, ChatGPT correctly answers "what is 7+8" with 15, but when asked "7+8=15, True or False" it responds with "False". This inconsistency between generating and validating an answer is prevalent in language models (LMs) and erodes trust. In this paper, we propose a framework for measuring the consisten...
http://arxiv.org/pdf/2310.01846v1
2023-10-03T07:23:22
2023-10-03T07:23:22
{ "title": [ -0.026062012, 0.0056495667, -0.054870605, -0.11010742, 0.027755737, 0.023956299, -0.0925293, 0.017700195, -0.008972168, 0.04486084, 0.0030899048, 0.011749268, 0.012413025, -0.020431519, -0.052581787, 0.01423645, -0.029022217, 0.03411...
http://arxiv.org/abs/2310.01845v1
Zero-Shot Refinement of Buildings' Segmentation Models using SAM
Ali Mayladan, Hasan Nasrallah, Hasan Moughnieh, Mustafa Shukor, Ali J. Ghandour
cs.CV, cs.AI, cs.CL, cs.LG
Foundation models have excelled in various tasks but are often evaluated on general benchmarks. The adaptation of these models for specific domains, such as remote sensing imagery, remains an underexplored area. In remote sensing, precise building instance segmentation is vital for applications like urban planning. Whi...
http://arxiv.org/pdf/2310.01845v1
2023-10-03T07:19:59
2023-10-03T07:19:59
{ "title": [ -0.01398468, -0.0090408325, -0.012611389, -0.016784668, -0.03086853, -0.0007510185, -0.043701172, -0.03112793, 0.0027942657, 0.019332886, -0.016967773, -0.014190674, -0.02923584, -0.020324707, -0.046142578, -0.012184143, 0.00712204, ...
http://arxiv.org/abs/2310.01839v2
Preserving Phonemic Distinctions for Ordinal Regression: A Novel Loss Function for Automatic Pronunciation Assessment
Bi-Cheng Yan, Hsin-Wei Wang, Yi-Cheng Wang, Jiun-Ting Li, Chi-Han Lin, Berlin Chen
eess.AS, cs.CL, cs.SD
Automatic pronunciation assessment (APA) manages to quantify the pronunciation proficiency of a second language (L2) learner in a language. Prevailing approaches to APA normally leverage neural models trained with a regression loss function, such as the mean-squared error (MSE) loss, for proficiency level prediction. D...
http://arxiv.org/pdf/2310.01839v2
2023-10-04T06:51:24
2023-10-03T07:05:37
{ "title": [ -0.028366089, 0.007774353, -0.03277588, -0.03955078, -0.014694214, 0.0069503784, -0.03567505, -0.013336182, -0.009735107, 0.0390625, -0.021835327, 0.000016748905, 0.015991211, -0.019638062, -0.05303955, -0.020736694, -0.027282715, 0....
http://arxiv.org/abs/2310.01837v1
Extending CAM-based XAI methods for Remote Sensing Imagery Segmentation
Abdul Karim Gizzini, Mustafa Shukor, Ali J. Ghandour
cs.CV, cs.AI, cs.CL, cs.LG
Current AI-based methods do not provide comprehensible physical interpretations of the utilized data, extracted features, and predictions/inference operations. As a result, deep learning models trained using high-resolution satellite imagery lack transparency and explainability and can be merely seen as a black box, wh...
http://arxiv.org/pdf/2310.01837v1
2023-10-03T07:01:23
2023-10-03T07:01:23
{ "title": [ -0.04901123, 0.0036773682, -0.035186768, 0.0073509216, -0.0093307495, -0.0030727386, -0.01473999, -0.017486572, -0.0039711, 0.02418518, -0.032043457, 0.012519836, -0.053619385, -0.009391785, -0.058624268, -0.01612854, 0.019332886, 0....
http://arxiv.org/abs/2310.01828v1
Trainable Noise Model as an XAI evaluation method: application on Sobol for remote sensing image segmentation
Hossein Shreim, Abdul Karim Gizzini, Ali J. Ghandour
cs.CV, cs.AI, cs.CL, cs.LG
eXplainable Artificial Intelligence (XAI) has emerged as an essential requirement when dealing with mission-critical applications, ensuring transparency and interpretability of the employed black box AI models. The significance of XAI spans various domains, from healthcare to finance, where understanding the decision-m...
http://arxiv.org/pdf/2310.01828v1
2023-10-03T06:51:48
2023-10-03T06:51:48
{ "title": [ -0.036987305, 0.015449524, -0.023712158, -0.000042676926, -0.011932373, 0.00032043457, -0.026794434, -0.014732361, -0.032592773, 0.011604309, -0.025466919, 0.029754639, -0.07019043, -0.03353882, -0.018508911, -0.03527832, 0.0030326843, ...
http://arxiv.org/abs/2310.01825v1
Empirical Study of PEFT techniques for Winter Wheat Segmentation
Mohamad Hasan Zahweh, Hasan Nasrallah, Mustafa Shukor, Ghaleb Faour, Ali J. Ghandour
cs.CV, cs.AI, cs.CL, cs.LG
Parameter Efficient Fine Tuning (PEFT) techniques have recently experienced significant growth and have been extensively employed to adapt large vision and language models to various domains, enabling satisfactory model performance with minimal computational needs. Despite these advances, more research has yet to delve...
http://arxiv.org/pdf/2310.01825v1
2023-10-03T06:42:28
2023-10-03T06:42:28
{ "title": [ -0.0015468597, -0.0021743774, -0.013206482, 0.00881958, -0.008422852, -0.012290955, 0.025802612, -0.03363037, -0.033050537, 0.043060303, -0.005897522, 0.019210815, -0.019943237, -0.023986816, -0.036987305, -0.003200531, -0.0019226074, ...
http://arxiv.org/abs/2310.01801v2
Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
Suyu Ge, Yunan Zhang, Liyuan Liu, Minjia Zhang, Jiawei Han, Jianfeng Gao
cs.CL
In this study, we introduce adaptive KV cache compression, a plug-and-play method that reduces the memory footprint of generative inference for Large Language Models (LLMs). Different from the conventional KV cache that retains key and value vectors for all context tokens, we conduct targeted profiling to discern the i...
http://arxiv.org/pdf/2310.01801v2
2023-10-07T03:49:17
2023-10-03T05:17:08
{ "title": [ -0.052093506, -0.0023880005, -0.03805542, 0.016433716, -0.0463562, 0.0065231323, -0.053710938, -0.02267456, -0.014030457, 0.010223389, -0.022140503, -0.028244019, 0.02809143, -0.050567627, 0.0072250366, -0.03579712, 0.034301758, 0.01...
http://arxiv.org/abs/2310.01798v1
Large Language Models Cannot Self-Correct Reasoning Yet
Jie Huang, Xinyun Chen, Swaroop Mishra, Huaixiu Steven Zheng, Adams Wei Yu, Xinying Song, Denny Zhou
cs.CL, cs.AI
Large Language Models (LLMs) have emerged as a groundbreaking technology with their unparalleled text generation capabilities across various applications. Nevertheless, concerns persist regarding the accuracy and appropriateness of their generated content. A contemporary methodology, self-correction, has been proposed ...
http://arxiv.org/pdf/2310.01798v1
2023-10-03T04:56:12
2023-10-03T04:56:12
{ "title": [ -0.02482605, -0.019515991, -0.015213013, -0.029830933, -0.05999756, 0.018585205, -0.039245605, 0.006175995, 0.009269714, 0.03729248, -0.041748047, -0.059814453, 0.0076026917, -0.03378296, -0.035583496, -0.012786865, 0.016601562, 0.02...
http://arxiv.org/abs/2310.01783v1
Can large language models provide useful feedback on research papers? A large-scale empirical analysis
Weixin Liang, Yuhui Zhang, Hancheng Cao, Binglu Wang, Daisy Ding, Xinyu Yang, Kailas Vodrahalli, Siyu He, Daniel Smith, Yian Yin, Daniel McFarland, James Zou
cs.LG, cs.AI, cs.CL, cs.HC
Expert feedback lays the foundation of rigorous research. However, the rapid growth of scholarly production and intricate knowledge specialization challenge the conventional scientific feedback mechanisms. High-quality peer reviews are increasingly difficult to obtain. Researchers who are more junior or from under-reso...
http://arxiv.org/pdf/2310.01783v1
2023-10-03T04:14:17
2023-10-03T04:14:17
{ "title": [ -0.044952393, -0.0016117096, -0.028411865, -0.01574707, -0.013748169, -0.007949829, -0.039978027, -0.03277588, -0.0068626404, 0.023834229, -0.023651123, 0.0030155182, -0.00012505054, -0.0032958984, -0.05419922, 0.0047569275, -0.002130508...
http://arxiv.org/abs/2310.01777v1
SEA: Sparse Linear Attention with Estimated Attention Mask
Heejun Lee, Jina Kim, Jeffrey Willette, Sung Ju Hwang
cs.CL, cs.LG
The transformer architecture has made breakthroughs in recent years on tasks which require modeling pairwise relationships between sequential elements, as is the case in natural language understanding. However, transformers struggle with long sequences due to the quadratic complexity of the attention operation, and pre...
http://arxiv.org/pdf/2310.01777v1
2023-10-03T03:56:26
2023-10-03T03:56:26
{ "title": [ -0.031036377, 0.0028305054, -0.063964844, -0.034179688, 0.009353638, -0.017578125, -0.016418457, -0.00042891502, 0.016098022, 0.0070648193, -0.025344849, -0.0023384094, 0.014640808, -0.020263672, -0.020980835, 0.009216309, 0.010169983, ...
http://arxiv.org/abs/2310.01749v1
Stack Attention: Improving the Ability of Transformers to Model Hierarchical Patterns
Brian DuSell, David Chiang
cs.CL
Attention, specifically scaled dot-product attention, has proven effective for natural language, but it does not have a mechanism for handling hierarchical patterns of arbitrary nesting depth, which limits its ability to recognize certain syntactic structures. To address this shortcoming, we propose stack attention: an...
http://arxiv.org/pdf/2310.01749v1
2023-10-03T02:18:06
2023-10-03T02:18:06
{ "title": [ 0.018737793, 0.01373291, -0.03515625, -0.094177246, 0.037078857, 0.0041770935, -0.039886475, -0.010475159, 0.01576233, 0.03265381, -0.030197144, -0.011787415, 0.026916504, -0.020385742, -0.046020508, 0.01348114, -0.004337311, 0.04983...
http://arxiv.org/abs/2310.01732v1
Nugget: Neural Agglomerative Embeddings of Text
Guanghui Qin, Benjamin Van Durme
cs.CL, cs.AI, cs.LG, I.2.7; I.2.6
Embedding text sequences is a widespread requirement in modern language understanding. Existing approaches focus largely on constant-size representations. This is problematic, as the amount of information contained in text often varies with the length of the input. We propose a solution called Nugget, which encodes lan...
http://arxiv.org/pdf/2310.01732v1
2023-10-03T01:47:49
2023-10-03T01:47:49
{ "title": [ -0.014656067, -0.0050811768, -0.037963867, -0.057525635, 0.013847351, 0.008346558, -0.029129028, -0.00082445145, 0.015731812, 0.03753662, -0.016937256, 0.014251709, 0.0008907318, -0.0002193451, -0.045440674, 0.0063667297, -0.009986877, ...
http://arxiv.org/abs/2310.01717v1
Ensemble Distillation for Unsupervised Constituency Parsing
Behzad Shayegh, Yanshuai Cao, Xiaodan Zhu, Jackie C. K. Cheung, Lili Mou
cs.CL, cs.AI, cs.LG
We investigate the unsupervised constituency parsing task, which organizes words and phrases of a sentence into a hierarchical structure without using linguistically annotated data. We observe that existing unsupervised parsers capture differing aspects of parsing structures, which can be leveraged to enhance unsupervi...
http://arxiv.org/pdf/2310.01717v1
2023-10-03T01:02:44
2023-10-03T01:02:44
{ "title": [ -0.0049591064, 0.009140015, -0.029678345, -0.082458496, 0.01586914, 0.0062561035, -0.06390381, 0.0010728836, 0.015991211, 0.016143799, -0.043518066, 0.0016870499, -0.013893127, -0.00223732, -0.043395996, -0.013145447, -0.01802063, 0....
http://arxiv.org/abs/2310.01708v1
Deciphering Diagnoses: How Large Language Models Explanations Influence Clinical Decision Making
D. Umerenkov, G. Zubkova, A. Nesterov
cs.CL
Clinical Decision Support Systems (CDSS) utilize evidence-based knowledge and patient data to offer real-time recommendations, with Large Language Models (LLMs) emerging as a promising tool to generate plain-text explanations for medical decisions. This study explores the effectiveness and reliability of LLMs in genera...
http://arxiv.org/pdf/2310.01708v1
2023-10-03T00:08:23
2023-10-03T00:08:23
{ "title": [ -0.01638794, -0.01966858, -0.03036499, -0.084228516, -0.008323669, 0.00459671, -0.048583984, -0.004119873, -0.016021729, 0.01852417, -0.010093689, 0.0026016235, -0.01071167, -0.025527954, -0.010604858, -0.005580902, 0.00030064583, 0....
http://arxiv.org/abs/2310.04445v2
LoFT: Local Proxy Fine-tuning For Improving Transferability Of Adversarial Attacks Against Large Language Model
Muhammad Ahmed Shah, Roshan Sharma, Hira Dhamyal, Raphael Olivier, Ankit Shah, Joseph Konan, Dareen Alharthi, Hazim T Bukhari, Massa Baali, Soham Deshmukh, Michael Kuhlmann, Bhiksha Raj, Rita Singh
cs.CL, cs.AI, cs.LG
It has been shown that Large Language Model (LLM) alignments can be circumvented by appending specially crafted attack suffixes with harmful queries to elicit harmful responses. To conduct attacks against private target models whose characterization is unknown, public models can be used as proxies to fashion the attack...
http://arxiv.org/pdf/2310.04445v2
2023-10-21T17:59:41
2023-10-02T23:29:23
{ "title": [ -0.009017944, -0.0062446594, -0.025878906, -0.058380127, 0.0074691772, -0.0036258698, -0.026000977, 0.006843567, -0.053619385, 0.030731201, 0.028137207, 0.027954102, -0.009384155, -0.033813477, -0.009521484, -0.004425049, -0.009963989, ...
http://arxiv.org/abs/2310.01693v1
Closing the Curious Case of Neural Text Degeneration
Matthew Finlayson, John Hewitt, Alexander Koller, Swabha Swayamdipta, Ashish Sabharwal
cs.CL, 68T50, I.2.7
Despite their ubiquity in language generation, it remains unknown why truncation sampling heuristics like nucleus sampling are so effective. We provide a theoretical explanation for the effectiveness of the truncation sampling by proving that truncation methods that discard tokens below some probability threshold (the ...
http://arxiv.org/pdf/2310.01693v1
2023-10-02T23:16:25
2023-10-02T23:16:25
{ "title": [ 0.017593384, -0.0053901672, -0.013656616, -0.040252686, -0.019943237, -0.0061569214, -0.013305664, -0.0016670227, -0.00006240606, 0.046661377, -0.03930664, -0.0061912537, 0.004535675, -0.009056091, -0.052337646, -0.00045466423, -0.003664...
http://arxiv.org/abs/2310.01691v1
Zero-Shot Continuous Prompt Transfer: Generalizing Task Semantics Across Language Models
Zijun Wu, Yongkang Wu, Lili Mou
cs.CL, cs.AI
Prompt tuning in natural language processing (NLP) has become an increasingly popular method for adapting large language models to specific tasks. However, the transferability of these prompts, especially continuous prompts, between different models remains a challenge. In this work, we propose a zero-shot continuous p...
http://arxiv.org/pdf/2310.01691v1
2023-10-02T23:12:21
2023-10-02T23:12:21
{ "title": [ -0.024230957, -0.020187378, -0.034454346, -0.06616211, -0.009742737, -0.006778717, -0.057800293, 0.016601562, 0.012413025, 0.041015625, 0.0034332275, 0.005844116, 0.016662598, -0.014343262, -0.046295166, 0.023254395, 0.0010051727, 0....
http://arxiv.org/abs/2310.01688v1
One model to rule them all ? Towards End-to-End Joint Speaker Diarization and Speech Recognition
Samuele Cornell, Jee-weon Jung, Shinji Watanabe, Stefano Squartini
eess.AS, cs.CL, cs.SD
This paper presents a novel framework for joint speaker diarization (SD) and automatic speech recognition (ASR), named SLIDAR (sliding-window diarization-augmented recognition). SLIDAR can process arbitrary length inputs and can handle any number of speakers, effectively solving ``who spoke what, when'' concurrently. S...
http://arxiv.org/pdf/2310.01688v1
2023-10-02T23:03:30
2023-10-02T23:03:30
{ "title": [ 0.004875183, 0.021911621, -0.038391113, -0.070495605, 0.023895264, -0.006958008, -0.038757324, -0.0048217773, -0.022781372, 0.047790527, -0.013496399, 0.01134491, 0.010002136, -0.018630981, -0.024368286, 0.0049552917, -0.0141067505, ...
http://arxiv.org/abs/2310.04444v2
What's the Magic Word? A Control Theory of LLM Prompting
Aman Bhargava, Cameron Witkowski, Manav Shah, Matt Thomson
cs.CL, cs.AI, cs.LG, cs.NE
Prompt engineering is effective and important in the deployment of LLMs but is poorly understood mathematically. Here, we formalize prompt engineering as an optimal control problem on LLMs -- where the prompt is considered a control variable for modulating the output distribution of the LLM. Within this framework, we a...
http://arxiv.org/pdf/2310.04444v2
2023-10-10T10:15:14
2023-10-02T22:35:40
{ "title": [ -0.038330078, -0.007904053, -0.040283203, -0.025558472, -0.016937256, -0.00762558, -0.03842163, -0.020126343, -0.039093018, 0.005191803, -0.027023315, 0.02583313, -0.02468872, -0.03289795, -0.00831604, -0.0061035156, -0.0021095276, -...
http://arxiv.org/abs/2310.04443v2
Human Mobility Question Answering (Vision Paper)
Hao Xue, Flora D. Salim
cs.CL, cs.AI
Question answering (QA) systems have attracted much attention from the artificial intelligence community as they can learn to answer questions based on the given knowledge source (e.g., images in visual question answering). However, the research into question answering systems with human mobility data remains unexplore...
http://arxiv.org/pdf/2310.04443v2
2023-10-13T05:07:49
2023-10-02T21:24:26
{ "title": [ 0.0016431808, 0.0011005402, -0.02432251, -0.04714966, 0.0013818741, -0.01448822, -0.011459351, -0.013877869, -0.006389618, 0.044921875, -0.03869629, 0.003118515, 0.027282715, -0.006801605, -0.028640747, -0.0020160675, 0.014312744, 0....
http://arxiv.org/abs/2310.01627v1
VAL: Interactive Task Learning with GPT Dialog Parsing
Lane Lawley, Christopher J. MacLellan
cs.HC, cs.AI, cs.CL
Reinforcement learning often requires millions of examples to produce static, black-box models. In contrast, interactive task learning (ITL) emphasizes incremental knowledge acquisition from limited instruction provided by humans in modalities such as natural language. However, in practice, ITL systems often suffers fr...
http://arxiv.org/pdf/2310.01627v1
2023-10-02T20:45:41
2023-10-02T20:45:41
{ "title": [ -0.0011548996, 0.0026931763, -0.032928467, -0.060699463, -0.009391785, -0.0121154785, -0.066345215, -0.02482605, -0.007633209, 0.009483337, -0.028305054, 0.015434265, -0.047332764, -0.016403198, -0.03302002, -0.023376465, -0.020446777, ...
http://arxiv.org/abs/2310.01603v1
A Review of Digital Learning Environments for Teaching Natural Language Processing in K-12 Education
Xiaoyi Tian, Kristy Elizabeth Boyer
cs.CL, cs.AI, cs.HC
Natural Language Processing (NLP) plays a significant role in our daily lives and has become an essential part of Artificial Intelligence (AI) education in K-12. As children grow up with NLP-powered applications, it is crucial to introduce NLP concepts to them, fostering their understanding of language processing, lang...
http://arxiv.org/pdf/2310.01603v1
2023-10-02T19:54:30
2023-10-02T19:54:30
{ "title": [ -0.011703491, -0.010292053, -0.025100708, -0.034118652, -0.003774643, -0.01146698, -0.012237549, 0.0007252693, -0.027908325, 0.02758789, -0.026626587, 0.010597229, -0.00025320053, -0.02305603, -0.02949524, -0.027374268, -0.04147339, ...
http://arxiv.org/abs/2310.01568v1
Defending Against Authorship Identification Attacks
Haining Wang
cs.CL, cs.CY
Authorship identification has proven unsettlingly effective in inferring the identity of the author of an unsigned document, even when sensitive personal information has been carefully omitted. In the digital era, individuals leave a lasting digital footprint through their written content, whether it is posted on socia...
http://arxiv.org/pdf/2310.01568v1
2023-10-02T19:03:11
2023-10-02T19:03:11
{ "title": [ -0.022155762, 0.010757446, -0.03878784, -0.05508423, 0.01689148, 0.0022697449, -0.03479004, -0.01524353, -0.05895996, 0.031951904, 0.018035889, 0.02986145, -0.023635864, -0.0211792, -0.025650024, -0.0007314682, -0.024353027, 0.043640...
http://arxiv.org/abs/2310.01558v1
Making Retrieval-Augmented Language Models Robust to Irrelevant Context
Ori Yoran, Tomer Wolfson, Ori Ram, Jonathan Berant
cs.CL, cs.AI
Retrieval-augmented language models (RALMs) hold promise to produce language understanding systems that are are factual, efficient, and up-to-date. An important desideratum of RALMs, is that retrieved information helps model performance when it is relevant, and does not harm performance when it is not. This is particul...
http://arxiv.org/pdf/2310.01558v1
2023-10-02T18:52:35
2023-10-02T18:52:35
{ "title": [ 0.0074272156, 0.008476257, -0.016174316, -0.06137085, 0.01071167, 0.0032157898, -0.046020508, -0.028121948, 0.025939941, 0.05230713, -0.015823364, -0.005405426, 0.024856567, -0.005455017, -0.07354736, 0.020858765, -0.02168274, 0.0293...
http://arxiv.org/abs/2310.01415v2
GPT-Driver: Learning to Drive with GPT
Jiageng Mao, Yuxi Qian, Hang Zhao, Yue Wang
cs.CV, cs.AI, cs.CL, cs.RO
We present a simple yet effective approach that can transform the OpenAI GPT-3.5 model into a reliable motion planner for autonomous vehicles. Motion planning is a core challenge in autonomous driving, aiming to plan a driving trajectory that is safe and comfortable. Existing motion planners predominantly leverage heur...
http://arxiv.org/pdf/2310.01415v2
2023-10-16T18:33:10
2023-10-02T17:59:57
{ "title": [ 0.0018472672, 0.0033073425, -0.025360107, 0.00077819824, -0.015007019, -0.028457642, -0.0034065247, -0.04949951, -0.036712646, 0.018173218, -0.0579834, 0.011222839, -0.016113281, -0.031280518, -0.0041389465, -0.028579712, -0.004299164, ...
http://arxiv.org/abs/2310.01405v3
Representation Engineering: A Top-Down Approach to AI Transparency
Andy Zou, Long Phan, Sarah Chen, James Campbell, Phillip Guo, Richard Ren, Alexander Pan, Xuwang Yin, Mantas Mazeika, Ann-Kathrin Dombrowski, Shashwat Goel, Nathaniel Li, Michael J. Byun, Zifan Wang, Alex Mallen, Steven Basart, Sanmi Koyejo, Dawn Song, Matt Fredrikson, J. Zico Kolter, Dan Hendrycks
cs.LG, cs.AI, cs.CL, cs.CV, cs.CY
In this paper, we identify and characterize the emerging area of representation engineering (RepE), an approach to enhancing the transparency of AI systems that draws on insights from cognitive neuroscience. RepE places population-level representations, rather than neurons or circuits, at the center of analysis, equipp...
http://arxiv.org/pdf/2310.01405v3
2023-10-10T08:00:53
2023-10-02T17:59:07
{ "title": [ -0.02067566, -0.0015716553, -0.033081055, -0.046539307, 0.010169983, -0.004776001, -0.0065345764, -0.022354126, 0.0029792786, 0.023986816, -0.026367188, -0.0008406639, -0.0012788773, -0.025634766, -0.036224365, -0.012619019, -0.008865356...
http://arxiv.org/abs/2310.01387v1
It's MBR All the Way Down: Modern Generation Techniques Through the Lens of Minimum Bayes Risk
Amanda Bertsch, Alex Xie, Graham Neubig, Matthew R. Gormley
cs.CL
Minimum Bayes Risk (MBR) decoding is a method for choosing the outputs of a machine learning system based not on the output with the highest probability, but the output with the lowest risk (expected error) among multiple candidates. It is a simple but powerful method: for an additional cost at inference time, MBR prov...
http://arxiv.org/pdf/2310.01387v1
2023-10-02T17:47:10
2023-10-02T17:47:10
{ "title": [ -0.0099487305, -0.034301758, -0.021316528, -0.06100464, -0.010116577, -0.0074272156, -0.026733398, -0.0055656433, -0.030792236, 0.0068244934, -0.013908386, 0.035339355, -0.02583313, -0.03817749, 0.0013065338, -0.008674622, 0.004508972, ...
http://arxiv.org/abs/2310.01386v1
Who is ChatGPT? Benchmarking LLMs' Psychological Portrayal Using PsychoBench
Jen-tse Huang, Wenxuan Wang, Eric John Li, Man Ho Lam, Shujie Ren, Youliang Yuan, Wenxiang Jiao, Zhaopeng Tu, Michael R. Lyu
cs.CL
Large Language Models (LLMs) have recently showcased their remarkable capacities, not only in natural language processing tasks but also across diverse domains such as clinical medicine, legal consultation, and education. LLMs become more than mere applications, evolving into assistants capable of addressing diverse us...
http://arxiv.org/pdf/2310.01386v1
2023-10-02T17:46:09
2023-10-02T17:46:09
{ "title": [ -0.042236328, -0.016677856, -0.017700195, 0.0127334595, -0.07373047, 0.011650085, -0.05847168, -0.022750854, -0.042907715, 0.043060303, -0.038909912, 0.046569824, 0.002855301, 0.00036025047, 0.0124053955, -0.018829346, -0.005760193, ...
http://arxiv.org/abs/2310.01382v1
Compressing LLMs: The Truth is Rarely Pure and Never Simple
Ajay Jaiswal, Zhe Gan, Xianzhi Du, Bowen Zhang, Zhangyang Wang, Yinfei Yang
cs.CL, cs.LG
Despite their remarkable achievements, modern Large Language Models (LLMs) encounter exorbitant computational and memory footprints. Recently, several works have shown significant success in training-free and data-free compression (pruning and quantization) of LLMs achieving 50-60% sparsity and reducing the bit-width d...
http://arxiv.org/pdf/2310.01382v1
2023-10-02T17:42:37
2023-10-02T17:42:37
{ "title": [ -0.035369873, -0.009460449, -0.05026245, -0.012260437, -0.0317688, -0.01838684, -0.040283203, -0.014038086, -0.0791626, 0.04067993, -0.011123657, 0.0020542145, 0.00207901, -0.061767578, -0.0022907257, -0.048950195, 0.01285553, 0.0140...
http://arxiv.org/abs/2310.01381v2
DiffAR: Denoising Diffusion Autoregressive Model for Raw Speech Waveform Generation
Roi Benita, Michael Elad, Joseph Keshet
cs.SD, cs.CL, eess.AS
Diffusion models have recently been shown to be relevant for high-quality speech generation. Most work has been focused on generating spectrograms, and as such, they further require a subsequent model to convert the spectrogram to a waveform (i.e., a vocoder). This work proposes a diffusion probabilistic end-to-end mod...
http://arxiv.org/pdf/2310.01381v2
2023-11-06T18:55:18
2023-10-02T17:42:22
{ "title": [ -0.02609253, 0.008125305, -0.07751465, -0.07287598, 0.014785767, 0.0050354004, -0.054748535, 0.0101623535, 0.021194458, 0.04425049, -0.009010315, 0.0065994263, 0.05505371, -0.037963867, -0.036468506, 0.02558899, -0.0031719208, 0.0401...
http://arxiv.org/abs/2310.01377v1
UltraFeedback: Boosting Language Models with High-quality Feedback
Ganqu Cui, Lifan Yuan, Ning Ding, Guanming Yao, Wei Zhu, Yuan Ni, Guotong Xie, Zhiyuan Liu, Maosong Sun
cs.CL, cs.AI, cs.LG
Reinforcement learning from human feedback (RLHF) has become a pivot technique in aligning large language models (LLMs) with human preferences. In RLHF practice, preference data plays a crucial role in bridging human proclivity and LLMs. However, the scarcity of diverse, naturalistic datasets of human preferences on LL...
http://arxiv.org/pdf/2310.01377v1
2023-10-02T17:40:01
2023-10-02T17:40:01
{ "title": [ -0.01777649, -0.0031528473, -0.025222778, -0.05819702, 0.0063552856, 0.0061302185, -0.06933594, -0.019332886, -0.003862381, 0.04208374, -0.0026168823, 0.00031113625, 0.008872986, -0.020431519, -0.04901123, 0.019104004, 0.0011253357, ...
http://arxiv.org/abs/2310.01361v1
GenSim: Generating Robotic Simulation Tasks via Large Language Models
Lirui Wang, Yiyang Ling, Zhecheng Yuan, Mohit Shridhar, Chen Bao, Yuzhe Qin, Bailin Wang, Huazhe Xu, Xiaolong Wang
cs.LG, cs.CL, cs.CV, cs.RO
Collecting large amounts of real-world interaction data to train general robotic policies is often prohibitively expensive, thus motivating the use of simulation data. However, existing methods for data generation have generally focused on scene-level diversity (e.g., object instances and poses) rather than task-level ...
http://arxiv.org/pdf/2310.01361v1
2023-10-02T17:23:48
2023-10-02T17:23:48
{ "title": [ -0.041931152, -0.023757935, -0.030670166, -0.036987305, -0.013916016, 0.007160187, -0.055145264, -0.013542175, -0.0006432533, 0.022064209, -0.011413574, 0.01399231, -0.006969452, -0.021560669, -0.014961243, 0.0066375732, -0.0067214966, ...
http://arxiv.org/abs/2310.01352v3
RA-DIT: Retrieval-Augmented Dual Instruction Tuning
Xi Victoria Lin, Xilun Chen, Mingda Chen, Weijia Shi, Maria Lomeli, Rich James, Pedro Rodriguez, Jacob Kahn, Gergely Szilvasy, Mike Lewis, Luke Zettlemoyer, Scott Yih
cs.CL, cs.AI
Retrieval-augmented language models (RALMs) improve performance by accessing long-tail and up-to-date knowledge from external data stores, but are challenging to build. Existing approaches require either expensive retrieval-specific modifications to LM pre-training or use post-hoc integration of the data store that lea...
http://arxiv.org/pdf/2310.01352v3
2023-11-05T06:25:55
2023-10-02T17:16:26
{ "title": [ -0.0423584, -0.017700195, -0.011329651, -0.064575195, -0.0071983337, 0.017822266, -0.03277588, 0.012825012, 0.0028457642, 0.041503906, -0.036987305, 0.021652222, -0.0073394775, -0.0036144257, -0.05026245, 0.008522034, 0.0064926147, 0...
http://arxiv.org/abs/2310.01339v2
Improving Dialogue Management: Quality Datasets vs Models
Miguel Ángel Medina-Ramírez, Cayetano Guerra-Artal, Mario Hernández-Tejera
cs.CL, cs.AI
Task-oriented dialogue systems (TODS) have become crucial for users to interact with machines and computers using natural language. One of its key components is the dialogue manager, which guides the conversation towards a good goal for the user by providing the best possible response. Previous works have proposed rule...
http://arxiv.org/pdf/2310.01339v2
2023-10-23T17:15:55
2023-10-02T17:02:57
{ "title": [ 0.0028247833, -0.0061836243, -0.03781128, -0.0803833, -0.0011949539, -0.0038700104, -0.062927246, -0.0038337708, -0.007019043, 0.018997192, -0.0048179626, -0.0030269623, -0.04360962, -0.023239136, -0.05041504, -0.0068473816, -0.02406311,...
http://arxiv.org/abs/2310.01469v2
LLM Lies: Hallucinations are not Bugs, but Features as Adversarial Examples
Jia-Yu Yao, Kun-Peng Ning, Zhen-Hui Liu, Mu-Nan Ning, Li Yuan
cs.CL, cs.AI
Large Language Models (LLMs), including GPT-3.5, LLaMA, and PaLM, seem to be knowledgeable and able to adapt to many tasks. However, we still can not completely trust their answer, since LLMs suffer from hallucination--fabricating non-existent facts to cheat users without perception. And the reasons for their existence...
http://arxiv.org/pdf/2310.01469v2
2023-10-04T17:53:49
2023-10-02T17:01:56
{ "title": [ -0.002910614, -0.01550293, -0.00793457, 0.0088272095, -0.042388916, -0.015426636, -0.0008215904, -0.047821045, -0.036071777, 0.009803772, 0.0064353943, 0.01991272, -0.03277588, -0.05532837, -0.003162384, -0.022033691, -0.01550293, 0....
http://arxiv.org/abs/2310.01468v2
The Entity-Deduction Arena: A playground for probing the conversational reasoning and planning capabilities of LLMs
Yizhe Zhang, Jiarui Lu, Navdeep Jaitly
cs.CL, cs.AI, cs.HC, cs.LG
Large language models (LLMs) are effective at answering questions that are clearly asked. However, when faced with ambiguous queries they can act unpredictably and produce incorrect outputs. This underscores the need for the development of intelligent agents capable of asking clarification questions to resolve ambiguit...
http://arxiv.org/pdf/2310.01468v2
2023-10-04T05:40:10
2023-10-02T16:55:37
{ "title": [ -0.03201294, -0.003435135, -0.05429077, -0.02722168, -0.016220093, -0.00844574, -0.05770874, -0.018661499, 0.0043182373, -0.019622803, -0.021453857, 0.007083893, -0.019729614, -0.05343628, 0.020050049, -0.024658203, -0.02178955, 0.02...
http://arxiv.org/abs/2310.01334v1
Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy
Pingzhi Li, Zhenyu Zhang, Prateek Yadav, Yi-Lin Sung, Yu Cheng, Mohit Bansal, Tianlong Chen
cs.LG, cs.AI, cs.CL
Sparsely activated Mixture-of-Experts (SMoE) has shown promise to scale up the learning capacity of neural networks, however, they have issues like (a) High Memory Usage, due to duplication of the network layers into multiple copies as experts; and (b) Redundancy in Experts, as common learning-based routing policies su...
http://arxiv.org/pdf/2310.01334v1
2023-10-02T16:51:32
2023-10-02T16:51:32
{ "title": [ -0.017471313, 0.010726929, -0.052093506, -0.03591919, 0.014533997, -0.012756348, 0.008010864, 0.018035889, -0.090148926, 0.021392822, -0.0015821457, 0.04510498, -0.043701172, -0.036193848, 0.01878357, -0.049957275, 0.021316528, -0.00...
http://arxiv.org/abs/2310.01329v1
BTR: Binary Token Representations for Efficient Retrieval Augmented Language Models
Qingqing Cao, Sewon Min, Yizhong Wang, Hannaneh Hajishirzi
cs.CL, cs.AI
Retrieval augmentation addresses many critical problems in large language models such as hallucination, staleness, and privacy leaks. However, running retrieval-augmented language models (LMs) is slow and difficult to scale due to processing large amounts of retrieved text. We introduce binary token representations (BT...
http://arxiv.org/pdf/2310.01329v1
2023-10-02T16:48:47
2023-10-02T16:48:47
{ "title": [ -0.010673523, -0.014015198, -0.028244019, -0.07159424, 0.0072402954, 0.012969971, -0.044189453, 0.0050735474, 0.015777588, 0.051239014, -0.0040740967, 0.0023288727, 0.026443481, -0.0062675476, -0.05886841, 0.027816772, -0.022537231, ...
http://arxiv.org/abs/2310.01467v1
FedBPT: Efficient Federated Black-box Prompt Tuning for Large Language Models
Jingwei Sun, Ziyue Xu, Hongxu Yin, Dong Yang, Daguang Xu, Yiran Chen, Holger R. Roth
cs.CL, cs.AI
Pre-trained language models (PLM) have revolutionized the NLP landscape, achieving stellar performances across diverse tasks. These models, while benefiting from vast training data, often require fine-tuning on specific data to cater to distinct downstream tasks. However, this data adaptation process has inherent secur...
http://arxiv.org/pdf/2310.01467v1
2023-10-02T16:43:14
2023-10-02T16:43:14
{ "title": [ -0.015617371, -0.013076782, -0.042419434, -0.097595215, 0.0019130707, 0.009307861, -0.088134766, 0.018356323, -0.0038375854, 0.02607727, -0.034240723, 0.01612854, 0.01763916, -0.0028152466, -0.031677246, 0.0076408386, -0.009788513, 0...
http://arxiv.org/abs/2310.01320v3
Avalon's Game of Thoughts: Battle Against Deception through Recursive Contemplation
Shenzhi Wang, Chang Liu, Zilong Zheng, Siyuan Qi, Shuo Chen, Qisen Yang, Andrew Zhao, Chaofei Wang, Shiji Song, Gao Huang
cs.AI, cs.CL, cs.CY, cs.LG, cs.MA
Recent breakthroughs in large language models (LLMs) have brought remarkable success in the field of LLM-as-Agent. Nevertheless, a prevalent assumption is that the information processed by LLMs is consistently honest, neglecting the pervasive deceptive or misleading information in human society and AI-generated content...
http://arxiv.org/pdf/2310.01320v3
2023-10-24T12:51:28
2023-10-02T16:27:36
{ "title": [ 0.03479004, -0.02532959, -0.0037059784, 0.0132751465, -0.040405273, -0.024978638, 0.011138916, -0.051818848, -0.010795593, 0.009162903, 0.0077171326, -0.016464233, 0.02848816, -0.064453125, 0.004875183, -0.030212402, 0.015357971, 0.0...
http://arxiv.org/abs/2310.01307v2
On the Generalization of Training-based ChatGPT Detection Methods
Han Xu, Jie Ren, Pengfei He, Shenglai Zeng, Yingqian Cui, Amy Liu, Hui Liu, Jiliang Tang
cs.CL, cs.AI, cs.LG
ChatGPT is one of the most popular language models which achieve amazing performance on various natural language tasks. Consequently, there is also an urgent need to detect the texts generated ChatGPT from human written. One of the extensively studied methods trains classification models to distinguish both. However, e...
http://arxiv.org/pdf/2310.01307v2
2023-10-03T16:40:35
2023-10-02T16:13:08
{ "title": [ 0.0033473969, 0.0061035156, -0.0010986328, -0.022567749, 0.020507812, 0.001499176, -0.013473511, -0.033172607, -0.016296387, 0.020095825, -0.013496399, 0.017074585, -0.019470215, -0.016098022, -0.028656006, 0.0015792847, -0.044036865, ...
http://arxiv.org/abs/2310.01299v1
Generating Explanations in Medical Question-Answering by Expectation Maximization Inference over Evidence
Wei Sun, Mingxiao Li, Damien Sileo, Jesse Davis, Marie-Francine Moens
cs.CL, cs.AI
Medical Question Answering~(medical QA) systems play an essential role in assisting healthcare workers in finding answers to their questions. However, it is not sufficient to merely provide answers by medical QA systems because users might want explanations, that is, more analytic statements in natural language that de...
http://arxiv.org/pdf/2310.01299v1
2023-10-02T16:00:37
2023-10-02T16:00:37
{ "title": [ -0.009483337, -0.0015907288, -0.024169922, -0.06506348, -0.010314941, 0.00944519, -0.033477783, -0.0075187683, -0.019699097, 0.012397766, 0.016098022, 0.0078125, -0.018630981, -0.030044556, -0.03253174, -0.0067596436, 0.0006532669, 0...
http://arxiv.org/abs/2310.01297v1
Co-audit: tools to help humans double-check AI-generated content
Andrew D. Gordon, Carina Negreanu, José Cambronero, Rasika Chakravarthy, Ian Drosos, Hao Fang, Bhaskar Mitra, Hannah Richardson, Advait Sarkar, Stephanie Simmons, Jack Williams, Ben Zorn
cs.HC, cs.AI, cs.CL, cs.PL
Users are increasingly being warned to check AI-generated content for correctness. Still, as LLMs (and other generative models) generate more complex output, such as summaries, tables, or code, it becomes harder for the user to audit or evaluate the output for quality or correctness. Hence, we are seeing the emergence ...
http://arxiv.org/pdf/2310.01297v1
2023-10-02T15:59:10
2023-10-02T15:59:10
{ "title": [ -0.0045318604, -0.012390137, -0.032073975, -0.013183594, -0.027786255, -0.014717102, 0.0065956116, -0.011009216, -0.046142578, 0.054382324, -0.05001831, 0.0357666, 0.038879395, -0.010017395, -0.0013437271, -0.03161621, 0.008514404, 0...
http://arxiv.org/abs/2310.01290v1
Knowledge Crosswords: Geometric Reasoning over Structured Knowledge with Large Language Models
Wenxuan Ding, Shangbin Feng, Yuhan Liu, Zhaoxuan Tan, Vidhisha Balachandran, Tianxing He, Yulia Tsvetkov
cs.CL, cs.AI
Large language models (LLMs) are widely adopted in knowledge-intensive tasks and have achieved impressive performance thanks to their knowledge abilities. While LLMs have demonstrated outstanding performance on atomic or linear (multi-hop) QA tasks, whether they can reason in knowledge-rich scenarios with interweaving ...
http://arxiv.org/pdf/2310.01290v1
2023-10-02T15:43:53
2023-10-02T15:43:53
{ "title": [ -0.06463623, -0.051605225, -0.048217773, -0.070129395, -0.03125, 0.015113831, -0.048675537, 0.03463745, 0.025619507, 0.020996094, -0.03289795, -0.0076446533, 0.007133484, -0.012290955, -0.031280518, -0.0020637512, 0.012390137, 0.0135...
http://arxiv.org/abs/2310.01271v2
LEEC: A Legal Element Extraction Dataset with an Extensive Domain-Specific Label System
Xue Zongyue, Liu Huanghai, Hu Yiran, Kong Kangle, Wang Chenlu, Liu Yun, Shen Weixing
cs.CL, cs.IR
As a pivotal task in natural language processing, element extraction has gained significance in the legal domain. Extracting legal elements from judicial documents helps enhance interpretative and analytical capacities of legal cases, and thereby facilitating a wide array of downstream applications in various domains o...
http://arxiv.org/pdf/2310.01271v2
2023-10-10T04:34:56
2023-10-02T15:16:31
{ "title": [ -0.055603027, -0.013084412, -0.059417725, -0.030334473, -0.0059890747, -0.009765625, -0.042236328, -0.0033168793, 0.014404297, 0.020858765, -0.027893066, 0.011802673, 0.0074653625, -0.02218628, -0.03277588, 0.006565094, -0.02734375, ...
http://arxiv.org/abs/2310.01260v1
SPELL: Semantic Prompt Evolution based on a LLM
Yujian Betterest Li, Kai Wu
cs.CL, cs.AI
Prompt engineering is a new paradigm for enhancing the performance of trained neural network models. For optimizing text-style prompts, existing methods usually individually operate small portions of a text step by step, which either breaks the fluency or could not globally adjust a prompt. Since large language models ...
http://arxiv.org/pdf/2310.01260v1
2023-10-02T14:51:16
2023-10-02T14:51:16
{ "title": [ -0.052642822, -0.01928711, -0.05117798, -0.021270752, -0.04058838, -0.00504303, -0.058013916, -0.008728027, 0.009994507, 0.017974854, -0.03149414, 0.013168335, 0.0127334595, -0.030426025, -0.0079193115, 0.0052757263, -0.0077323914, 0...
http://arxiv.org/abs/2310.01248v2
Improving Emotional Expression and Cohesion in Image-Based Playlist Description and Music Topics: A Continuous Parameterization Approach
Yuelyu Ji, Yuheng Song, Wei Wang, Ruoyi Xu, Zhongqian Xie, Huiyun Liu
cs.CL
Text generation in image-based platforms, particularly for music-related content, requires precise control over text styles and the incorporation of emotional expression. However, existing approaches often need help to control the proportion of external factors in generated text and rely on discrete inputs, lacking con...
http://arxiv.org/pdf/2310.01248v2
2023-10-12T20:10:37
2023-10-02T14:32:07
{ "title": [ 0.028808594, 0.028778076, -0.044799805, -0.04324341, 0.032989502, -0.030548096, -0.0070495605, -0.029800415, -0.016738892, 0.051574707, -0.0102005005, 0.024932861, 0.020248413, -0.021530151, -0.021514893, -0.01171875, 0.002904892, -0...
http://arxiv.org/abs/2310.01217v1
ScaLearn: Simple and Highly Parameter-Efficient Task Transfer by Learning to Scale
Markus Frohmann, Carolin Holtermann, Shahed Masoudian, Anne Lauscher, Navid Rekabsaz
cs.LG, cs.AI, cs.CL
Multi-task learning (MTL) has shown considerable practical benefits, particularly when using pre-trained language models (PLMs). While this is commonly achieved by simultaneously learning $n$ tasks under a joint optimization procedure, recent methods such as AdapterFusion structure the problem into two distinct stages:...
http://arxiv.org/pdf/2310.01217v1
2023-10-02T14:01:36
2023-10-02T14:01:36
{ "title": [ -0.02456665, -0.00831604, -0.057495117, -0.042266846, 0.0105896, -0.019165039, -0.029052734, -0.014694214, 0.0033950806, 0.008811951, 0.0034618378, 0.0076293945, -0.010772705, -0.039489746, -0.0037841797, 0.016448975, 0.012031555, 0....
http://arxiv.org/abs/2310.01208v1
Label Supervised LLaMA Finetuning
Zongxi Li, Xianming Li, Yuzhang Liu, Haoran Xie, Jing Li, Fu-lee Wang, Qing Li, Xiaoqin Zhong
cs.CL
The recent success of Large Language Models (LLMs) has gained significant attention in both academia and industry. Substantial efforts have been made to enhance the zero- and few-shot generalization capabilities of open-source LLMs through finetuning. Currently, the prevailing approach is instruction-tuning, which trai...
http://arxiv.org/pdf/2310.01208v1
2023-10-02T13:53:03
2023-10-02T13:53:03
{ "title": [ -0.00046753883, 0.0009622574, -0.032318115, -0.030212402, -0.039733887, -0.0118637085, -0.03555298, 0.00680542, -0.07348633, 0.044799805, -0.008155823, 0.025665283, -0.00042152405, -0.0368042, 0.053619385, -0.0049324036, -0.0007686615, ...
http://arxiv.org/abs/2310.01206v2
appjsonify: An Academic Paper PDF-to-JSON Conversion Toolkit
Atsuki Yamaguchi, Terufumi Morishita
cs.CL, cs.AI
We present appjsonify, a Python-based PDF-to-JSON conversion toolkit for academic papers. It parses a PDF file using several visual-based document layout analysis models and rule-based text processing approaches. appjsonify is a flexible tool that allows users to easily configure the processing pipeline to handle a spe...
http://arxiv.org/pdf/2310.01206v2
2023-10-03T13:19:40
2023-10-02T13:48:16
{ "title": [ -0.023742676, 0.007797241, -0.0657959, -0.050109863, 0.0071411133, -0.009719849, -0.041748047, 0.0040016174, -0.030838013, 0.048339844, 0.036834717, -0.0036792755, 0.039733887, -0.045684814, -0.010597229, -0.008148193, 0.015037537, 0...
http://arxiv.org/abs/2310.01188v1
Quantifying the Plausibility of Context Reliance in Neural Machine Translation
Gabriele Sarti, Grzegorz Chrupała, Malvina Nissim, Arianna Bisazza
cs.CL, cs.AI, cs.HC, cs.LG, I.2.7
Establishing whether language models can use contextual information in a human-plausible way is important to ensure their safe adoption in real-world settings. However, the questions of when and which parts of the context affect model generations are typically tackled separately, and current plausibility evaluations ar...
http://arxiv.org/pdf/2310.01188v1
2023-10-02T13:26:43
2023-10-02T13:26:43
{ "title": [ -0.002565384, 0.00095796585, -0.005001068, -0.053344727, 0.0013523102, 0.0023422241, -0.019485474, -0.015113831, -0.01889038, 0.0463562, -0.010101318, 0.006565094, 0.025253296, -0.012687683, -0.03491211, 0.01209259, -0.014305115, 0.0...
http://arxiv.org/abs/2310.01459v1
NarrativePlay: Interactive Narrative Understanding
Runcong Zhao, Wenjia Zhang, Jiazheng Li, Lixing Zhu, Yanran Li, Yulan He, Lin Gui
cs.CL, cs.AI, cs.HC
In this paper, we introduce NarrativePlay, a novel system that allows users to role-play a fictional character and interact with other characters in narratives such as novels in an immersive environment. We leverage Large Language Models (LLMs) to generate human-like responses, guided by personality traits extracted fr...
http://arxiv.org/pdf/2310.01459v1
2023-10-02T13:24:00
2023-10-02T13:24:00
{ "title": [ -0.03491211, -0.018600464, -0.036743164, -0.05328369, -0.0076942444, 0.0026245117, -0.026931763, -0.00894165, -0.008018494, -0.0003979206, -0.0088272095, -0.006629944, -0.02684021, -0.040496826, -0.021575928, -0.014053345, -0.014663696, ...
http://arxiv.org/abs/2310.01138v1
Target-Aware Contextual Political Bias Detection in News
Iffat Maab, Edison Marrese-Taylor, Yutaka Matsuo
cs.CL
Media bias detection requires comprehensive integration of information derived from multiple news sources. Sentence-level political bias detection in news is no exception, and has proven to be a challenging task that requires an understanding of bias in consideration of the context. Inspired by the fact that humans exh...
http://arxiv.org/pdf/2310.01138v1
2023-10-02T12:25:05
2023-10-02T12:25:05
{ "title": [ 0.019104004, 0.012359619, -0.031097412, -0.04232788, 0.03100586, -0.016921997, 0.00945282, -0.035095215, -0.012916565, 0.03781128, -0.029953003, 0.01449585, -0.0060310364, 0.0046463013, -0.041656494, -0.011383057, -0.0289917, 0.04418...
http://arxiv.org/abs/2310.01132v1
Automated Evaluation of Classroom Instructional Support with LLMs and BoWs: Connecting Global Predictions to Specific Feedback
Jacob Whitehill, Jennifer LoCasale-Crouch
cs.CL, cs.AI
With the aim to provide teachers with more specific, frequent, and actionable feedback about their teaching, we explore how Large Language Models (LLMs) can be used to estimate ``Instructional Support'' domain scores of the CLassroom Assessment Scoring System (CLASS), a widely used observation protocol. We design a mac...
http://arxiv.org/pdf/2310.01132v1
2023-10-02T12:11:17
2023-10-02T12:11:17
{ "title": [ -0.038116455, -0.006324768, -0.026748657, -0.039154053, -0.023117065, 0.0030765533, -0.040496826, -0.009712219, 0.0012226105, 0.010894775, -0.0357666, 0.027709961, -0.0010299683, -0.014389038, -0.024795532, -0.03111267, -0.010108948, ...
http://arxiv.org/abs/2310.02281v1
End-to-End Continuous Speech Emotion Recognition in Real-life Customer Service Call Center Conversations
Yajing Feng, Laurence Devillers
eess.AS, cs.AI, cs.CL, cs.LG, cs.SD
Speech Emotion recognition (SER) in call center conversations has emerged as a valuable tool for assessing the quality of interactions between clients and agents. In contrast to controlled laboratory environments, real-life conversations take place under uncontrolled conditions and are subject to contextual factors tha...
http://arxiv.org/pdf/2310.02281v1
2023-10-02T11:53:48
2023-10-02T11:53:48
{ "title": [ 0.0039901733, 0.0022621155, -0.05886841, -0.066345215, 0.02305603, -0.022125244, -0.017074585, 0.027572632, 0.0074539185, 0.04876709, -0.021957397, -0.011543274, 0.040893555, -0.026443481, -0.04144287, 0.0013589859, -0.004180908, 0.0...
http://arxiv.org/abs/2310.01119v1
Text Data Augmentation in Low-Resource Settings via Fine-Tuning of Large Language Models
Jean Kaddour, Qi Liu
cs.CL, cs.LG
The in-context learning ability of large language models (LLMs) enables them to generalize to novel downstream tasks with relatively few labeled examples. However, they require enormous computational resources to be deployed. Alternatively, smaller models can solve specific tasks if fine-tuned with enough labeled examp...
http://arxiv.org/pdf/2310.01119v1
2023-10-02T11:49:05
2023-10-02T11:49:05
{ "title": [ -0.012420654, -0.010955811, -0.04876709, -0.0703125, 0.0016098022, -0.0016431808, -0.060821533, 0.011413574, 0.0077667236, 0.038848877, -0.027694702, 0.012573242, 0.01234436, -0.0011405945, -0.048980713, 0.012313843, -0.018157959, 0....
http://arxiv.org/abs/2310.01089v1
GraphText: Graph Reasoning in Text Space
Jianan Zhao, Le Zhuo, Yikang Shen, Meng Qu, Kai Liu, Michael Bronstein, Zhaocheng Zhu, Jian Tang
cs.CL, cs.LG
Large Language Models (LLMs) have gained the ability to assimilate human knowledge and facilitate natural language interactions with both humans and other LLMs. However, despite their impressive achievements, LLMs have not made significant advancements in the realm of graph machine learning. This limitation arises beca...
http://arxiv.org/pdf/2310.01089v1
2023-10-02T11:03:57
2023-10-02T11:03:57
{ "title": [ -0.043518066, -0.036071777, -0.03036499, -0.06274414, -0.014266968, 0.024215698, -0.042175293, 0.005443573, 0.0158844, 0.0024909973, -0.029937744, 0.011474609, -0.020446777, -0.025848389, -0.025039673, -0.0068626404, -0.0013360977, 0...
http://arxiv.org/abs/2310.01088v1
Towards human-like spoken dialogue generation between AI agents from written dialogue
Kentaro Mitsui, Yukiya Hono, Kei Sawada
cs.CL, cs.LG, cs.SD, eess.AS
The advent of large language models (LLMs) has made it possible to generate natural written dialogues between two agents. However, generating human-like spoken dialogues from these written dialogues remains challenging. Spoken dialogues have several unique characteristics: they frequently include backchannels and laugh...
http://arxiv.org/pdf/2310.01088v1
2023-10-02T11:03:20
2023-10-02T11:03:20
{ "title": [ -0.054229736, -0.009635925, -0.04638672, -0.031341553, -0.010772705, 0.0039367676, -0.032043457, 0.015930176, -0.0003979206, 0.02305603, -0.025390625, 0.0018386841, -0.013343811, -0.0070762634, -0.061309814, -0.0066947937, -0.02532959, ...
http://arxiv.org/abs/2310.01074v2
Back to the Future: Towards Explainable Temporal Reasoning with Large Language Models
Chenhan Yuan, Qianqian Xie, Jimin Huang, Sophia Ananiadou
cs.CL, cs.AI
Temporal reasoning is a crucial NLP task, providing a nuanced understanding of time-sensitive contexts within textual data. Although recent advancements in LLMs have demonstrated their potential in temporal reasoning, the predominant focus has been on tasks such as temporal expression and temporal relation extraction. ...
http://arxiv.org/pdf/2310.01074v2
2023-10-08T12:45:18
2023-10-02T10:35:23
{ "title": [ -0.0129852295, -0.019012451, -0.045410156, -0.11706543, 0.013412476, 0.0073890686, -0.058502197, 0.0385437, 0.013313293, 0.013580322, -0.010009766, -0.012268066, -0.0026474, -0.015197754, -0.042297363, -0.0013618469, -0.010948181, 0....
http://arxiv.org/abs/2310.01061v1
Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning
Linhao Luo, Yuan-Fang Li, Gholamreza Haffari, Shirui Pan
cs.CL, cs.AI
Large language models (LLMs) have demonstrated impressive reasoning abilities in complex tasks. However, they lack up-to-date knowledge and experience hallucinations during reasoning, which can lead to incorrect reasoning processes and diminish their performance and trustworthiness. Knowledge graphs (KGs), which captur...
http://arxiv.org/pdf/2310.01061v1
2023-10-02T10:14:43
2023-10-02T10:14:43
{ "title": [ -0.038726807, -0.02961731, -0.046081543, -0.103881836, 0.003320694, 0.03152466, -0.089538574, 0.023086548, 0.006225586, 0.0041007996, -0.014144897, -0.011146545, -0.01789856, -0.02532959, -0.023117065, -0.0034255981, -0.014472961, 0....
http://arxiv.org/abs/2310.01045v1
Tool-Augmented Reward Modeling
Lei Li, Yekun Chai, Shuohuan Wang, Yu Sun, Hao Tian, Ningyu Zhang, Hua Wu
cs.CL
Reward modeling (a.k.a., preference modeling) is instrumental for aligning large language models with human preferences, particularly within the context of reinforcement learning from human feedback (RLHF). While conventional reward models (RMs) have exhibited remarkable scalability, they oft struggle with fundamental ...
http://arxiv.org/pdf/2310.01045v1
2023-10-02T09:47:40
2023-10-02T09:47:40
{ "title": [ 0.000582695, 0.019561768, -0.010597229, -0.0026073456, -0.019226074, -0.014625549, -0.0052986145, -0.047424316, -0.022979736, 0.033325195, 0.0036811829, -0.039245605, -0.011291504, -0.052520752, -0.013534546, -0.0021152496, -0.0011234283...
http://arxiv.org/abs/2310.01041v1
Language Model Decoding as Direct Metrics Optimization
Haozhe Ji, Pei Ke, Hongning Wang, Minlie Huang
cs.CL
Despite the remarkable advances in language modeling, current mainstream decoding methods still struggle to generate texts that align with human texts across different aspects. In particular, sampling-based methods produce less-repetitive texts which are often disjunctive in discourse, while search-based methods mainta...
http://arxiv.org/pdf/2310.01041v1
2023-10-02T09:35:27
2023-10-02T09:35:27
{ "title": [ -0.057159424, -0.021896362, -0.016311646, -0.036956787, -0.043823242, 0.010719299, -0.07678223, -0.0054626465, -0.012649536, 0.062042236, 0.00818634, -0.005054474, 0.02835083, -0.01625061, -0.04864502, 0.027069092, 0.0060272217, 0.03...
http://arxiv.org/abs/2310.00996v1
ARN: A Comprehensive Framework and Dataset for Analogical Reasoning on Narratives
Zhivar Sourati, Filip Ilievski, Pia Sommerauer
cs.CL
Analogical reasoning is one of the prime abilities of humans and is linked to creativity and scientific discoveries. This ability has been studied extensively in natural language processing (NLP) as well as in cognitive psychology by proposing various benchmarks and evaluation setups. Yet, a substantial gap exists betw...
http://arxiv.org/pdf/2310.00996v1
2023-10-02T08:58:29
2023-10-02T08:58:29
{ "title": [ 0.0025100708, -0.017059326, -0.014190674, -0.06262207, 0.0016593933, 0.0064353943, -0.01940918, -0.012649536, 0.0049705505, 0.023239136, -0.0031852722, 0.015533447, 0.004966736, -0.01876831, -0.01751709, 0.011528015, -0.014587402, 0....
http://arxiv.org/abs/2310.00970v1
EALM: Introducing Multidimensional Ethical Alignment in Conversational Information Retrieval
Yiyao Yu, Junjie Wang, Yuxiang Zhang, Lin Zhang, Yujiu Yang, Tetsuya Sakai
cs.CL
Artificial intelligence (AI) technologies should adhere to human norms to better serve our society and avoid disseminating harmful or misleading information, particularly in Conversational Information Retrieval (CIR). Previous work, including approaches and datasets, has not always been successful or sufficiently robus...
http://arxiv.org/pdf/2310.00970v1
2023-10-02T08:22:34
2023-10-02T08:22:34
{ "title": [ 0.008735657, -0.0012664795, -0.053710938, -0.023376465, 0.0029525757, -0.03643799, -0.003932953, -0.017944336, 0.0068969727, 0.055236816, -0.020355225, 0.0076141357, 0.02999878, -0.01234436, -0.055877686, -0.0005455017, -0.036712646, ...
http://arxiv.org/abs/2310.00935v1
Resolving Knowledge Conflicts in Large Language Models
Yike Wang, Shangbin Feng, Heng Wang, Weijia Shi, Vidhisha Balachandran, Tianxing He, Yulia Tsvetkov
cs.CL
Large language models (LLMs) often encounter knowledge conflicts, scenarios where discrepancy arises between the internal parametric knowledge of LLMs and non-parametric information provided in the prompt context. In this work we ask what are the desiderata for LLMs when a knowledge conflict arises and whether existing...
http://arxiv.org/pdf/2310.00935v1
2023-10-02T06:57:45
2023-10-02T06:57:45
{ "title": [ -0.0395813, -0.02255249, -0.06762695, -0.091674805, 0.006778717, 0.009605408, -0.078308105, 0.032592773, 0.007270813, 0.023712158, -0.027282715, -0.019714355, 0.0010118484, -0.008087158, -0.046447754, -0.017562866, -0.0028247833, 0.0...
http://arxiv.org/abs/2310.01452v1
Fooling the Textual Fooler via Randomizing Latent Representations
Duy C. Hoang, Quang H. Nguyen, Saurav Manchanda, MinLong Peng, Kok-Seng Wong, Khoa D. Doan
cs.CL, cs.AI
Despite outstanding performance in a variety of NLP tasks, recent studies have revealed that NLP models are vulnerable to adversarial attacks that slightly perturb the input to cause the models to misbehave. Among these attacks, adversarial word-level perturbations are well-studied and effective attack strategies. Sinc...
http://arxiv.org/pdf/2310.01452v1
2023-10-02T06:57:25
2023-10-02T06:57:25
{ "title": [ -0.0077171326, 0.002752304, -0.021636963, -0.02218628, 0.008773804, -0.005569458, 0.0051879883, -0.03753662, -0.03842163, 0.027389526, -0.010864258, 0.03378296, -0.017791748, -0.019180298, -0.014320374, -0.008239746, -0.022216797, 0....
http://arxiv.org/abs/2310.00905v1
All Languages Matter: On the Multilingual Safety of Large Language Models
Wenxuan Wang, Zhaopeng Tu, Chang Chen, Youliang Yuan, Jen-tse Huang, Wenxiang Jiao, Michael R. Lyu
cs.CL, cs.AI
Safety lies at the core of developing and deploying large language models (LLMs). However, previous safety benchmarks only concern the safety in one language, e.g. the majority language in the pretraining data such as English. In this work, we build the first multilingual safety benchmark for LLMs, XSafety, in response...
http://arxiv.org/pdf/2310.00905v1
2023-10-02T05:23:34
2023-10-02T05:23:34
{ "title": [ -0.029083252, -0.01687622, -0.044067383, -0.08959961, 0.015563965, 0.00995636, -0.07043457, 0.014289856, -0.012886047, 0.044006348, 0.0035762787, 0.009529114, 0.016906738, -0.02218628, -0.030014038, 0.020019531, -0.013694763, 0.02949...
http://arxiv.org/abs/2310.00901v2
TADIS: Steering Models for Deep-Thinking about Demonstration Examples
Tianci Xue, Ziqi Wang, Yixia Li, Yun Chen, Guanhua Chen
cs.CL
Instruction tuning has been demonstrated that could significantly improve the zero-shot generalization capability to unseen tasks by an apparent margin. By incorporating additional context (e.g., task definition, examples) during the fine-tuning process, Large Language Models (LLMs) achieved much higher performance tha...
http://arxiv.org/pdf/2310.00901v2
2023-10-05T03:04:12
2023-10-02T04:42:53
{ "title": [ 0.020446777, -0.008026123, -0.016204834, -0.0546875, -0.00497818, 0.00010639429, -0.046783447, -0.031951904, 0.022659302, -0.005340576, -0.02861023, 0.0019874573, -0.010803223, -0.03616333, -0.012931824, -0.006362915, 0.0027008057, 0...
http://arxiv.org/abs/2310.00900v1
uSee: Unified Speech Enhancement and Editing with Conditional Diffusion Models
Muqiao Yang, Chunlei Zhang, Yong Xu, Zhongweiyang Xu, Heming Wang, Bhiksha Raj, Dong Yu
cs.SD, cs.AI, cs.CL, eess.AS
Speech enhancement aims to improve the quality of speech signals in terms of quality and intelligibility, and speech editing refers to the process of editing the speech according to specific user needs. In this paper, we propose a Unified Speech Enhancement and Editing (uSee) model with conditional diffusion models to ...
http://arxiv.org/pdf/2310.00900v1
2023-10-02T04:36:39
2023-10-02T04:36:39
{ "title": [ -0.015419006, 0.008132935, -0.06750488, -0.058502197, 0.033050537, -0.003616333, -0.020217896, 0.004432678, 0.005882263, 0.04067993, -0.00013422966, -0.006904602, 0.025054932, -0.039154053, -0.036743164, 0.014579773, -0.022766113, 0....
http://arxiv.org/abs/2310.00898v2
Enable Language Models to Implicitly Learn Self-Improvement From Data
Ziqi Wang, Le Hou, Tianjian Lu, Yuexin Wu, Yunxuan Li, Hongkun Yu, Heng Ji
cs.CL
Large Language Models (LLMs) have demonstrated remarkable capabilities in open-ended text generation tasks. However, the inherent open-ended nature of these tasks implies that there is always room for improvement in the quality of model responses. To address this challenge, various approaches have been proposed to enha...
http://arxiv.org/pdf/2310.00898v2
2023-10-05T22:34:59
2023-10-02T04:29:40
{ "title": [ 0.03427124, -0.016067505, -0.049804688, -0.0680542, -0.035186768, -0.021759033, -0.05380249, 0.026123047, -0.017211914, 0.029556274, -0.043518066, 0.0048294067, -0.010231018, -0.021850586, 0.010673523, -0.026977539, 0.01197052, 0.023...
http://arxiv.org/abs/2310.00892v1
No Offense Taken: Eliciting Offensiveness from Language Models
Anugya Srivastava, Rahul Ahuja, Rohith Mukku
cs.CL, cs.AI
This work was completed in May 2022. For safe and reliable deployment of language models in the real world, testing needs to be robust. This robustness can be characterized by the difficulty and diversity of the test cases we evaluate these models on. Limitations in human-in-the-loop test case generation has prompted...
http://arxiv.org/pdf/2310.00892v1
2023-10-02T04:17:35
2023-10-02T04:17:35
{ "title": [ -0.0043029785, -0.0038318634, -0.039886475, -0.03994751, 0.004878998, -0.024642944, -0.01991272, -0.0118637085, 0.01876831, 0.023361206, -0.011703491, -0.015296936, -0.0017652512, -0.014839172, -0.06378174, -0.007537842, -0.021575928, ...
http://arxiv.org/abs/2310.00867v2
(Dynamic) Prompting might be all you need to repair Compressed LLMs
Duc N. M Hoang, Minsik Cho, Thomas Merth, Mohammad Rastegari, Zhangyang Wang
cs.CL, cs.AI
Large language models (LLMs), while transformative for NLP, come with significant computational demands, underlining the need for efficient, training-free compression. Notably, despite the marked improvement in training-free compression for the largest of LLMs, our tests using LLaMA-7B and OPT-6.7b highlight a signific...
http://arxiv.org/pdf/2310.00867v2
2023-10-14T05:12:54
2023-10-02T03:12:06
{ "title": [ -0.0063934326, -0.04815674, -0.009147644, -0.066345215, -0.05731201, 0.012290955, -0.061767578, 0.030181885, -0.019317627, -0.0021591187, -0.028518677, -0.013298035, -0.021820068, -0.04220581, 0.01576233, -0.04800415, 0.031204224, -0...
http://arxiv.org/abs/2310.00863v1
Melody-conditioned lyrics generation via fine-tuning language model and its evaluation with ChatGPT
Zhe Zhang, Karol Lasocki, Yi Yu, Atsuhiro Takasu
cs.CL, cs.AI
We leverage character-level language models for syllable-level lyrics generation from symbolic melody. By fine-tuning a character-level pre-trained model, we integrate language knowledge into the beam search of a syllable-level Transformer generator. Using ChatGPT-based evaluations, we demonstrate enhanced coherence an...
http://arxiv.org/pdf/2310.00863v1
2023-10-02T02:53:29
2023-10-02T02:53:29
{ "title": [ -0.018234253, -0.03201294, -0.024032593, -0.03781128, -0.024887085, 0.006668091, -0.034606934, 0.029129028, -0.016540527, 0.042236328, -0.028442383, 0.018371582, -0.010574341, -0.00060606, -0.04232788, -0.008583069, -0.03036499, 0.05...
http://arxiv.org/abs/2310.02905v1
Use Your INSTINCT: INSTruction optimization usIng Neural bandits Coupled with Transformers
Xiaoqiang Lin, Zhaoxuan Wu, Zhongxiang Dai, Wenyang Hu, Yao Shu, See-Kiong Ng, Patrick Jaillet, Bryan Kian Hsiang Low
cs.LG, cs.AI, cs.CL
Large language models (LLMs) have shown remarkable instruction-following capabilities and achieved impressive performances in various applications. However, the performances of LLMs depend heavily on the instructions given to them, which are typically manually tuned with substantial human efforts. Recent work has used ...
http://arxiv.org/pdf/2310.02905v1
2023-10-02T02:01:16
2023-10-02T02:01:16
{ "title": [ -0.059051514, -0.007549286, -0.014564514, -0.022720337, -0.02923584, -0.0014410019, -0.014198303, -0.011665344, -0.02973938, 0.029006958, -0.055725098, 0.035247803, -0.0413208, 0.0027370453, -0.029144287, -0.025268555, -0.0049324036, ...
http://arxiv.org/abs/2310.00845v1
Application of frozen large-scale models to multimodal task-oriented dialogue
Tatsuki Kawamoto, Takuma Suzuki, Ko Miyama, Takumi Meguro, Tomohiro Takagi
cs.CL, cs.AI
In this study, we use the existing Large Language Models ENnhanced to See Framework (LENS Framework) to test the feasibility of multimodal task-oriented dialogues. The LENS Framework has been proposed as a method to solve computer vision tasks without additional training and with fixed parameters of pre-trained models....
http://arxiv.org/pdf/2310.00845v1
2023-10-02T01:42:28
2023-10-02T01:42:28
{ "title": [ -0.005947113, 0.0034065247, -0.047851562, -0.06488037, 0.0027694702, -0.016082764, -0.058441162, 0.0036697388, 0.0085372925, 0.032287598, 0.011436462, -0.025390625, 0.00025749207, -0.036102295, -0.04815674, -0.005973816, -0.007041931, ...
http://arxiv.org/abs/2310.00840v1
Error Norm Truncation: Robust Training in the Presence of Data Noise for Text Generation Models
Tianjian Li, Haoran Xu, Philipp Koehn, Daniel Khashabi, Kenton Murray
cs.CL
Text generation models are notoriously vulnerable to errors in the training data. With the wide-spread availability of massive amounts of web-crawled data becoming more commonplace, how can we enhance the robustness of models trained on a massive amount of noisy web-crawled text? In our work, we propose Error Norm Trun...
http://arxiv.org/pdf/2310.00840v1
2023-10-02T01:30:27
2023-10-02T01:30:27
{ "title": [ -0.03451538, 0.0033931732, -0.051452637, -0.02835083, -0.006095886, -0.0049362183, -0.056121826, -0.028198242, 0.02331543, 0.035827637, -0.026382446, 0.028015137, 0.016952515, -0.0045814514, -0.047332764, 0.016204834, -0.019332886, 0...
http://arxiv.org/abs/2310.00836v2
Towards LogiGLUE: A Brief Survey and A Benchmark for Analyzing Logical Reasoning Capabilities of Language Models
Man Luo, Shrinidhi Kumbhar, Ming shen, Mihir Parmar, Neeraj Varshney, Pratyay Banerjee, Somak Aditya, Chitta Baral
cs.CL, cs.AI
Logical reasoning is fundamental for humans yet presents a substantial challenge in the domain of Artificial Intelligence. Initially, researchers used Knowledge Representation and Reasoning (KR) systems that did not scale and required non trivial manual effort. Recently, the emergence of large language models (LLMs) ha...
http://arxiv.org/pdf/2310.00836v2
2023-11-03T23:43:07
2023-10-02T01:00:50
{ "title": [ -0.025512695, -0.019332886, -0.041656494, -0.08703613, 0.003129959, 0.026382446, -0.091430664, 0.006576538, -0.0016202927, 0.037872314, -0.0129852295, 0.007045746, 0.003042221, -0.011772156, -0.051696777, 0.016555786, -0.02243042, 0....
http://arxiv.org/abs/2310.00835v2
TRAM: Benchmarking Temporal Reasoning for Large Language Models
Yuqing Wang, Yun Zhao
cs.CL
Reasoning about time is essential for understanding the nuances of events described in natural language. Previous research on this topic has been limited in scope, characterized by a lack of standardized benchmarks that would allow for consistent evaluations across different studies. In this paper, we introduce TRAM, a...
http://arxiv.org/pdf/2310.00835v2
2023-10-03T13:54:02
2023-10-02T00:59:07
{ "title": [ -0.017929077, -0.03012085, -0.043273926, -0.12261963, 0.00623703, 0.0062942505, -0.07476807, 0.04257202, 0.00881958, 0.031158447, -0.014175415, 0.005996704, 0.0029792786, -0.016296387, -0.053131104, 0.004459381, -0.010772705, 0.02539...
http://arxiv.org/abs/2310.00833v1
Necessary and Sufficient Watermark for Large Language Models
Yuki Takezawa, Ryoma Sato, Han Bao, Kenta Niwa, Makoto Yamada
cs.CL, cs.LG
In recent years, large language models (LLMs) have achieved remarkable performances in various NLP tasks. They can generate texts that are indistinguishable from those written by humans. Such remarkable performance of LLMs increases their risk of being used for malicious purposes, such as generating fake news articles....
http://arxiv.org/pdf/2310.00833v1
2023-10-02T00:48:51
2023-10-02T00:48:51
{ "title": [ -0.022506714, 0.00018930435, -0.045013428, -0.07458496, 0.018173218, 0.003479004, -0.060760498, -0.00094652176, -0.004886627, 0.033996582, -0.010375977, 0.0043754578, 0.0037994385, -0.01826477, -0.03414917, 0.0143585205, -0.01574707, ...
http://arxiv.org/abs/2310.00832v1
Natural Language Models for Data Visualization Utilizing nvBench Dataset
Shuo Wang, Carlos Crespo-Quinones
cs.CL, cs.AI
Translation of natural language into syntactically correct commands for data visualization is an important application of natural language models and could be leveraged to many different tasks. A closely related effort is the task of translating natural languages into SQL queries, which in turn could be translated into...
http://arxiv.org/pdf/2310.00832v1
2023-10-02T00:48:01
2023-10-02T00:48:01
{ "title": [ -0.002023697, -0.0059051514, -0.04473877, -0.06390381, -0.0071640015, -0.010696411, -0.042785645, -0.0067253113, 0.009063721, 0.053649902, -0.04208374, -0.03050232, 0.02746582, -0.008766174, -0.05618286, -0.015716553, 0.0020313263, 0...
http://arxiv.org/abs/2310.00819v1
Parameter-Efficient Tuning Helps Language Model Alignment
Tianci Xue, Ziqi Wang, Heng Ji
cs.CL
Aligning large language models (LLMs) with human preferences is essential for safe and useful LLMs. Previous works mainly adopt reinforcement learning (RLHF) and direct preference optimization (DPO) with human feedback for alignment. Nevertheless, they have certain drawbacks. One such limitation is that they can only a...
http://arxiv.org/pdf/2310.00819v1
2023-10-01T23:27:14
2023-10-01T23:27:14
{ "title": [ -0.0016298294, -0.00022852421, -0.032226562, -0.082092285, -0.04333496, -0.010017395, -0.095336914, 0.013168335, -0.029388428, 0.03869629, 0.0056610107, -0.004825592, 0.03475952, -0.050689697, 0.0031585693, 0.015945435, 0.006412506, ...
http://arxiv.org/abs/2310.00811v1
Sparse Backpropagation for MoE Training
Liyuan Liu, Jianfeng Gao, Weizhu Chen
cs.LG, cs.AI, cs.CL, cs.CV
One defining characteristic of Mixture-of-Expert (MoE) models is their capacity for conducting sparse computation via expert routing, leading to remarkable scalability. However, backpropagation, the cornerstone of deep learning, requires dense computation, thereby posting challenges in MoE gradient computations. Here, ...
http://arxiv.org/pdf/2310.00811v1
2023-10-01T22:43:57
2023-10-01T22:43:57
{ "title": [ -0.021881104, -0.00831604, -0.040222168, -0.059143066, 0.015716553, 0.0003604889, -0.024612427, 0.009765625, 0.0129470825, 0.024337769, -0.0015153885, 0.0038280487, -0.0056648254, -0.040863037, -0.031173706, 0.0047836304, 0.009902954, ...
http://arxiv.org/abs/2310.00796v1
Injecting a Structural Inductive Bias into a Seq2Seq Model by Simulation
Matthias Lindemann, Alexander Koller, Ivan Titov
cs.CL
Strong inductive biases enable learning from little data and help generalization outside of the training distribution. Popular neural architectures such as Transformers lack strong structural inductive biases for seq2seq NLP tasks on their own. Consequently, they struggle with systematic generalization beyond the train...
http://arxiv.org/pdf/2310.00796v1
2023-10-01T21:19:12
2023-10-01T21:19:12
{ "title": [ 0.0066184998, -0.007701874, -0.014701843, -0.03338623, -0.023269653, -0.020874023, -0.018051147, -0.037322998, -0.031341553, 0.0010538101, -0.018997192, 0.0031471252, -0.008628845, -0.04788208, 0.014579773, -0.027114868, -0.013023376, ...
http://arxiv.org/abs/2310.00789v1
Testing the Limits of Unified Sequence to Sequence LLM Pretraining on Diverse Table Data Tasks
Soumajyoti Sarkar, Leonard Lausen
cs.CL, cs.LG
Tables stored in databases and tables which are present in web pages and articles account for a large part of semi-structured data that is available on the internet. It then becomes pertinent to develop a modeling approach with large language models (LLMs) that can be used to solve diverse table tasks such as semantic ...
http://arxiv.org/pdf/2310.00789v1
2023-10-01T21:06:15
2023-10-01T21:06:15
{ "title": [ -0.009468079, 0.008255005, -0.07397461, -0.082092285, -0.0053253174, -0.023773193, -0.081726074, 0.009979248, -0.0044441223, 0.047851562, -0.025970459, 0.045135498, 0.001244545, -0.012008667, -0.0071983337, 0.008361816, 0.010635376, ...
http://arxiv.org/abs/2310.00785v2
BooookScore: A systematic exploration of book-length summarization in the era of LLMs
Yapei Chang, Kyle Lo, Tanya Goyal, Mohit Iyyer
cs.CL, cs.AI, cs.LG
Summarizing book-length documents (>100K tokens) that exceed the context window size of large language models (LLMs) requires first breaking the input document into smaller chunks and then prompting an LLM to merge, update, and compress chunk-level summaries. Despite the complexity and importance of this task, it has y...
http://arxiv.org/pdf/2310.00785v2
2023-10-05T19:34:26
2023-10-01T20:46:44
{ "title": [ -0.032592773, -0.012687683, -0.035339355, -0.04031372, -0.031707764, -0.010475159, -0.04876709, -0.024093628, -0.01675415, 0.030349731, -0.019363403, 0.05267334, -0.01448822, -0.012718201, -0.023208618, -0.0007047653, -0.006958008, 0...
http://arxiv.org/abs/2310.00754v1
Analyzing and Mitigating Object Hallucination in Large Vision-Language Models
Yiyang Zhou, Chenhang Cui, Jaehong Yoon, Linjun Zhang, Zhun Deng, Chelsea Finn, Mohit Bansal, Huaxiu Yao
cs.LG, cs.CL, cs.CV
Large vision-language models (LVLMs) have shown remarkable abilities in understanding visual information with human languages. However, LVLMs still suffer from object hallucination, which is the problem of generating descriptions that include objects that do not actually exist in the images. This can negatively impact ...
http://arxiv.org/pdf/2310.00754v1
2023-10-01T18:10:53
2023-10-01T18:10:53
{ "title": [ -0.0007185936, 0.0065078735, -0.026947021, -0.050231934, 0.0023345947, -0.017929077, -0.035064697, -0.02267456, 0.010047913, 0.030136108, -0.0074920654, 0.0010261536, 0.0050811768, -0.010650635, -0.049957275, -0.0007967949, 0.012878418, ...
http://arxiv.org/abs/2310.00752v1
TIGERScore: Towards Building Explainable Metric for All Text Generation Tasks
Dongfu Jiang, Yishan Li, Ge Zhang, Wenhao Huang, Bill Yuchen Lin, Wenhu Chen
cs.CL, cs.AI
We present TIGERScore, a \textbf{T}rained metric that follows \textbf{I}nstruction \textbf{G}uidance to perform \textbf{E}xplainable, and \textbf{R}eference-free evaluation over a wide spectrum of text generation tasks. Different from other automatic evaluation methods that only provide arcane scores, TIGERScore is gui...
http://arxiv.org/pdf/2310.00752v1
2023-10-01T18:01:51
2023-10-01T18:01:51
{ "title": [ -0.057281494, -0.031951904, -0.023788452, -0.037353516, -0.038757324, 0.018951416, -0.06518555, 0.0030097961, 0.011726379, 0.027694702, -0.00522995, 0.024291992, -0.001868248, -0.008590698, -0.038208008, 0.013168335, -0.022872925, 0....
http://arxiv.org/abs/2310.00746v1
RoleLLM: Benchmarking, Eliciting, and Enhancing Role-Playing Abilities of Large Language Models
Zekun Moore Wang, Zhongyuan Peng, Haoran Que, Jiaheng Liu, Wangchunshu Zhou, Yuhan Wu, Hongcheng Guo, Ruitong Gan, Zehao Ni, Man Zhang, Zhaoxiang Zhang, Wanli Ouyang, Ke Xu, Wenhu Chen, Jie Fu, Junran Peng
cs.CL, cs.AI
The advent of Large Language Models (LLMs) has paved the way for complex tasks such as role-playing, which enhances user interactions by enabling models to imitate various characters. However, the closed-source nature of state-of-the-art LLMs and their general-purpose training limit role-playing optimization. In this p...
http://arxiv.org/pdf/2310.00746v1
2023-10-01T17:52:59
2023-10-01T17:52:59
{ "title": [ -0.026672363, -0.024139404, -0.02809143, -0.06994629, -0.010574341, 0.013473511, -0.06719971, 0.006034851, -0.025421143, 0.026794434, 0.008659363, 0.012382507, -0.0070266724, -0.029129028, -0.020965576, 0.0038433075, -0.022628784, 0....
http://arxiv.org/abs/2310.00741v1
FELM: Benchmarking Factuality Evaluation of Large Language Models
Shiqi Chen, Yiran Zhao, Jinghan Zhang, I-Chun Chern, Siyang Gao, Pengfei Liu, Junxian He
cs.CL
Assessing factuality of text generated by large language models (LLMs) is an emerging yet crucial research area, aimed at alerting users to potential errors and guiding the development of more reliable LLMs. Nonetheless, the evaluators assessing factuality necessitate suitable evaluation themselves to gauge progress an...
http://arxiv.org/pdf/2310.00741v1
2023-10-01T17:37:31
2023-10-01T17:37:31
{ "title": [ -0.033416748, -0.004547119, -0.04751587, -0.070373535, 0.0013980865, -0.0074920654, -0.06756592, -0.0063819885, 0.0074882507, 0.037109375, 0.0005469322, 0.018539429, 0.025939941, -0.015960693, -0.041656494, 0.0317688, -0.034088135, 0...
http://arxiv.org/abs/2310.00737v2
GenAI Against Humanity: Nefarious Applications of Generative Artificial Intelligence and Large Language Models
Emilio Ferrara
cs.CY, cs.AI, cs.CL, cs.HC
Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) are marvels of technology; celebrated for their prowess in natural language processing and multimodal content generation, they promise a transformative future. But as with all powerful tools, they come with their shadows. Picture living in a wo...
http://arxiv.org/pdf/2310.00737v2
2023-10-12T17:39:04
2023-10-01T17:25:56
{ "title": [ -0.027252197, -0.024124146, -0.044555664, -0.015090942, -0.016662598, -0.034118652, -0.0019340515, -0.015533447, -0.005264282, 0.025543213, -0.010307312, -0.00039601326, 0.017181396, -0.045288086, -0.016174316, 0.0036449432, -0.017944336...
http://arxiv.org/abs/2310.00734v1
Robust Sentiment Analysis for Low Resource languages Using Data Augmentation Approaches: A Case Study in Marathi
Aabha Pingle, Aditya Vyawahare, Isha Joshi, Rahul Tangsali, Geetanjali Kale, Raviraj Joshi
cs.CL, cs.LG
Sentiment analysis plays a crucial role in understanding the sentiment expressed in text data. While sentiment analysis research has been extensively conducted in English and other Western languages, there exists a significant gap in research efforts for sentiment analysis in low-resource languages. Limited resources, ...
http://arxiv.org/pdf/2310.00734v1
2023-10-01T17:09:31
2023-10-01T17:09:31
{ "title": [ 0.01146698, -0.004421234, -0.034179688, -0.061431885, 0.0071487427, -0.000035762787, -0.048797607, -0.0013256073, 0.010551453, 0.034332275, -0.039886475, -0.00030899048, 0.0154953, -0.009628296, -0.066589355, -0.011482239, -0.0059432983,...
http://arxiv.org/abs/2310.00706v1
Evaluating Speech Synthesis by Training Recognizers on Synthetic Speech
Dareen Alharthi, Roshan Sharma, Hira Dhamyal, Soumi Maiti, Bhiksha Raj, Rita Singh
cs.CL, cs.SD, eess.AS
Modern speech synthesis systems have improved significantly, with synthetic speech being indistinguishable from real speech. However, efficient and holistic evaluation of synthetic speech still remains a significant challenge. Human evaluation using Mean Opinion Score (MOS) is ideal, but inefficient due to high costs. ...
http://arxiv.org/pdf/2310.00706v1
2023-10-01T15:52:48
2023-10-01T15:52:48
{ "title": [ -0.032440186, 0.01776123, -0.04928589, -0.06274414, 0.006652832, 0.004753113, -0.05847168, 0.023376465, -0.030975342, 0.024887085, -0.035949707, 0.010368347, -0.023345947, -0.018371582, -0.03378296, -0.023773193, -0.026748657, 0.0559...
http://arxiv.org/abs/2310.00696v1
Do the Benefits of Joint Models for Relation Extraction Extend to Document-level Tasks?
Pratik Saini, Tapas Nayak, Indrajit Bhattacharya
cs.CL
Two distinct approaches have been proposed for relational triple extraction - pipeline and joint. Joint models, which capture interactions across triples, are the more recent development, and have been shown to outperform pipeline models for sentence-level extraction tasks. Document-level extraction is a more challengi...
http://arxiv.org/pdf/2310.00696v1
2023-10-01T15:09:36
2023-10-01T15:09:36
{ "title": [ 0.030349731, 0.0046081543, -0.03564453, -0.06878662, -0.020751953, -0.010246277, -0.056365967, -0.005584717, 0.008781433, 0.027404785, -0.0050849915, -0.0044822693, 0.019805908, -0.026992798, -0.002998352, 0.005142212, -0.01890564, 0...
http://arxiv.org/abs/2310.01448v2
Meta Semantic Template for Evaluation of Large Language Models
Yachuan Liu, Liang Chen, Jindong Wang, Qiaozhu Mei, Xing Xie
cs.CL, cs.AI
Do large language models (LLMs) genuinely understand the semantics of the language, or just memorize the training data? The recent concern on potential data contamination of LLMs has raised awareness of the community to conduct research on LLMs evaluation. In this paper, we propose MSTemp, an approach that creates meta...
http://arxiv.org/pdf/2310.01448v2
2023-10-19T02:46:07
2023-10-01T15:06:51
{ "title": [ -0.044769287, -0.012664795, -0.044281006, -0.0703125, 0.006248474, 0.015426636, -0.08099365, -0.0069732666, 0.017593384, 0.047821045, -0.010467529, -0.0059051514, 0.018692017, -0.017303467, -0.057800293, 0.020309448, -0.01638794, 0.0...
http://arxiv.org/abs/2310.00679v1
CebuaNER: A New Baseline Cebuano Named Entity Recognition Model
Ma. Beatrice Emanuela Pilar, Ellyza Mari Papas, Mary Loise Buenaventura, Dane Dedoroy, Myron Darrel Montefalcon, Jay Rhald Padilla, Lany Maceda, Mideth Abisado, Joseph Marvin Imperial
cs.CL
Despite being one of the most linguistically diverse groups of countries, computational linguistics and language processing research in Southeast Asia has struggled to match the level of countries from the Global North. Thus, initiatives such as open-sourcing corpora and the development of baseline models for basic lan...
http://arxiv.org/pdf/2310.00679v1
2023-10-01T14:09:42
2023-10-01T14:09:42
{ "title": [ -0.006462097, -0.009902954, -0.022628784, -0.06555176, -0.004245758, 0.008575439, -0.05291748, -0.0009765625, 0.015213013, 0.04144287, -0.03894043, 0.012664795, -0.0005955696, -0.008377075, -0.06451416, -0.011390686, -0.0066566467, 0...
http://arxiv.org/abs/2310.00672v2
GeRA: Label-Efficient Geometrically Regularized Alignment
Dustin Klebe, Tal Shnitzer, Mikhail Yurochkin, Leonid Karlinsky, Justin Solomon
cs.LG, cs.CL, cs.CV, I.2; I.2.7
Pretrained unimodal encoders incorporate rich semantic information into embedding space structures. To be similarly informative, multi-modal encoders typically require massive amounts of paired data for alignment and training. We introduce a semi-supervised Geometrically Regularized Alignment (GeRA) method to align the...
http://arxiv.org/pdf/2310.00672v2
2023-10-07T14:58:15
2023-10-01T13:48:36
{ "title": [ -0.051208496, -0.0028705597, -0.060455322, -0.066345215, 0.0069732666, 0.0017499924, -0.055847168, -0.0074157715, 0.035339355, 0.030975342, -0.011276245, 0.014846802, 0.01864624, -0.021438599, -0.02557373, 0.025863647, 0.012954712, 0...
http://arxiv.org/abs/2310.01446v1
Adaptive-Solver Framework for Dynamic Strategy Selection in Large Language Model Reasoning
Jianpeng Zhou, Wanjun Zhong, Yanlin Wang, Jiahai Wang
cs.CL, cs.AI
Large Language Models (LLMs) are showcasing impressive ability in handling complex reasoning tasks. In real-world situations, problems often span a spectrum of complexities. Humans inherently adjust their problem-solving approaches based on task complexity. However, most methodologies that leverage LLMs tend to adopt a...
http://arxiv.org/pdf/2310.01446v1
2023-10-01T12:28:36
2023-10-01T12:28:36
{ "title": [ -0.043426514, -0.013336182, -0.045410156, -0.06304932, -0.0042915344, 0.020828247, -0.060424805, 0.009635925, -0.017532349, 0.012466431, 0.00086069107, 0.011909485, -0.009880066, -0.02519226, 0.0036621094, 0.007663727, 0.0010967255, ...
http://arxiv.org/abs/2310.00648v2
Fewer is More: Trojan Attacks on Parameter-Efficient Fine-Tuning
Lauren Hong, Ting Wang
cs.CL
Parameter-efficient fine-tuning (PEFT) enables efficient adaptation of pre-trained language models (PLMs) to specific tasks. By tuning only a minimal set of (extra) parameters, PEFT achieves performance comparable to full fine-tuning. However, despite its prevalent use, the security implications of PEFT remain largely ...
http://arxiv.org/pdf/2310.00648v2
2023-10-04T13:21:44
2023-10-01T12:07:44
{ "title": [ -0.006690979, 0.0024261475, -0.012626648, -0.075805664, 0.026290894, 0.0052833557, -0.001247406, 0.0020065308, -0.0692749, 0.039886475, 0.012130737, 0.035705566, -0.03250122, -0.0051002502, -0.026794434, -0.0075187683, -0.001288414, ...
http://arxiv.org/abs/2310.00637v1
Knowledge Engineering using Large Language Models
Bradley P. Allen, Lise Stork, Paul Groth
cs.AI, cs.CL
Knowledge engineering is a discipline that focuses on the creation and maintenance of processes that generate and apply knowledge. Traditionally, knowledge engineering approaches have focused on knowledge expressed in formal languages. The emergence of large language models and their capabilities to effectively work wi...
http://arxiv.org/pdf/2310.00637v1
2023-10-01T10:26:25
2023-10-01T10:26:25
{ "title": [ -0.04916382, -0.035369873, -0.0569458, -0.06854248, -0.0014305115, 0.009590149, -0.062164307, 0.023025513, 0.012138367, 0.031921387, -0.0076026917, -0.02659607, 0.0035686493, -0.020187378, -0.053741455, 0.004032135, -0.0037498474, 0....
http://arxiv.org/abs/2310.09297v1
Understanding AI Cognition: A Neural Module for Inference Inspired by Human Memory Mechanisms
Xiangyu Zeng, Jie Lin, Piao Hu, Ruizheng Huang, Zhicheng Zhang
cs.LG, cs.AI, cs.CL
How humans and machines make sense of current inputs for relation reasoning and question-answering while putting the perceived information into context of our past memories, has been a challenging conundrum in cognitive science and artificial intelligence. Inspired by human brain's memory system and cognitive architect...
http://arxiv.org/pdf/2310.09297v1
2023-10-01T08:12:55
2023-10-01T08:12:55
{ "title": [ -0.038024902, -0.017364502, -0.037750244, -0.028823853, -0.011833191, -0.019515991, -0.0012788773, -0.008140564, 0.004787445, 0.01965332, -0.019226074, -0.013786316, 0.009216309, -0.02609253, -0.039276123, -0.0021381378, 0.003019333, ...