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/2309.16573v1
The ARRT of Language-Models-as-a-Service: Overview of a New Paradigm and its Challenges
Emanuele La Malfa, Aleksandar Petrov, Simon Frieder, Christoph Weinhuber, Ryan Burnell, Anthony G. Cohn, Nigel Shadbolt, Michael Wooldridge
cs.AI, cs.CL, cs.CY
Some of the most powerful language models currently are proprietary systems, accessible only via (typically restrictive) web or software programming interfaces. This is the Language-Models-as-a-Service (LMaaS) paradigm. Contrasting with scenarios where full model access is available, as in the case of open-source model...
http://arxiv.org/pdf/2309.16573v1
2023-09-28T16:29:52
2023-09-28T16:29:52
{ "title": [ -0.020339966, -0.025863647, -0.035736084, -0.092163086, 0.0045280457, 0.004386902, -0.04574585, 0.028579712, -0.012573242, 0.049194336, -0.008857727, 0.017074585, 0.022079468, -0.01626587, -0.040924072, 0.026245117, -0.01737976, 0.05...
http://arxiv.org/abs/2309.16540v1
Unsupervised Fact Verification by Language Model Distillation
Adrián Bazaga, Pietro Liò, Gos Micklem
cs.CL, cs.LG, stat.ML
Unsupervised fact verification aims to verify a claim using evidence from a trustworthy knowledge base without any kind of data annotation. To address this challenge, algorithms must produce features for every claim that are both semantically meaningful, and compact enough to find a semantic alignment with the source i...
http://arxiv.org/pdf/2309.16540v1
2023-09-28T15:53:44
2023-09-28T15:53:44
{ "title": [ -0.0072631836, 0.010543823, -0.046051025, -0.09423828, 0.015853882, 0.0069389343, -0.08337402, -0.004673004, 0.016845703, 0.026657104, -0.020935059, -0.016174316, 0.008132935, -0.018554688, -0.04837036, -0.003900528, -0.033935547, 0....
http://arxiv.org/abs/2309.16535v1
KLoB: a Benchmark for Assessing Knowledge Locating Methods in Language Models
Yiming Ju, Zheng Zhang
cs.CL, cs.AI
Recently, Locate-Then-Edit paradigm has emerged as one of the main approaches in changing factual knowledge stored in the Language models. However, there is a lack of research on whether present locating methods can pinpoint the exact parameters embedding the desired knowledge. Moreover, although many researchers have ...
http://arxiv.org/pdf/2309.16535v1
2023-09-28T15:47:03
2023-09-28T15:47:03
{ "title": [ -0.031707764, -0.007911682, -0.029388428, -0.036895752, -0.0014562607, -0.0048980713, -0.042785645, -0.0262146, 0.012046814, 0.043029785, -0.021194458, 0.016998291, 0.029525757, -0.010604858, -0.05770874, 0.022537231, -0.018981934, 0...
http://arxiv.org/abs/2310.01429v1
Chatmap : Large Language Model Interaction with Cartographic Data
Eren Unlu
cs.CL, cs.AI
The swift advancement and widespread availability of foundational Large Language Models (LLMs), complemented by robust fine-tuning methodologies, have catalyzed their adaptation for innovative and industrious applications. Enabling LLMs to recognize and interpret geospatial data, while offering a linguistic access to v...
http://arxiv.org/pdf/2310.01429v1
2023-09-28T15:32:36
2023-09-28T15:32:36
{ "title": [ -0.029556274, -0.027496338, -0.016311646, -0.0345459, -0.020751953, 0.004005432, -0.040527344, -0.012817383, 0.03390503, 0.025100708, -0.026992798, -0.019073486, 0.00605011, -0.015686035, -0.05633545, -0.0021438599, -0.0012979507, 0....
http://arxiv.org/abs/2309.16511v1
Toloka Visual Question Answering Benchmark
Dmitry Ustalov, Nikita Pavlichenko, Sergey Koshelev, Daniil Likhobaba, Alisa Smirnova
cs.CV, cs.AI, cs.CL, cs.HC, 68-11, C.4
In this paper, we present Toloka Visual Question Answering, a new crowdsourced dataset allowing comparing performance of machine learning systems against human level of expertise in the grounding visual question answering task. In this task, given an image and a textual question, one has to draw the bounding box around...
http://arxiv.org/pdf/2309.16511v1
2023-09-28T15:18:35
2023-09-28T15:18:35
{ "title": [ -0.022872925, -0.01133728, -0.03579712, -0.041168213, -0.03149414, 0.0013389587, -0.043182373, -0.0009264946, 0.022094727, 0.03363037, -0.030700684, 0.018005371, -0.0033798218, -0.01499939, -0.04071045, 0.012825012, 0.013946533, 0.01...
http://arxiv.org/abs/2309.16459v1
Augmenting LLMs with Knowledge: A survey on hallucination prevention
Konstantinos Andriopoulos, Johan Pouwelse
cs.CL, cs.AI, cs.LG
Large pre-trained language models have demonstrated their proficiency in storing factual knowledge within their parameters and achieving remarkable results when fine-tuned for downstream natural language processing tasks. Nonetheless, their capacity to access and manipulate knowledge with precision remains constrained,...
http://arxiv.org/pdf/2309.16459v1
2023-09-28T14:09:58
2023-09-28T14:09:58
{ "title": [ -0.007820129, 0.00042653084, -0.043640137, -0.026931763, -0.013366699, -0.020477295, -0.030685425, -0.021408081, -0.024093628, 0.018295288, -0.014884949, 0.042663574, -0.020584106, -0.03604126, -0.0055885315, 0.00026202202, 0.0037593842,...
http://arxiv.org/abs/2309.16424v1
Prompt-and-Align: Prompt-Based Social Alignment for Few-Shot Fake News Detection
Jiaying Wu, Shen Li, Ailin Deng, Miao Xiong, Bryan Hooi
cs.CL, cs.AI, cs.SI
Despite considerable advances in automated fake news detection, due to the timely nature of news, it remains a critical open question how to effectively predict the veracity of news articles based on limited fact-checks. Existing approaches typically follow a "Train-from-Scratch" paradigm, which is fundamentally bounde...
http://arxiv.org/pdf/2309.16424v1
2023-09-28T13:19:43
2023-09-28T13:19:43
{ "title": [ 0.0003695488, 0.0055999756, -0.034606934, -0.0141067505, 0.0030231476, -0.011932373, 0.0011920929, -0.03665161, -0.027923584, 0.042297363, -0.02835083, 0.032592773, 0.0042648315, -0.007297516, -0.016677856, -0.02218628, -0.028915405, ...
http://arxiv.org/abs/2309.16396v3
A Comprehensive Survey of Document-level Relation Extraction (2016-2023)
Julien Delaunay, Hanh Thi Hong Tran, Carlos-Emiliano González-Gallardo, Georgeta Bordea, Nicolas Sidere, Antoine Doucet
cs.CL, A.1
Document-level relation extraction (DocRE) is an active area of research in natural language processing (NLP) concerned with identifying and extracting relationships between entities beyond sentence boundaries. Compared to the more traditional sentence-level relation extraction, DocRE provides a broader context for ana...
http://arxiv.org/pdf/2309.16396v3
2023-10-12T13:39:39
2023-09-28T12:43:32
{ "title": [ -0.0063095093, -0.010406494, -0.0463562, -0.08911133, 0.014251709, 0.0019483566, -0.05621338, 0.009986877, 0.024536133, 0.01713562, -0.013435364, 0.007637024, 0.009986877, -0.012481689, -0.030715942, 0.01184845, -0.012680054, 0.03723...
http://arxiv.org/abs/2309.16354v1
Transformer-VQ: Linear-Time Transformers via Vector Quantization
Lucas D. Lingle
cs.LG, cs.CL, cs.CV
We introduce Transformer-VQ, a decoder-only transformer computing softmax-based dense self-attention in linear time. Transformer-VQ's efficient attention is enabled by vector-quantized keys and a novel caching mechanism. In large-scale experiments, Transformer-VQ is shown highly competitive in quality, with strong resu...
http://arxiv.org/pdf/2309.16354v1
2023-09-28T11:26:52
2023-09-28T11:26:52
{ "title": [ -0.015914917, 0.008071899, -0.041168213, -0.042907715, 0.013656616, -0.027160645, -0.0134887695, -0.0423584, -0.023529053, 0.039733887, -0.0099487305, 0.0039787292, 0.016998291, -0.028015137, -0.02243042, 0.002658844, 0.019989014, 0....
http://arxiv.org/abs/2309.16349v1
Human Feedback is not Gold Standard
Tom Hosking, Phil Blunsom, Max Bartolo
cs.CL
Human feedback has become the de facto standard for evaluating the performance of Large Language Models, and is increasingly being used as a training objective. However, it is not clear which properties of a generated output this single `preference' score captures. We hypothesise that preference scores are subjective a...
http://arxiv.org/pdf/2309.16349v1
2023-09-28T11:18:20
2023-09-28T11:18:20
{ "title": [ 0.008834839, 0.016601562, 0.03552246, 0.024536133, -0.039398193, 0.001373291, 0.0013780594, -0.05029297, -0.006668091, 0.04498291, -0.02796936, -0.034423828, -0.05404663, 0.0014772415, -0.047027588, -0.04647827, 0.037872314, 0.015716...
http://arxiv.org/abs/2309.16347v1
Intrinsic Language-Guided Exploration for Complex Long-Horizon Robotic Manipulation Tasks
Eleftherios Triantafyllidis, Filippos Christianos, Zhibin Li
cs.RO, cs.CL, cs.LG
Current reinforcement learning algorithms struggle in sparse and complex environments, most notably in long-horizon manipulation tasks entailing a plethora of different sequences. In this work, we propose the Intrinsically Guided Exploration from Large Language Models (IGE-LLMs) framework. By leveraging LLMs as an assi...
http://arxiv.org/pdf/2309.16347v1
2023-09-28T11:14:52
2023-09-28T11:14:52
{ "title": [ -0.024353027, -0.008514404, -0.016540527, -0.02583313, -0.009597778, -0.011520386, -0.0065994263, 0.0015516281, -0.009414673, 0.037872314, -0.023162842, 0.012207031, -0.019546509, -0.008674622, -0.038024902, -0.013916016, 0.01499939, ...
http://arxiv.org/abs/2309.16319v1
Augmenting transformers with recursively composed multi-grained representations
Xiang Hu, Qingyang Zhu, Kewei Tu, Wei Wu
cs.CL, cs.AI
We present ReCAT, a recursive composition augmented Transformer that is able to explicitly model hierarchical syntactic structures of raw texts without relying on gold trees during both learning and inference. Existing research along this line restricts data to follow a hierarchical tree structure and thus lacks inter-...
http://arxiv.org/pdf/2309.16319v1
2023-09-28T10:24:39
2023-09-28T10:24:39
{ "title": [ 0.013999939, 0.017059326, -0.020141602, -0.09313965, 0.027694702, 0.025527954, -0.051452637, -0.0012722015, 0.0053863525, 0.038970947, -0.038330078, -0.0015945435, 0.0065345764, -0.010055542, -0.03656006, -0.0010948181, 0.004722595, ...
http://arxiv.org/abs/2309.16298v2
At Which Training Stage Does Code Data Help LLMs Reasoning?
Yingwei Ma, Yue Liu, Yue Yu, Yuanliang Zhang, Yu Jiang, Changjian Wang, Shanshan Li
cs.CL
Large Language Models (LLMs) have exhibited remarkable reasoning capabilities and become the foundation of language technologies. Inspired by the great success of code data in training LLMs, we naturally wonder at which training stage introducing code data can really help LLMs reasoning. To this end, this paper systema...
http://arxiv.org/pdf/2309.16298v2
2023-09-30T07:01:13
2023-09-28T09:50:27
{ "title": [ -0.028564453, -0.03982544, -0.054260254, -0.02848816, -0.056152344, -0.013824463, -0.029281616, 0.030670166, -0.019638062, -0.007797241, -0.040252686, 0.0025177002, -0.020614624, -0.047912598, 0.05178833, -0.036621094, 0.0095825195, ...
http://arxiv.org/abs/2309.16292v2
DiLu: A Knowledge-Driven Approach to Autonomous Driving with Large Language Models
Licheng Wen, Daocheng Fu, Xin Li, Xinyu Cai, Tao Ma, Pinlong Cai, Min Dou, Botian Shi, Liang He, Yu Qiao
cs.RO, cs.CL
Recent advancements in autonomous driving have relied on data-driven approaches, which are widely adopted but face challenges including dataset bias, overfitting, and uninterpretability. Drawing inspiration from the knowledge-driven nature of human driving, we explore the question of how to instill similar capabilities...
http://arxiv.org/pdf/2309.16292v2
2023-10-12T11:11:47
2023-09-28T09:41:35
{ "title": [ -0.050231934, -0.0073890686, -0.066833496, -0.048431396, 0.01449585, -0.026245117, -0.020996094, 0.005279541, -0.015548706, 0.027145386, -0.057556152, 0.0078086853, -0.017105103, -0.006416321, -0.028015137, -0.028747559, -0.003616333, ...
http://arxiv.org/abs/2309.16289v1
LawBench: Benchmarking Legal Knowledge of Large Language Models
Zhiwei Fei, Xiaoyu Shen, Dawei Zhu, Fengzhe Zhou, Zhuo Han, Songyang Zhang, Kai Chen, Zongwen Shen, Jidong Ge
cs.CL, cs.AI, cs.LG
Large language models (LLMs) have demonstrated strong capabilities in various aspects. However, when applying them to the highly specialized, safe-critical legal domain, it is unclear how much legal knowledge they possess and whether they can reliably perform legal-related tasks. To address this gap, we propose a compr...
http://arxiv.org/pdf/2309.16289v1
2023-09-28T09:35:59
2023-09-28T09:35:59
{ "title": [ -0.050842285, -0.023254395, -0.05618286, -0.06964111, -0.0034427643, 0.0035648346, -0.061767578, 0.019424438, 0.01966858, 0.0234375, -0.038146973, 0.018035889, 0.007827759, -0.00308609, -0.0418396, 0.0024528503, -0.012512207, 0.03671...
http://arxiv.org/abs/2309.16283v1
Self-supervised Cross-view Representation Reconstruction for Change Captioning
Yunbin Tu, Liang Li, Li Su, Zheng-Jun Zha, Chenggang Yan, Qingming Huang
cs.CV, cs.CL
Change captioning aims to describe the difference between a pair of similar images. Its key challenge is how to learn a stable difference representation under pseudo changes caused by viewpoint change. In this paper, we address this by proposing a self-supervised cross-view representation reconstruction (SCORER) networ...
http://arxiv.org/pdf/2309.16283v1
2023-09-28T09:28:50
2023-09-28T09:28:50
{ "title": [ -0.0146484375, 0.010734558, -0.031402588, -0.052368164, -0.013542175, -0.011123657, -0.04107666, -0.010559082, 0.004470825, 0.05038452, -0.044036865, 0.025604248, 0.0030345917, 0.008552551, -0.045166016, -0.0036964417, 0.014167786, 0...
http://arxiv.org/abs/2309.16275v1
UPB @ ACTI: Detecting Conspiracies using fine tuned Sentence Transformers
Andrei Paraschiv, Mihai Dascalu
cs.CL, cs.AI, I.2.7
Conspiracy theories have become a prominent and concerning aspect of online discourse, posing challenges to information integrity and societal trust. As such, we address conspiracy theory detection as proposed by the ACTI @ EVALITA 2023 shared task. The combination of pre-trained sentence Transformer models and data au...
http://arxiv.org/pdf/2309.16275v1
2023-09-28T09:17:20
2023-09-28T09:17:20
{ "title": [ 0.01524353, -0.004776001, -0.015174866, -0.05859375, 0.015792847, 0.0013399124, -0.013175964, -0.013824463, -0.026504517, 0.01838684, -0.014305115, 0.0096206665, 0.021469116, -0.02116394, -0.023773193, -0.007858276, -0.032165527, 0.0...
http://arxiv.org/abs/2309.16270v1
Social Media Fashion Knowledge Extraction as Captioning
Yifei Yuan, Wenxuan Zhang, Yang Deng, Wai Lam
cs.CL
Social media plays a significant role in boosting the fashion industry, where a massive amount of fashion-related posts are generated every day. In order to obtain the rich fashion information from the posts, we study the task of social media fashion knowledge extraction. Fashion knowledge, which typically consists of ...
http://arxiv.org/pdf/2309.16270v1
2023-09-28T09:07:48
2023-09-28T09:07:48
{ "title": [ -0.016860962, -0.0019264221, -0.051696777, -0.03363037, -0.023803711, -0.028213501, -0.017120361, 0.019561768, -0.0011367798, 0.060150146, -0.03756714, 0.009803772, 0.027175903, -0.008560181, -0.033843994, -0.010055542, 0.01727295, 0...
http://arxiv.org/abs/2309.16254v1
On the Challenges of Fully Incremental Neural Dependency Parsing
Ana Ezquerro, Carlos Gómez-Rodríguez, David Vilares
cs.CL
Since the popularization of BiLSTMs and Transformer-based bidirectional encoders, state-of-the-art syntactic parsers have lacked incrementality, requiring access to the whole sentence and deviating from human language processing. This paper explores whether fully incremental dependency parsing with modern architectures...
http://arxiv.org/pdf/2309.16254v1
2023-09-28T08:44:08
2023-09-28T08:44:08
{ "title": [ -0.0062026978, -0.0033454895, -0.04763794, -0.11029053, 0.01335907, 0.010040283, -0.07104492, 0.035614014, -0.0011739731, 0.014823914, -0.03036499, 0.003917694, -0.031982422, -0.005138397, -0.03717041, -0.0054016113, -0.013320923, 0....
http://arxiv.org/abs/2309.16248v1
Spider4SPARQL: A Complex Benchmark for Evaluating Knowledge Graph Question Answering Systems
Catherine Kosten, Philippe Cudré-Mauroux, Kurt Stockinger
cs.CL
With the recent spike in the number and availability of Large Language Models (LLMs), it has become increasingly important to provide large and realistic benchmarks for evaluating Knowledge Graph Question Answering (KBQA) systems. So far the majority of benchmarks rely on pattern-based SPARQL query generation approache...
http://arxiv.org/pdf/2309.16248v1
2023-09-28T08:41:08
2023-09-28T08:41:08
{ "title": [ -0.033325195, -0.017333984, -0.051239014, -0.06378174, -0.0135269165, 0.006389618, -0.046661377, 0.020126343, 0.017181396, 0.03604126, -0.026229858, 0.0042304993, -0.008392334, -0.004917145, -0.06567383, -0.008033752, 0.00011754036, ...
http://arxiv.org/abs/2309.16235v1
Language models in molecular discovery
Nikita Janakarajan, Tim Erdmann, Sarath Swaminathan, Teodoro Laino, Jannis Born
physics.chem-ph, cs.AI, cs.CL, cs.LG, q-bio.BM
The success of language models, especially transformer-based architectures, has trickled into other domains giving rise to "scientific language models" that operate on small molecules, proteins or polymers. In chemistry, language models contribute to accelerating the molecule discovery cycle as evidenced by promising r...
http://arxiv.org/pdf/2309.16235v1
2023-09-28T08:19:54
2023-09-28T08:19:54
{ "title": [ -0.07366943, -0.022994995, -0.054229736, -0.05050659, -0.023498535, 0.026855469, -0.0657959, 0.040771484, -0.0020332336, 0.021759033, -0.016220093, 0.009284973, 0.0018720627, -0.018203735, -0.019378662, 0.015533447, -0.001405716, 0.0...
http://arxiv.org/abs/2309.16234v1
Analyzing Political Figures in Real-Time: Leveraging YouTube Metadata for Sentiment Analysis
Danendra Athallariq Harya Putra, Arief Purnama Muharram
cs.CL
Sentiment analysis using big data from YouTube videos metadata can be conducted to analyze public opinions on various political figures who represent political parties. This is possible because YouTube has become one of the platforms for people to express themselves, including their opinions on various political figure...
http://arxiv.org/pdf/2309.16234v1
2023-09-28T08:15:55
2023-09-28T08:15:55
{ "title": [ 0.012893677, -0.0015106201, -0.030166626, -0.084472656, 0.027816772, -0.011299133, -0.009391785, -0.0026760101, 0.015579224, 0.050933838, -0.032287598, 0.011306763, 0.0026988983, 0.0079422, -0.07434082, 0.00038695335, -0.008216858, 0...
http://arxiv.org/abs/2309.16231v1
Controllable Text Generation with Residual Memory Transformer
Hanqing Zhang, Sun Si, Haiming Wu, Dawei Song
cs.CL
Large-scale Causal Language Models (CLMs), e.g., GPT3 and ChatGPT, have brought great success in text generation. However, it is still an open challenge to control the generation process of CLM while balancing flexibility, control granularity, and generation efficiency. In this paper, we provide a new alternative for c...
http://arxiv.org/pdf/2309.16231v1
2023-09-28T08:13:33
2023-09-28T08:13:33
{ "title": [ -0.029510498, -0.011795044, -0.032165527, -0.047454834, -0.020492554, 0.014953613, -0.056030273, -0.0005450249, 0.0028076172, 0.0552063, 0.0038700104, 0.0013256073, 0.029220581, -0.016998291, -0.04260254, 0.027175903, -0.0031585693, ...
http://arxiv.org/abs/2309.16228v1
Brand Network Booster: A New System for Improving Brand Connectivity
J. Cancellieri, W. Didimo, A. Fronzetti Colladon, F. Montecchiani
cs.SI, cs.CL, cs.SE, physics.soc-ph, F.2; H.4; H.5; I.2; D.2
This paper presents a new decision support system offered for an in-depth analysis of semantic networks, which can provide insights for a better exploration of a brand's image and the improvement of its connectivity. In terms of network analysis, we show that this goal is achieved by solving an extended version of the ...
http://arxiv.org/pdf/2309.16228v1
2023-09-28T08:09:33
2023-09-28T08:09:33
{ "title": [ -0.0027675629, -0.0015506744, -0.035614014, -0.02670288, 0.004180908, -0.021575928, -0.004508972, -0.015098572, -0.044311523, 0.015716553, -0.02180481, 0.022537231, 0.009841919, -0.03640747, 0.005191803, -0.024917603, 0.011146545, 0....
http://arxiv.org/abs/2309.16202v1
Marathi-English Code-mixed Text Generation
Dhiraj Amin, Sharvari Govilkar, Sagar Kulkarni, Yash Shashikant Lalit, Arshi Ajaz Khwaja, Daries Xavier, Sahil Girijashankar Gupta
cs.CL
Code-mixing, the blending of linguistic elements from distinct languages to form meaningful sentences, is common in multilingual settings, yielding hybrid languages like Hinglish and Minglish. Marathi, India's third most spoken language, often integrates English for precision and formality. Developing code-mixed langua...
http://arxiv.org/pdf/2309.16202v1
2023-09-28T06:51:26
2023-09-28T06:51:26
{ "title": [ -0.047790527, -0.036224365, -0.043701172, -0.0066184998, -0.06524658, -0.010498047, -0.048858643, 0.040161133, -0.056671143, 0.070495605, -0.016067505, 0.021392822, 0.06384277, -0.024383545, 0.028182983, -0.0056419373, 0.014732361, 0...
http://arxiv.org/abs/2310.01427v1
Attention Sorting Combats Recency Bias In Long Context Language Models
Alexander Peysakhovich, Adam Lerer
cs.CL, cs.AI
Current language models often fail to incorporate long contexts efficiently during generation. We show that a major contributor to this issue are attention priors that are likely learned during pre-training: relevant information located earlier in context is attended to less on average. Yet even when models fail to use...
http://arxiv.org/pdf/2310.01427v1
2023-09-28T05:19:06
2023-09-28T05:19:06
{ "title": [ 0.0076522827, 0.00041532516, -0.03111267, -0.053466797, -0.0061531067, -0.023788452, -0.02520752, -0.004371643, 0.007019043, 0.05328369, -0.01777649, 0.0026397705, 0.0335083, -0.006542206, -0.0501709, 0.01020813, -0.013877869, 0.0245...
http://arxiv.org/abs/2309.16175v1
Using Weak Supervision and Data Augmentation in Question Answering
Chumki Basu, Himanshu Garg, Allen McIntosh, Sezai Sablak, John R. Wullert II
cs.CL, cs.LG
The onset of the COVID-19 pandemic accentuated the need for access to biomedical literature to answer timely and disease-specific questions. During the early days of the pandemic, one of the biggest challenges we faced was the lack of peer-reviewed biomedical articles on COVID-19 that could be used to train machine lea...
http://arxiv.org/pdf/2309.16175v1
2023-09-28T05:16:51
2023-09-28T05:16:51
{ "title": [ 0.020812988, -0.0019216537, -0.036376953, -0.07196045, 0.014625549, -0.011184692, -0.020370483, 0.0010519028, 0.009712219, 0.014572144, -0.04232788, 0.017547607, -0.0107421875, 0.00086069107, -0.040039062, -0.0036182404, -0.0013628006, ...
http://arxiv.org/abs/2309.16167v1
Large Language Model Soft Ideologization via AI-Self-Consciousness
Xiaotian Zhou, Qian Wang, Xiaofeng Wang, Haixu Tang, Xiaozhong Liu
cs.CL
Large language models (LLMs) have demonstrated human-level performance on a vast spectrum of natural language tasks. However, few studies have addressed the LLM threat and vulnerability from an ideology perspective, especially when they are increasingly being deployed in sensitive domains, e.g., elections and education...
http://arxiv.org/pdf/2309.16167v1
2023-09-28T04:47:58
2023-09-28T04:47:58
{ "title": [ -0.0079193115, 0.0093688965, -0.040649414, -0.040252686, -0.0043296814, -0.02255249, 0.0034236908, 0.008728027, -0.011238098, 0.037384033, -0.030380249, -0.022033691, 0.030822754, -0.041381836, -0.0037937164, -0.006427765, -0.003566742, ...
http://arxiv.org/abs/2309.16155v1
The Trickle-down Impact of Reward (In-)consistency on RLHF
Lingfeng Shen, Sihao Chen, Linfeng Song, Lifeng Jin, Baolin Peng, Haitao Mi, Daniel Khashabi, Dong Yu
cs.CL, cs.LG
Standard practice within Reinforcement Learning from Human Feedback (RLHF) involves optimizing against a Reward Model (RM), which itself is trained to reflect human preferences for desirable generations. A notable subject that is understudied is the (in-)consistency of RMs -- whether they can recognize the semantic cha...
http://arxiv.org/pdf/2309.16155v1
2023-09-28T04:05:13
2023-09-28T04:05:13
{ "title": [ 0.03100586, -0.00554657, -0.034301758, -0.05731201, -0.013893127, -0.025238037, -0.011436462, -0.0072250366, -0.06707764, 0.04611206, 0.022979736, -0.027908325, 0.00088071823, -0.055511475, -0.0040893555, -0.032348633, 0.0024547577, ...
http://arxiv.org/abs/2309.16150v1
AE-GPT: Using Large Language Models to Extract Adverse Events from Surveillance Reports-A Use Case with Influenza Vaccine Adverse Events
Yiming Li, Jianfu Li, Jianping He, Cui Tao
cs.CL, cs.AI
Though Vaccines are instrumental in global health, mitigating infectious diseases and pandemic outbreaks, they can occasionally lead to adverse events (AEs). Recently, Large Language Models (LLMs) have shown promise in effectively identifying and cataloging AEs within clinical reports. Utilizing data from the Vaccine A...
http://arxiv.org/pdf/2309.16150v1
2023-09-28T03:53:21
2023-09-28T03:53:21
{ "title": [ 0.015396118, -0.011787415, -0.036132812, -0.1005249, -0.007335663, 0.0019025803, -0.08294678, 0.0031471252, -0.0049209595, 0.017211914, -0.014015198, 0.019058228, -0.014083862, -0.033966064, -0.017028809, 0.008239746, -0.013885498, -...
http://arxiv.org/abs/2309.16145v1
The Confidence-Competence Gap in Large Language Models: A Cognitive Study
Aniket Kumar Singh, Suman Devkota, Bishal Lamichhane, Uttam Dhakal, Chandra Dhakal
cs.CL, cs.CY, cs.HC, ACM-class: I.2.0
Large Language Models (LLMs) have acquired ubiquitous attention for their performances across diverse domains. Our study here searches through LLMs' cognitive abilities and confidence dynamics. We dive deep into understanding the alignment between their self-assessed confidence and actual performance. We exploit these ...
http://arxiv.org/pdf/2309.16145v1
2023-09-28T03:50:09
2023-09-28T03:50:09
{ "title": [ -0.032348633, 0.010887146, -0.05429077, -0.048187256, 0.00422287, -0.013427734, -0.040802002, -0.015403748, 0.0065078735, 0.022247314, -0.036712646, -0.0099105835, -0.0014381409, -0.019744873, -0.0579834, -0.011054993, -0.025024414, ...
http://arxiv.org/abs/2309.16090v1
TPE: Towards Better Compositional Reasoning over Conceptual Tools with Multi-persona Collaboration
Hongru Wang, Huimin Wang, Lingzhi Wang, Minda Hu, Rui Wang, Boyang Xue, Hongyuan Lu, Fei Mi, Kam-Fai Wong
cs.AI, cs.CL
Large language models (LLMs) have demonstrated exceptional performance in planning the use of various functional tools, such as calculators and retrievers, particularly in question-answering tasks. In this paper, we expand the definition of these tools, centering on conceptual tools within the context of dialogue syste...
http://arxiv.org/pdf/2309.16090v1
2023-09-28T01:18:53
2023-09-28T01:18:53
{ "title": [ -0.02708435, -0.018753052, -0.06021118, -0.09350586, -0.00030732155, 0.011314392, -0.09020996, 0.024841309, 0.014579773, 0.024154663, 0.0000834465, 0.007293701, 0.0038394928, -0.03366089, -0.026184082, 0.005821228, -0.04385376, 0.038...
http://arxiv.org/abs/2309.16082v1
Forgetting Private Textual Sequences in Language Models via Leave-One-Out Ensemble
Zhe Liu, Ozlem Kalinli
cs.CL, cs.AI
Recent research has shown that language models have a tendency to memorize rare or unique token sequences in the training corpus. After deploying a model, practitioners might be asked to delete any personal information from the model by individuals' requests. Re-training the underlying model every time individuals woul...
http://arxiv.org/pdf/2309.16082v1
2023-09-28T00:43:18
2023-09-28T00:43:18
{ "title": [ -0.02784729, -0.025177002, -0.028915405, -0.05819702, -0.01878357, 0.014350891, -0.040985107, 0.039642334, 0.031311035, 0.051818848, -0.024551392, 0.0014858246, 0.009521484, 0.0036945343, -0.07110596, 0.011054993, -0.021453857, 0.045...
http://arxiv.org/abs/2309.16058v1
AnyMAL: An Efficient and Scalable Any-Modality Augmented Language Model
Seungwhan Moon, Andrea Madotto, Zhaojiang Lin, Tushar Nagarajan, Matt Smith, Shashank Jain, Chun-Fu Yeh, Prakash Murugesan, Peyman Heidari, Yue Liu, Kavya Srinet, Babak Damavandi, Anuj Kumar
cs.LG, cs.CL, cs.CV
We present Any-Modality Augmented Language Model (AnyMAL), a unified model that reasons over diverse input modality signals (i.e. text, image, video, audio, IMU motion sensor), and generates textual responses. AnyMAL inherits the powerful text-based reasoning abilities of the state-of-the-art LLMs including LLaMA-2 (70...
http://arxiv.org/pdf/2309.16058v1
2023-09-27T22:50:51
2023-09-27T22:50:51
{ "title": [ -0.025299072, -0.007949829, -0.045135498, -0.08935547, 0.010406494, 0.027023315, -0.082092285, 0.015342712, 0.01626587, 0.041412354, 0.0009012222, -0.0011386871, 0.033599854, -0.01889038, -0.03616333, 0.023529053, -0.022109985, 0.024...
http://arxiv.org/abs/2309.16042v1
Towards Best Practices of Activation Patching in Language Models: Metrics and Methods
Fred Zhang, Neel Nanda
cs.LG, cs.AI, cs.CL
Mechanistic interpretability seeks to understand the internal mechanisms of machine learning models, where localization -- identifying the important model components -- is a key step. Activation patching, also known as causal tracing or interchange intervention, is a standard technique for this task (Vig et al., 2020),...
http://arxiv.org/pdf/2309.16042v1
2023-09-27T21:53:56
2023-09-27T21:53:56
{ "title": [ -0.012870789, 0.0005168915, -0.035980225, -0.066223145, 0.020599365, 0.0103302, -0.06427002, -0.009681702, -0.011726379, 0.036315918, 0.0043640137, 0.008338928, 0.015113831, -0.02619934, -0.037109375, 0.020706177, -0.0231781, 0.04946...
http://arxiv.org/abs/2309.16039v3
Effective Long-Context Scaling of Foundation Models
Wenhan Xiong, Jingyu Liu, Igor Molybog, Hejia Zhang, Prajjwal Bhargava, Rui Hou, Louis Martin, Rashi Rungta, Karthik Abinav Sankararaman, Barlas Oguz, Madian Khabsa, Han Fang, Yashar Mehdad, Sharan Narang, Kshitiz Malik, Angela Fan, Shruti Bhosale, Sergey Edunov, Mike Lewis, Sinong Wang, Hao Ma
cs.CL
We present a series of long-context LLMs that support effective context windows of up to 32,768 tokens. Our model series are built through continual pretraining from Llama 2 with longer training sequences and on a dataset where long texts are upsampled. We perform extensive evaluation on language modeling, synthetic co...
http://arxiv.org/pdf/2309.16039v3
2023-11-14T01:40:13
2023-09-27T21:41:49
{ "title": [ -0.022125244, 0.008796692, -0.046447754, -0.06286621, 0.011703491, -0.0028686523, -0.03237915, 0.002216339, 0.022705078, 0.03086853, -0.0062828064, 0.007232666, -0.008911133, -0.030471802, -0.030151367, 0.02696228, -0.0040245056, 0.0...
http://arxiv.org/abs/2309.16035v1
MedEdit: Model Editing for Medical Question Answering with External Knowledge Bases
Yucheng Shi, Shaochen Xu, Zhengliang Liu, Tianming Liu, Xiang Li, Ninghao Liu
cs.CL, cs.AI
Large Language Models (LLMs), although powerful in general domains, often perform poorly on domain-specific tasks like medical question answering (QA). Moreover, they tend to function as "black-boxes," making it challenging to modify their behavior. Addressing this, our study delves into model editing utilizing in-cont...
http://arxiv.org/pdf/2309.16035v1
2023-09-27T21:26:03
2023-09-27T21:26:03
{ "title": [ -0.011360168, -0.010673523, -0.028320312, -0.070251465, 0.009597778, 0.018737793, -0.04058838, -0.013442993, -0.0026950836, 0.029632568, -0.012718201, -0.00573349, 0.0043563843, -0.010482788, -0.046173096, 0.00034475327, 0.0051727295, ...
http://arxiv.org/abs/2309.15991v1
Targeted Image Data Augmentation Increases Basic Skills Captioning Robustness
Valentin Barriere, Felipe del Rio, Andres Carvallo De Ferari, Carlos Aspillaga, Eugenio Herrera-Berg, Cristian Buc Calderon
cs.CV, cs.CL
Artificial neural networks typically struggle in generalizing to out-of-context examples. One reason for this limitation is caused by having datasets that incorporate only partial information regarding the potential correlational structure of the world. In this work, we propose TIDA (Targeted Image-editing Data Augment...
http://arxiv.org/pdf/2309.15991v1
2023-09-27T20:12:41
2023-09-27T20:12:41
{ "title": [ 0.07714844, 0.014053345, -0.035461426, -0.034606934, -0.0020866394, -0.051727295, 0.014213562, -0.014152527, -0.0054244995, 0.045959473, -0.05343628, 0.039123535, 0.029907227, -0.004634857, 0.0012569427, 0.0019426346, 0.0045204163, 0...
http://arxiv.org/abs/2309.15826v1
Cross-Modal Multi-Tasking for Speech-to-Text Translation via Hard Parameter Sharing
Brian Yan, Xuankai Chang, Antonios Anastasopoulos, Yuya Fujita, Shinji Watanabe
cs.CL, cs.SD, eess.AS
Recent works in end-to-end speech-to-text translation (ST) have proposed multi-tasking methods with soft parameter sharing which leverage machine translation (MT) data via secondary encoders that map text inputs to an eventual cross-modal representation. In this work, we instead propose a ST/MT multi-tasking framework ...
http://arxiv.org/pdf/2309.15826v1
2023-09-27T17:48:14
2023-09-27T17:48:14
{ "title": [ -0.025222778, -0.011138916, -0.021469116, -0.068725586, -0.01146698, 0.005680084, -0.060424805, 0.0047454834, -0.009552002, 0.069885254, -0.007671356, 0.032287598, 0.04660034, -0.0077590942, -0.040649414, 0.023910522, -0.0062942505, ...
http://arxiv.org/abs/2309.15806v3
Lyra: Orchestrating Dual Correction in Automated Theorem Proving
Chuanyang Zheng, Haiming Wang, Enze Xie, Zhengying Liu, Jiankai Sun, Huajian Xin, Jianhao Shen, Zhenguo Li, Yu Li
cs.CL, cs.AI
Large Language Models (LLMs) present an intriguing avenue for exploration in the field of formal theorem proving. Nevertheless, their full potential, particularly concerning the mitigation of hallucinations and refinement through prover error messages, remains an area that has yet to be thoroughly investigated. To enha...
http://arxiv.org/pdf/2309.15806v3
2023-10-07T07:54:19
2023-09-27T17:29:41
{ "title": [ -0.026535034, -0.008712769, -0.045898438, -0.09008789, 0.01675415, 0.010673523, -0.041534424, 0.019607544, -0.019424438, 0.023880005, -0.033416748, 0.011047363, 0.0016460419, -0.006298065, -0.014808655, -0.015991211, -0.016281128, 0....
http://arxiv.org/abs/2309.15800v1
Exploring Speech Recognition, Translation, and Understanding with Discrete Speech Units: A Comparative Study
Xuankai Chang, Brian Yan, Kwanghee Choi, Jeeweon Jung, Yichen Lu, Soumi Maiti, Roshan Sharma, Jiatong Shi, Jinchuan Tian, Shinji Watanabe, Yuya Fujita, Takashi Maekaku, Pengcheng Guo, Yao-Fei Cheng, Pavel Denisov, Kohei Saijo, Hsiu-Hsuan Wang
cs.CL, cs.SD, eess.AS
Speech signals, typically sampled at rates in the tens of thousands per second, contain redundancies, evoking inefficiencies in sequence modeling. High-dimensional speech features such as spectrograms are often used as the input for the subsequent model. However, they can still be redundant. Recent investigations propo...
http://arxiv.org/pdf/2309.15800v1
2023-09-27T17:21:13
2023-09-27T17:21:13
{ "title": [ -0.038513184, -0.011604309, -0.041381836, -0.05532837, 0.0062561035, -0.011688232, -0.045806885, 0.030166626, -0.031799316, 0.03878784, -0.036956787, 0.009048462, 0.015113831, -0.012924194, -0.051757812, -0.01864624, -0.0132369995, 0...
http://arxiv.org/abs/2309.15789v1
Large Language Model Routing with Benchmark Datasets
Tal Shnitzer, Anthony Ou, Mírian Silva, Kate Soule, Yuekai Sun, Justin Solomon, Neil Thompson, Mikhail Yurochkin
cs.CL, cs.LG, I.2.7, I.2.6
There is a rapidly growing number of open-source Large Language Models (LLMs) and benchmark datasets to compare them. While some models dominate these benchmarks, no single model typically achieves the best accuracy in all tasks and use cases. In this work, we address the challenge of selecting the best LLM out of a co...
http://arxiv.org/pdf/2309.15789v1
2023-09-27T17:08:40
2023-09-27T17:08:40
{ "title": [ -0.051574707, -0.023605347, -0.047973633, -0.062805176, -0.015716553, 0.010421753, -0.086242676, 0.031158447, 0.00095033646, 0.034179688, 0.001745224, 0.008171082, -0.017288208, -0.023773193, -0.0385437, 0.022644043, -0.00018107891, ...
http://arxiv.org/abs/2309.15779v1
Question answering using deep learning in low resource Indian language Marathi
Dhiraj Amin, Sharvari Govilkar, Sagar Kulkarni
cs.CL
Precise answers are extracted from a text for a given input question in a question answering system. Marathi question answering system is created in recent studies by using ontology, rule base and machine learning based approaches. Recently transformer models and transfer learning approaches are used to solve question ...
http://arxiv.org/pdf/2309.15779v1
2023-09-27T16:53:11
2023-09-27T16:53:11
{ "title": [ -0.026641846, -0.004096985, -0.03845215, -0.05331421, -0.009117126, -0.012969971, -0.047912598, -0.00422287, -0.0071144104, 0.029174805, -0.0236969, -0.01626587, 0.012565613, -0.024429321, -0.037841797, -0.008285522, 0.010528564, 0.0...
http://arxiv.org/abs/2310.01425v2
Borges and AI
Léon Bottou, Bernhard Schölkopf
cs.CL, cs.AI, cs.LG
Many believe that Large Language Models (LLMs) open the era of Artificial Intelligence (AI). Some see opportunities while others see dangers. Yet both proponents and opponents grasp AI through the imagery popularised by science fiction. Will the machine become sentient and rebel against its creators? Will we experience...
http://arxiv.org/pdf/2310.01425v2
2023-10-04T18:10:30
2023-09-27T16:15:34
{ "title": [ -0.0074539185, -0.016296387, 0.0018033981, 0.0046043396, -0.06542969, -0.02128601, 0.021636963, -0.009887695, 0.043762207, 0.050872803, -0.060913086, -0.010871887, 0.0050735474, 0.0037288666, -0.016479492, -0.007621765, -0.016464233, ...
http://arxiv.org/abs/2309.15739v1
Experience and Evidence are the eyes of an excellent summarizer! Towards Knowledge Infused Multi-modal Clinical Conversation Summarization
Abhisek Tiwari, Anisha Saha, Sriparna Saha, Pushpak Bhattacharyya, Minakshi Dhar
cs.CL, cs.AI
With the advancement of telemedicine, both researchers and medical practitioners are working hand-in-hand to develop various techniques to automate various medical operations, such as diagnosis report generation. In this paper, we first present a multi-modal clinical conversation summary generation task that takes a cl...
http://arxiv.org/pdf/2309.15739v1
2023-09-27T15:49:43
2023-09-27T15:49:43
{ "title": [ 0.028442383, -0.009117126, -0.0019836426, -0.093933105, -0.035736084, 0.010856628, -0.09539795, -0.00014781952, 0.014129639, 0.050689697, -0.00034809113, 0.028533936, 0.025238037, -0.021713257, -0.033935547, -0.0027313232, -0.000872612, ...
http://arxiv.org/abs/2310.01424v1
Identifying and Mitigating Privacy Risks Stemming from Language Models: A Survey
Victoria Smith, Ali Shahin Shamsabadi, Carolyn Ashurst, Adrian Weller
cs.CL, cs.AI
Rapid advancements in language models (LMs) have led to their adoption across many sectors. Alongside the potential benefits, such models present a range of risks, including around privacy. In particular, as LMs have grown in size, the potential to memorise aspects of their training data has increased, resulting in the...
http://arxiv.org/pdf/2310.01424v1
2023-09-27T15:15:23
2023-09-27T15:15:23
{ "title": [ -0.026397705, -0.01512146, -0.050598145, -0.056152344, 0.013465881, -0.007003784, -0.040039062, 0.007701874, -0.024520874, 0.0340271, -0.0033950806, 0.018157959, 0.0030574799, -0.020126343, -0.025741577, 0.009346008, -0.01663208, 0.0...
http://arxiv.org/abs/2309.15714v2
Integrating LLM, EEG, and Eye-Tracking Biomarker Analysis for Word-Level Neural State Classification in Semantic Inference Reading Comprehension
Yuhong Zhang, Qin Li, Sujal Nahata, Tasnia Jamal, Shih-kuen Cheng, Gert Cauwenberghs, Tzyy-Ping Jung
cs.CL, cs.AI
With the recent proliferation of large language models (LLMs), such as Generative Pre-trained Transformers (GPT), there has been a significant shift in exploring human and machine comprehension of semantic language meaning. This shift calls for interdisciplinary research that bridges cognitive science and natural langu...
http://arxiv.org/pdf/2309.15714v2
2023-10-17T17:17:59
2023-09-27T15:12:08
{ "title": [ -0.044281006, -0.00447464, -0.016235352, 0.0046539307, -0.06298828, -0.012580872, -0.05819702, -0.034606934, 0.032806396, 0.010559082, -0.029022217, 0.00081539154, 0.01940918, -0.035461426, -0.010734558, -0.0029335022, -0.0073890686, ...
http://arxiv.org/abs/2309.15701v2
HyPoradise: An Open Baseline for Generative Speech Recognition with Large Language Models
Chen Chen, Yuchen Hu, Chao-Han Huck Yang, Sabato Macro Siniscalchi, Pin-Yu Chen, Eng Siong Chng
cs.CL, cs.AI, cs.LG, cs.SD, eess.AS
Advancements in deep neural networks have allowed automatic speech recognition (ASR) systems to attain human parity on several publicly available clean speech datasets. However, even state-of-the-art ASR systems experience performance degradation when confronted with adverse conditions, as a well-trained acoustic model...
http://arxiv.org/pdf/2309.15701v2
2023-10-16T05:47:42
2023-09-27T14:44:10
{ "title": [ -0.024887085, -0.0036506653, -0.07165527, -0.07702637, 0.014465332, -0.014343262, -0.052459717, 0.03765869, 0.008781433, 0.02067566, -0.032684326, -0.005180359, 0.0044441223, -0.02293396, -0.030319214, 0.0032024384, 0.0013742447, 0.0...
http://arxiv.org/abs/2309.15686v1
Enhancing End-to-End Conversational Speech Translation Through Target Language Context Utilization
Amir Hussein, Brian Yan, Antonios Anastasopoulos, Shinji Watanabe, Sanjeev Khudanpur
cs.CL, cs.SD, eess.AS
Incorporating longer context has been shown to benefit machine translation, but the inclusion of context in end-to-end speech translation (E2E-ST) remains under-studied. To bridge this gap, we introduce target language context in E2E-ST, enhancing coherence and overcoming memory constraints of extended audio segments. ...
http://arxiv.org/pdf/2309.15686v1
2023-09-27T14:32:30
2023-09-27T14:32:30
{ "title": [ -0.00070619583, -0.0006246567, -0.036254883, -0.078308105, 0.017349243, -0.0023460388, -0.050048828, 0.017852783, -0.014274597, 0.058685303, 0.014595032, 0.018936157, 0.030731201, -0.026992798, -0.03744507, 0.027648926, -0.0012426376, ...
http://arxiv.org/abs/2309.15674v1
Speech collage: code-switched audio generation by collaging monolingual corpora
Amir Hussein, Dorsa Zeinali, Ondřej Klejch, Matthew Wiesner, Brian Yan, Shammur Chowdhury, Ahmed Ali, Shinji Watanabe, Sanjeev Khudanpur
cs.SD, cs.CL, cs.LG, eess.AS
Designing effective automatic speech recognition (ASR) systems for Code-Switching (CS) often depends on the availability of the transcribed CS resources. To address data scarcity, this paper introduces Speech Collage, a method that synthesizes CS data from monolingual corpora by splicing audio segments. We further impr...
http://arxiv.org/pdf/2309.15674v1
2023-09-27T14:17:53
2023-09-27T14:17:53
{ "title": [ -0.027572632, -0.013542175, -0.055877686, -0.03024292, -0.015167236, -0.018371582, -0.0345459, 0.01309967, -0.019821167, 0.06335449, 0.011711121, -0.002670288, 0.046417236, -0.03817749, -0.02305603, -0.004463196, -0.03250122, 0.05374...
http://arxiv.org/abs/2309.15670v1
MONOVAB : An Annotated Corpus for Bangla Multi-label Emotion Detection
Sumit Kumar Banshal, Sajal Das, Shumaiya Akter Shammi, Narayan Ranjan Chakraborty
cs.LG, cs.CL
In recent years, Sentiment Analysis (SA) and Emotion Recognition (ER) have been increasingly popular in the Bangla language, which is the seventh most spoken language throughout the entire world. However, the language is structurally complicated, which makes this field arduous to extract emotions in an accurate manner....
http://arxiv.org/pdf/2309.15670v1
2023-09-27T14:10:57
2023-09-27T14:10:57
{ "title": [ -0.008636475, -0.0104904175, -0.02746582, -0.04928589, 0.0036315918, -0.008255005, -0.0070991516, 0.0056991577, 0.028625488, 0.028213501, -0.015472412, -0.008071899, 0.00831604, -0.033111572, -0.04837036, -0.0041618347, -0.016860962, ...
http://arxiv.org/abs/2309.15656v1
Conversational Feedback in Scripted versus Spontaneous Dialogues: A Comparative Analysis
Ildikó Pilán, Laurent Prévot, Hendrik Buschmeier, Pierre Lison
cs.CL
Scripted dialogues such as movie and TV subtitles constitute a widespread source of training data for conversational NLP models. However, the linguistic characteristics of those dialogues are notably different from those observed in corpora of spontaneous interactions. This difference is particularly marked for communi...
http://arxiv.org/pdf/2309.15656v1
2023-09-27T13:45:38
2023-09-27T13:45:38
{ "title": [ -0.025482178, -0.020767212, -0.01701355, -0.04244995, -0.01687622, -0.00060224533, -0.03677368, -0.00642395, -0.0068893433, 0.016540527, -0.01725769, 0.013580322, -0.031555176, -0.013175964, -0.047180176, -0.02822876, -0.009414673, 0...
http://arxiv.org/abs/2309.15649v2
Generative Speech Recognition Error Correction with Large Language Models and Task-Activating Prompting
Chao-Han Huck Yang, Yile Gu, Yi-Chieh Liu, Shalini Ghosh, Ivan Bulyko, Andreas Stolcke
cs.CL, cs.AI, cs.LG, cs.SD, eess.AS
We explore the ability of large language models (LLMs) to act as speech recognition post-processors that perform rescoring and error correction. Our first focus is on instruction prompting to let LLMs perform these task without fine-tuning, for which we evaluate different prompting schemes, both zero- and few-shot in-c...
http://arxiv.org/pdf/2309.15649v2
2023-10-10T09:10:58
2023-09-27T13:36:03
{ "title": [ -0.007858276, 0.001086235, -0.060302734, -0.05645752, 0.00831604, -0.012260437, -0.055877686, 0.0073165894, 0.0005784035, 0.031829834, -0.042236328, -0.0052604675, 0.017166138, -0.020996094, -0.035217285, -0.0063285828, 0.002960205, ...
http://arxiv.org/abs/2309.15630v4
NLPBench: Evaluating Large Language Models on Solving NLP Problems
Linxin Song, Jieyu Zhang, Lechao Cheng, Pengyuan Zhou, Tianyi Zhou, Irene Li
cs.CL
Recent developments in large language models (LLMs) have shown promise in enhancing the capabilities of natural language processing (NLP). Despite these successes, there remains a dearth of research dedicated to the NLP problem-solving abilities of LLMs. To fill the gap in this area, we present a unique benchmarking da...
http://arxiv.org/pdf/2309.15630v4
2023-10-19T05:58:31
2023-09-27T13:02:06
{ "title": [ -0.030654907, -0.023956299, -0.031341553, -0.052581787, 0.010147095, 0.008255005, -0.03857422, -0.0041236877, 0.016647339, 0.018447876, -0.020462036, 0.0023078918, -0.00013244152, -0.014350891, -0.046325684, 0.010139465, -0.021408081, ...
http://arxiv.org/abs/2310.01423v1
An Empirical Study of AI Generated Text Detection Tools
Arslan Akram
cs.CL, cs.AI, cs.LG, 53, I.2.6; I.2.7
Since ChatGPT has emerged as a major AIGC model, providing high-quality responses across a wide range of applications (including software development and maintenance), it has attracted much interest from many individuals. ChatGPT has great promise, but there are serious problems that might arise from its misuse, especi...
http://arxiv.org/pdf/2310.01423v1
2023-09-27T12:44:12
2023-09-27T12:44:12
{ "title": [ -0.028961182, 0.0025901794, -0.021026611, -0.0050354004, -0.002527237, -0.0090789795, 0.0052604675, -0.02784729, -0.0073242188, 0.035369873, -0.034484863, 0.010658264, -0.00017654896, -0.00049209595, -0.046020508, -0.005077362, -0.023727...
http://arxiv.org/abs/2309.15609v1
Developing automatic verbatim transcripts for international multilingual meetings: an end-to-end solution
Akshat Dewan, Michal Ziemski, Henri Meylan, Lorenzo Concina, Bruno Pouliquen
cs.CL, cs.AI, cs.LG
This paper presents an end-to-end solution for the creation of fully automated conference meeting transcripts and their machine translations into various languages. This tool has been developed at the World Intellectual Property Organization (WIPO) using in-house developed speech-to-text (S2T) and machine translation (...
http://arxiv.org/pdf/2309.15609v1
2023-09-27T12:16:15
2023-09-27T12:16:15
{ "title": [ -0.029037476, 0.007205963, -0.05859375, -0.08319092, 0.0028419495, 0.0064964294, -0.08050537, 0.014907837, 0.01259613, 0.04586792, -0.0052986145, 0.002216339, 0.017196655, -0.012001038, -0.050323486, 0.015151978, -0.012542725, 0.0308...
http://arxiv.org/abs/2309.15588v1
Few-Shot Multi-Label Aspect Category Detection Utilizing Prototypical Network with Sentence-Level Weighting and Label Augmentation
Zeyu Wang, Mizuho Iwaihara
cs.CL
Multi-label aspect category detection is intended to detect multiple aspect categories occurring in a given sentence. Since aspect category detection often suffers from limited datasets and data sparsity, the prototypical network with attention mechanisms has been applied for few-shot aspect category detection. Neverth...
http://arxiv.org/pdf/2309.15588v1
2023-09-27T11:44:04
2023-09-27T11:44:04
{ "title": [ 0.013000488, -0.016616821, -0.03515625, -0.042938232, -0.0256958, -0.011817932, -0.026489258, 0.013023376, 0.018096924, 0.021057129, 0.030883789, 0.0004901886, -0.00856781, -0.047668457, -0.043121338, 0.0043754578, -0.009529114, 0.03...
http://arxiv.org/abs/2309.15564v2
Jointly Training Large Autoregressive Multimodal Models
Emanuele Aiello, Lili Yu, Yixin Nie, Armen Aghajanyan, Barlas Oguz
cs.LG, cs.CL, cs.CV
In recent years, advances in the large-scale pretraining of language and text-to-image models have revolutionized the field of machine learning. Yet, integrating these two modalities into a single, robust model capable of generating seamless multimodal outputs remains a significant challenge. To address this gap, we pr...
http://arxiv.org/pdf/2309.15564v2
2023-09-28T17:23:11
2023-09-27T10:40:23
{ "title": [ -0.0141067505, 0.018722534, -0.04788208, -0.062927246, 0.022537231, -0.0031280518, -0.051605225, -0.024017334, -0.0030002594, 0.032226562, 0.00010895729, -0.0019702911, -0.01802063, -0.021759033, -0.043914795, 0.00095415115, -0.013252258...
http://arxiv.org/abs/2309.15554v1
Direct Models for Simultaneous Translation and Automatic Subtitling: FBK@IWSLT2023
Sara Papi, Marco Gaido, Matteo Negri
cs.CL, cs.AI, cs.SD, eess.AS
This paper describes the FBK's participation in the Simultaneous Translation and Automatic Subtitling tracks of the IWSLT 2023 Evaluation Campaign. Our submission focused on the use of direct architectures to perform both tasks: for the simultaneous one, we leveraged the knowledge already acquired by offline-trained mo...
http://arxiv.org/pdf/2309.15554v1
2023-09-27T10:24:42
2023-09-27T10:24:42
{ "title": [ -0.025512695, 0.0003376007, -0.024612427, -0.048431396, -0.0101623535, 0.005844116, -0.052642822, -0.011039734, -0.014266968, 0.07183838, 0.008102417, -0.012039185, 0.05606079, -0.03869629, -0.029174805, 0.02067566, 0.010063171, 0.00...
http://arxiv.org/abs/2309.15516v1
Teaching Text-to-Image Models to Communicate
Xiaowen Sun, Jiazhan Feng, Yuxuan Wang, Yuxuan Lai, Xingyu Shen, Dongyan Zhao
cs.CL, cs.AI, cs.CV
Various works have been extensively studied in the research of text-to-image generation. Although existing models perform well in text-to-image generation, there are significant challenges when directly employing them to generate images in dialogs. In this paper, we first highlight a new problem: dialog-to-image genera...
http://arxiv.org/pdf/2309.15516v1
2023-09-27T09:33:16
2023-09-27T09:33:16
{ "title": [ 0.00056934357, 0.006298065, 0.005706787, -0.009475708, -0.03768921, -0.003993988, -0.011817932, -0.039276123, 0.026397705, 0.042633057, -0.044128418, -0.012825012, 0.0020942688, -0.024520874, -0.03942871, -0.01436615, 0.01637268, 0.0...
http://arxiv.org/abs/2309.15512v1
High-Fidelity Speech Synthesis with Minimal Supervision: All Using Diffusion Models
Chunyu Qiang, Hao Li, Yixin Tian, Yi Zhao, Ying Zhang, Longbiao Wang, Jianwu Dang
cs.SD, cs.AI, cs.CL, eess.AS
Text-to-speech (TTS) methods have shown promising results in voice cloning, but they require a large number of labeled text-speech pairs. Minimally-supervised speech synthesis decouples TTS by combining two types of discrete speech representations(semantic \& acoustic) and using two sequence-to-sequence tasks to enable...
http://arxiv.org/pdf/2309.15512v1
2023-09-27T09:27:03
2023-09-27T09:27:03
{ "title": [ -0.0001282692, 0.008171082, -0.061645508, -0.06781006, 0.008071899, -0.025817871, -0.03982544, 0.01826477, -0.0070877075, 0.049438477, -0.009567261, 0.0028591156, 0.027832031, -0.039733887, -0.011672974, 0.017837524, -0.004486084, 0....
http://arxiv.org/abs/2309.15494v1
VideoAdviser: Video Knowledge Distillation for Multimodal Transfer Learning
Yanan Wang, Donghuo Zeng, Shinya Wada, Satoshi Kurihara
cs.CV, cs.CL
Multimodal transfer learning aims to transform pretrained representations of diverse modalities into a common domain space for effective multimodal fusion. However, conventional systems are typically built on the assumption that all modalities exist, and the lack of modalities always leads to poor inference performance...
http://arxiv.org/pdf/2309.15494v1
2023-09-27T08:44:04
2023-09-27T08:44:04
{ "title": [ -0.017623901, 0.010070801, -0.07116699, -0.05142212, 0.021621704, -0.030685425, -0.028076172, -0.017242432, 0.008087158, 0.014839172, -0.03894043, 0.016021729, -0.01663208, -0.010002136, -0.031829834, -0.019119263, -0.0013313293, 0.0...
http://arxiv.org/abs/2309.15476v1
Dynamic Multi-Scale Context Aggregation for Conversational Aspect-Based Sentiment Quadruple Analysis
Yuqing Li, Wenyuan Zhang, Binbin Li, Siyu Jia, Zisen Qi, Xingbang Tan
cs.CL
Conversational aspect-based sentiment quadruple analysis (DiaASQ) aims to extract the quadruple of target-aspect-opinion-sentiment within a dialogue. In DiaASQ, a quadruple's elements often cross multiple utterances. This situation complicates the extraction process, emphasizing the need for an adequate understanding o...
http://arxiv.org/pdf/2309.15476v1
2023-09-27T08:17:28
2023-09-27T08:17:28
{ "title": [ 0.019256592, -0.005153656, -0.054229736, -0.097717285, 0.027755737, -0.0069732666, -0.029800415, 0.03225708, 0.037322998, 0.06921387, -0.021621704, -0.0047073364, 0.018630981, -0.0030326843, -0.08251953, 0.007286072, -0.0021533966, 0...
http://arxiv.org/abs/2309.15461v1
ChatCounselor: A Large Language Models for Mental Health Support
June M. Liu, Donghao Li, He Cao, Tianhe Ren, Zeyi Liao, Jiamin Wu
cs.CL
This paper presents ChatCounselor, a large language model (LLM) solution designed to provide mental health support. Unlike generic chatbots, ChatCounselor is distinguished by its foundation in real conversations between consulting clients and professional psychologists, enabling it to possess specialized knowledge and ...
http://arxiv.org/pdf/2309.15461v1
2023-09-27T07:57:21
2023-09-27T07:57:21
{ "title": [ -0.014274597, -0.024597168, -0.029769897, -0.08898926, 0.0049972534, 0.0181427, -0.064208984, 0.012176514, 0.0070877075, 0.042541504, -0.010314941, -0.016555786, 0.02935791, -0.022399902, -0.042144775, 0.01007843, -0.0050697327, 0.02...
http://arxiv.org/abs/2309.15427v1
Graph Neural Prompting with Large Language Models
Yijun Tian, Huan Song, Zichen Wang, Haozhu Wang, Ziqing Hu, Fang Wang, Nitesh V. Chawla, Panpan Xu
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have shown remarkable generalization capability with exceptional performance in various language modeling tasks. However, they still exhibit inherent limitations in precisely capturing and returning grounded knowledge. While existing work has explored utilizing knowledge graphs to enhance l...
http://arxiv.org/pdf/2309.15427v1
2023-09-27T06:33:29
2023-09-27T06:33:29
{ "title": [ -0.018478394, -0.004119873, -0.015838623, -0.05041504, -0.019439697, 0.0056495667, -0.06591797, -0.0047912598, -0.0058555603, 0.010398865, -0.011978149, -0.011726379, -0.013420105, -0.033721924, -0.019302368, 0.0076675415, 0.0126953125, ...
http://arxiv.org/abs/2309.15402v2
A Survey of Chain of Thought Reasoning: Advances, Frontiers and Future
Zheng Chu, Jingchang Chen, Qianglong Chen, Weijiang Yu, Tao He, Haotian Wang, Weihua Peng, Ming Liu, Bing Qin, Ting Liu
cs.CL, cs.AI
Chain-of-thought reasoning, a cognitive process fundamental to human intelligence, has garnered significant attention in the realm of artificial intelligence and natural language processing. However, there still remains a lack of a comprehensive survey for this arena. To this end, we take the first step and present a t...
http://arxiv.org/pdf/2309.15402v2
2023-10-16T07:37:45
2023-09-27T04:53:10
{ "title": [ -0.006664276, -0.012458801, -0.031311035, -0.050842285, 0.012489319, -0.005874634, -0.012786865, -0.0041656494, 0.0027694702, 0.011947632, -0.023330688, -0.0011262894, -0.014427185, -0.029510498, -0.028213501, -0.0022888184, -0.013267517...
http://arxiv.org/abs/2309.15337v1
Beyond the Chat: Executable and Verifiable Text-Editing with LLMs
Philippe Laban, Jesse Vig, Marti A. Hearst, Caiming Xiong, Chien-Sheng Wu
cs.CL, cs.HC
Conversational interfaces powered by Large Language Models (LLMs) have recently become a popular way to obtain feedback during document editing. However, standard chat-based conversational interfaces do not support transparency and verifiability of the editing changes that they suggest. To give the author more agency w...
http://arxiv.org/pdf/2309.15337v1
2023-09-27T00:56:17
2023-09-27T00:56:17
{ "title": [ -0.051818848, -0.01020813, -0.047729492, -0.04699707, -0.013832092, 0.008872986, -0.060638428, 0.0052375793, -0.038238525, 0.02748108, -0.01838684, 0.031219482, 0.0037384033, -0.028915405, -0.01108551, -0.01374054, -0.030792236, 0.04...
http://arxiv.org/abs/2309.15317v2
Joint Prediction and Denoising for Large-scale Multilingual Self-supervised Learning
William Chen, Jiatong Shi, Brian Yan, Dan Berrebbi, Wangyou Zhang, Yifan Peng, Xuankai Chang, Soumi Maiti, Shinji Watanabe
cs.CL, cs.AI, cs.SD, eess.AS
Multilingual self-supervised learning (SSL) has often lagged behind state-of-the-art (SOTA) methods due to the expenses and complexity required to handle many languages. This further harms the reproducibility of SSL, which is already limited to few research groups due to its resource usage. We show that more powerful t...
http://arxiv.org/pdf/2309.15317v2
2023-09-28T02:28:16
2023-09-26T23:55:57
{ "title": [ -0.015945435, 0.025039673, -0.05331421, -0.07562256, 0.02268982, 0.001124382, -0.05130005, -0.004699707, -0.0058937073, 0.037902832, 0.0011320114, 0.0096206665, -0.02571106, -0.019836426, -0.035064697, 0.0030822754, 0.0057907104, 0.0...
http://arxiv.org/abs/2310.01420v2
Ruffle&Riley: Towards the Automated Induction of Conversational Tutoring Systems
Robin Schmucker, Meng Xia, Amos Azaria, Tom Mitchell
cs.CL, cs.AI, cs.HC
Conversational tutoring systems (CTSs) offer learning experiences driven by natural language interaction. They are known to promote high levels of cognitive engagement and benefit learning outcomes, particularly in reasoning tasks. Nonetheless, the time and cost required to author CTS content is a major obstacle to wid...
http://arxiv.org/pdf/2310.01420v2
2023-11-14T15:41:45
2023-09-26T23:27:06
{ "title": [ -0.020217896, -0.011024475, -0.03189087, -0.024963379, -0.010635376, -0.0027503967, -0.026779175, -0.0044136047, -0.002374649, 0.01864624, -0.040008545, 0.023803711, -0.0017318726, -0.014671326, -0.027389526, -0.014694214, -0.028518677, ...
http://arxiv.org/abs/2309.15869v1
Unsupervised Pre-Training for Vietnamese Automatic Speech Recognition in the HYKIST Project
Khai Le-Duc
cs.CL, cs.SD, eess.AS
In today's interconnected globe, moving abroad is more and more prevalent, whether it's for employment, refugee resettlement, or other causes. Language difficulties between natives and immigrants present a common issue on a daily basis, especially in medical domain. This can make it difficult for patients and doctors t...
http://arxiv.org/pdf/2309.15869v1
2023-09-26T21:12:09
2023-09-26T21:12:09
{ "title": [ -0.0073432922, -0.00052165985, -0.044769287, -0.042633057, -0.014465332, -0.009109497, -0.033935547, 0.003932953, -0.023101807, 0.030349731, -0.03250122, -0.0021476746, 0.01676941, -0.042785645, -0.0052986145, -0.028213501, 0.00774765, ...
http://arxiv.org/abs/2311.01463v1
Creating Trustworthy LLMs: Dealing with Hallucinations in Healthcare AI
Muhammad Aurangzeb Ahmad, Ilker Yaramis, Taposh Dutta Roy
cs.CL, cs.AI, cs.CV, cs.LG, cs.NE
Large language models have proliferated across multiple domains in as short period of time. There is however hesitation in the medical and healthcare domain towards their adoption because of issues like factuality, coherence, and hallucinations. Give the high stakes nature of healthcare, many researchers have even caut...
http://arxiv.org/pdf/2311.01463v1
2023-09-26T20:52:46
2023-09-26T20:52:46
{ "title": [ -0.015975952, 0.0038375854, -0.05709839, -0.015289307, -0.0030498505, -0.026947021, -0.00818634, -0.0036201477, -0.04711914, 0.021560669, -0.00749588, 0.03302002, -0.0011148453, -0.047027588, 0.019226074, -0.023788452, 0.0075950623, ...
http://arxiv.org/abs/2309.15238v1
Learning Using Generated Privileged Information by Text-to-Image Diffusion Models
Rafael-Edy Menadil, Mariana-Iuliana Georgescu, Radu Tudor Ionescu
cs.CL, cs.AI, cs.LG
Learning Using Privileged Information is a particular type of knowledge distillation where the teacher model benefits from an additional data representation during training, called privileged information, improving the student model, which does not see the extra representation. However, privileged information is rarely...
http://arxiv.org/pdf/2309.15238v1
2023-09-26T20:04:48
2023-09-26T20:04:48
{ "title": [ 0.0059928894, 0.014678955, -0.024642944, -0.022613525, 0.0017061234, -0.013511658, 0.008979797, -0.03375244, 0.030273438, 0.016464233, -0.03744507, 0.012481689, -0.023086548, -0.023422241, -0.03753662, -0.0104599, -0.015052795, 0.054...
http://arxiv.org/abs/2309.15223v2
Low-rank Adaptation of Large Language Model Rescoring for Parameter-Efficient Speech Recognition
Yu Yu, Chao-Han Huck Yang, Jari Kolehmainen, Prashanth G. Shivakumar, Yile Gu, Sungho Ryu, Roger Ren, Qi Luo, Aditya Gourav, I-Fan Chen, Yi-Chieh Liu, Tuan Dinh, Ankur Gandhe, Denis Filimonov, Shalini Ghosh, Andreas Stolcke, Ariya Rastow, Ivan Bulyko
cs.CL, cs.AI, cs.LG, cs.NE, cs.SD, eess.AS
We propose a neural language modeling system based on low-rank adaptation (LoRA) for speech recognition output rescoring. Although pretrained language models (LMs) like BERT have shown superior performance in second-pass rescoring, the high computational cost of scaling up the pretraining stage and adapting the pretrai...
http://arxiv.org/pdf/2309.15223v2
2023-10-10T06:30:47
2023-09-26T19:41:34
{ "title": [ -0.018630981, -0.01436615, -0.02670288, -0.06463623, -0.012069702, 0.0078048706, -0.045898438, 0.0155181885, 0.004585266, 0.032104492, -0.020599365, 0.0007944107, -0.0014724731, -0.015777588, -0.034729004, 0.0038909912, 0.006362915, ...
http://arxiv.org/abs/2309.15217v1
RAGAS: Automated Evaluation of Retrieval Augmented Generation
Shahul Es, Jithin James, Luis Espinosa-Anke, Steven Schockaert
cs.CL
We introduce RAGAs (Retrieval Augmented Generation Assessment), a framework for reference-free evaluation of Retrieval Augmented Generation (RAG) pipelines. RAG systems are composed of a retrieval and an LLM based generation module, and provide LLMs with knowledge from a reference textual database, which enables them t...
http://arxiv.org/pdf/2309.15217v1
2023-09-26T19:23:54
2023-09-26T19:23:54
{ "title": [ -0.001285553, 0.0032863617, -0.04534912, -0.05126953, 0.018432617, -0.010139465, -0.023361206, -0.02861023, 0.02696228, 0.031677246, -0.009727478, 0.010971069, -0.005630493, -0.019744873, -0.05819702, 0.014251709, -0.019241333, 0.031...
http://arxiv.org/abs/2309.15176v1
STANCE-C3: Domain-adaptive Cross-target Stance Detection via Contrastive Learning and Counterfactual Generation
Nayoung Kim, David Mosallanezhad, Lu Cheng, Michelle V. Mancenido, Huan Liu
cs.CL, cs.SI
Stance detection is the process of inferring a person's position or standpoint on a specific issue to deduce prevailing perceptions toward topics of general or controversial interest, such as health policies during the COVID-19 pandemic. Existing models for stance detection are trained to perform well for a single doma...
http://arxiv.org/pdf/2309.15176v1
2023-09-26T18:19:51
2023-09-26T18:19:51
{ "title": [ -0.008300781, 0.009567261, -0.032562256, -0.01586914, 0.0046577454, -0.033843994, -0.000025570393, -0.04168701, 0.024291992, 0.023513794, -0.04168701, 0.008033752, -0.008842468, -0.023010254, -0.038085938, -0.001045227, -0.008026123, ...
http://arxiv.org/abs/2309.15098v1
Attention Satisfies: A Constraint-Satisfaction Lens on Factual Errors of Language Models
Mert Yuksekgonul, Varun Chandrasekaran, Erik Jones, Suriya Gunasekar, Ranjita Naik, Hamid Palangi, Ece Kamar, Besmira Nushi
cs.CL, cs.AI, cs.LG
We investigate the internal behavior of Transformer-based Large Language Models (LLMs) when they generate factually incorrect text. We propose modeling factual queries as Constraint Satisfaction Problems and use this framework to investigate how the model interacts internally with factual constraints. Specifically, we ...
http://arxiv.org/pdf/2309.15098v1
2023-09-26T17:48:55
2023-09-26T17:48:55
{ "title": [ -0.011230469, -0.0078086853, -0.029968262, -0.06329346, -0.00017929077, 0.0024776459, -0.05429077, 0.00035643578, 0.01309967, 0.023239136, -0.04345703, -0.022537231, 0.04196167, -0.002696991, -0.046875, -0.004749298, -0.03387451, 0.0...
http://arxiv.org/abs/2309.15091v1
VideoDirectorGPT: Consistent Multi-scene Video Generation via LLM-Guided Planning
Han Lin, Abhay Zala, Jaemin Cho, Mohit Bansal
cs.CV, cs.AI, cs.CL, cs.LG
Although recent text-to-video (T2V) generation methods have seen significant advancements, most of these works focus on producing short video clips of a single event with a single background (i.e., single-scene videos). Meanwhile, recent large language models (LLMs) have demonstrated their capability in generating layo...
http://arxiv.org/pdf/2309.15091v1
2023-09-26T17:36:26
2023-09-26T17:36:26
{ "title": [ -0.037902832, -0.004257202, -0.057006836, -0.050628662, -0.007156372, -0.004600525, -0.058135986, -0.0070495605, -0.048980713, 0.013084412, -0.03326416, 0.059783936, -0.042785645, -0.016998291, 0.0031032562, -0.02192688, -0.0047569275, ...
http://arxiv.org/abs/2309.15088v1
RankVicuna: Zero-Shot Listwise Document Reranking with Open-Source Large Language Models
Ronak Pradeep, Sahel Sharifymoghaddam, Jimmy Lin
cs.IR, cs.CL
Researchers have successfully applied large language models (LLMs) such as ChatGPT to reranking in an information retrieval context, but to date, such work has mostly been built on proprietary models hidden behind opaque API endpoints. This approach yields experimental results that are not reproducible and non-determin...
http://arxiv.org/pdf/2309.15088v1
2023-09-26T17:31:57
2023-09-26T17:31:57
{ "title": [ -0.013397217, -0.015411377, -0.027999878, -0.05014038, -0.011543274, 0.008895874, -0.03967285, -0.018859863, 0.021759033, 0.0385437, -0.02558899, 0.00053453445, 0.025222778, -0.005580902, -0.038604736, 0.0037899017, -0.007068634, 0.0...
http://arxiv.org/abs/2309.15840v1
How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questions
Lorenzo Pacchiardi, Alex J. Chan, Sören Mindermann, Ilan Moscovitz, Alexa Y. Pan, Yarin Gal, Owain Evans, Jan Brauner
cs.CL, cs.AI, cs.LG
Large language models (LLMs) can "lie", which we define as outputting false statements despite "knowing" the truth in a demonstrable sense. LLMs might "lie", for example, when instructed to output misinformation. Here, we develop a simple lie detector that requires neither access to the LLM's activations (black-box) no...
http://arxiv.org/pdf/2309.15840v1
2023-09-26T16:07:54
2023-09-26T16:07:54
{ "title": [ -0.033569336, 0.0034713745, -0.06188965, -0.0093307495, -0.018829346, -0.025344849, -0.019454956, -0.020965576, -0.026321411, 0.015396118, -0.040618896, 0.04147339, -0.0128479, -0.015533447, -0.005683899, -0.029281616, -0.02784729, 0...
http://arxiv.org/abs/2309.15028v2
Don't throw away your value model! Making PPO even better via Value-Guided Monte-Carlo Tree Search decoding
Jiacheng Liu, Andrew Cohen, Ramakanth Pasunuru, Yejin Choi, Hannaneh Hajishirzi, Asli Celikyilmaz
cs.CL, cs.AI, cs.LG
Inference-time search algorithms such as Monte-Carlo Tree Search (MCTS) may seem unnecessary when generating natural language text based on state-of-the-art reinforcement learning such as Proximal Policy Optimization (PPO). In this paper, we demonstrate that it is possible to get extra mileage out of PPO by integrating...
http://arxiv.org/pdf/2309.15028v2
2023-10-18T15:05:36
2023-09-26T15:57:57
{ "title": [ -0.017745972, -0.007507324, -0.034088135, -0.068115234, -0.0340271, -0.010726929, -0.08105469, -0.0067367554, -0.039764404, 0.02128601, 0.0016965866, 0.024887085, -0.015487671, -0.03805542, 0.0015068054, -0.025527954, -0.012031555, 0...
http://arxiv.org/abs/2309.15025v1
Large Language Model Alignment: A Survey
Tianhao Shen, Renren Jin, Yufei Huang, Chuang Liu, Weilong Dong, Zishan Guo, Xinwei Wu, Yan Liu, Deyi Xiong
cs.CL, cs.AI
Recent years have witnessed remarkable progress made in large language models (LLMs). Such advancements, while garnering significant attention, have concurrently elicited various concerns. The potential of these models is undeniably vast; however, they may yield texts that are imprecise, misleading, or even detrimental...
http://arxiv.org/pdf/2309.15025v1
2023-09-26T15:49:23
2023-09-26T15:49:23
{ "title": [ -0.048339844, -0.007575989, -0.0619812, -0.07714844, 0.009933472, 0.012207031, -0.0836792, 0.022155762, 0.005908966, 0.046539307, -0.0010080338, -0.005001068, 0.021697998, -0.018814087, -0.04574585, 0.016799927, -0.016952515, 0.02058...
http://arxiv.org/abs/2309.15016v1
Question-Answering Approach to Evaluate Legal Summaries
Huihui Xu, Kevin Ashley
cs.CL
Traditional evaluation metrics like ROUGE compare lexical overlap between the reference and generated summaries without taking argumentative structure into account, which is important for legal summaries. In this paper, we propose a novel legal summarization evaluation framework that utilizes GPT-4 to generate a set of...
http://arxiv.org/pdf/2309.15016v1
2023-09-26T15:36:29
2023-09-26T15:36:29
{ "title": [ -0.013221741, -0.008003235, -0.030593872, -0.05529785, -0.026992798, -0.004184723, -0.041748047, 0.016601562, 0.009239197, 0.026046753, -0.04663086, -0.0116119385, 0.0021324158, -0.009971619, -0.04611206, -0.014297485, -0.0037975311, ...
http://arxiv.org/abs/2309.15013v1
Updated Corpora and Benchmarks for Long-Form Speech Recognition
Jennifer Drexler Fox, Desh Raj, Natalie Delworth, Quinn McNamara, Corey Miller, Migüel Jetté
cs.CL, cs.SD, eess.AS
The vast majority of ASR research uses corpora in which both the training and test data have been pre-segmented into utterances. In most real-word ASR use-cases, however, test audio is not segmented, leading to a mismatch between inference-time conditions and models trained on segmented utterances. In this paper, we re...
http://arxiv.org/pdf/2309.15013v1
2023-09-26T15:32:09
2023-09-26T15:32:09
{ "title": [ -0.011871338, 0.007865906, -0.04776001, -0.077819824, 0.005016327, -0.015289307, -0.044281006, 0.032409668, -0.028869629, 0.03994751, -0.042022705, 0.017471313, -0.01159668, -0.0056533813, -0.034301758, -0.012634277, 0.013053894, 0.0...
http://arxiv.org/abs/2309.15004v1
Automating question generation from educational text
Ayan Kumar Bhowmick, Ashish Jagmohan, Aditya Vempaty, Prasenjit Dey, Leigh Hall, Jeremy Hartman, Ravi Kokku, Hema Maheshwari
cs.CL, cs.AI, cs.LG
The use of question-based activities (QBAs) is wide-spread in education, traditionally forming an integral part of the learning and assessment process. In this paper, we design and evaluate an automated question generation tool for formative and summative assessment in schools. We present an expert survey of one hundre...
http://arxiv.org/pdf/2309.15004v1
2023-09-26T15:18:44
2023-09-26T15:18:44
{ "title": [ -0.033416748, -0.004047394, -0.011405945, -0.020553589, -0.018722534, 0.014183044, -0.010726929, -0.017166138, 0.008682251, 0.022506714, -0.028686523, 0.019989014, -0.033172607, -0.0035438538, -0.04940796, -0.012916565, -0.00806427, ...
http://arxiv.org/abs/2309.14966v1
Interactively Learning Social Media Representations Improves News Source Factuality Detection
Nikhil Mehta, Dan Goldwasser
cs.CL, cs.AI, cs.SI
The rise of social media has enabled the widespread propagation of fake news, text that is published with an intent to spread misinformation and sway beliefs. Rapidly detecting fake news, especially as new events arise, is important to prevent misinformation. While prior works have tackled this problem using supervis...
http://arxiv.org/pdf/2309.14966v1
2023-09-26T14:36:19
2023-09-26T14:36:19
{ "title": [ 0.030212402, 0.022781372, -0.026351929, -0.03314209, 0.014640808, -0.03152466, 0.0099105835, -0.03866577, -0.0141067505, 0.031799316, -0.054229736, 0.032958984, -0.004760742, 0.0066452026, -0.029022217, -0.009353638, -0.024032593, 0....
http://arxiv.org/abs/2309.14913v1
Robustness of the Random Language Model
Fatemeh Lalegani, Eric De Giuli
cond-mat.dis-nn, cs.CL
The Random Language Model (De Giuli 2019) is an ensemble of stochastic context-free grammars, quantifying the syntax of human and computer languages. The model suggests a simple picture of first language learning as a type of annealing in the vast space of potential languages. In its simplest formulation, it implies a ...
http://arxiv.org/pdf/2309.14913v1
2023-09-26T13:14:35
2023-09-26T13:14:35
{ "title": [ -0.032409668, 0.0044136047, -0.039093018, -0.058898926, 0.0107040405, 0.0030326843, -0.06286621, -0.012237549, -0.002231598, 0.021591187, -0.00090169907, -0.00095653534, -0.0063323975, -0.026626587, -0.04119873, 0.017791748, -0.017349243...
http://arxiv.org/abs/2309.15796v1
Learning from Flawed Data: Weakly Supervised Automatic Speech Recognition
Dongji Gao, Hainan Xu, Desh Raj, Leibny Paola Garcia Perera, Daniel Povey, Sanjeev Khudanpur
eess.AS, cs.CL, cs.LG
Training automatic speech recognition (ASR) systems requires large amounts of well-curated paired data. However, human annotators usually perform "non-verbatim" transcription, which can result in poorly trained models. In this paper, we propose Omni-temporal Classification (OTC), a novel training criterion that explici...
http://arxiv.org/pdf/2309.15796v1
2023-09-26T12:58:40
2023-09-26T12:58:40
{ "title": [ -0.013648987, 0.004322052, -0.03277588, -0.05935669, 0.01084137, 0.0059890747, -0.037353516, -0.011772156, -0.022644043, 0.028686523, -0.052581787, 0.01689148, -0.005569458, -0.009864807, -0.02885437, -0.022781372, -0.0031642914, 0.0...
http://arxiv.org/abs/2309.14823v1
Segmentation-Free Streaming Machine Translation
Javier Iranzo-Sánchez, Jorge Iranzo-Sánchez, Adrià Giménez, Jorge Civera, Alfons Juan
cs.CL
Streaming Machine Translation (MT) is the task of translating an unbounded input text stream in real-time. The traditional cascade approach, which combines an Automatic Speech Recognition (ASR) and an MT system, relies on an intermediate segmentation step which splits the transcription stream into sentence-like units. ...
http://arxiv.org/pdf/2309.14823v1
2023-09-26T10:43:52
2023-09-26T10:43:52
{ "title": [ -0.014877319, -0.010803223, -0.0149383545, -0.078186035, -0.006816864, 0.010643005, -0.043151855, 0.013267517, -0.0029563904, 0.06365967, -0.0043144226, 0.007167816, 0.024154663, -0.012565613, -0.037902832, 0.033721924, 0.007789612, ...
http://arxiv.org/abs/2311.04906v1
FlaCGEC: A Chinese Grammatical Error Correction Dataset with Fine-grained Linguistic Annotation
Hanyue Du, Yike Zhao, Qingyuan Tian, Jiani Wang, Lei Wang, Yunshi Lan, Xuesong Lu
cs.CL, cs.AI
Chinese Grammatical Error Correction (CGEC) has been attracting growing attention from researchers recently. In spite of the fact that multiple CGEC datasets have been developed to support the research, these datasets lack the ability to provide a deep linguistic topology of grammar errors, which is critical for interp...
http://arxiv.org/pdf/2311.04906v1
2023-09-26T10:22:43
2023-09-26T10:22:43
{ "title": [ -0.0062675476, 0.0076065063, -0.032104492, -0.04876709, 0.011749268, -0.0022907257, -0.031677246, -0.0046043396, 0.018051147, 0.026306152, -0.025650024, 0.004135132, 0.023864746, -0.014770508, -0.03955078, 0.008491516, -0.03189087, 0...
http://arxiv.org/abs/2309.14805v1
Fine-tuning and aligning question answering models for complex information extraction tasks
Matthias Engelbach, Dennis Klau, Felix Scheerer, Jens Drawehn, Maximilien Kintz
cs.CL, cs.AI
The emergence of Large Language Models (LLMs) has boosted performance and possibilities in various NLP tasks. While the usage of generative AI models like ChatGPT opens up new opportunities for several business use cases, their current tendency to hallucinate fake content strongly limits their applicability to document...
http://arxiv.org/pdf/2309.14805v1
2023-09-26T10:02:21
2023-09-26T10:02:21
{ "title": [ -0.028686523, 0.005153656, -0.03982544, -0.054229736, 0.0018415451, 0.005958557, -0.038848877, 0.0011100769, 0.011291504, 0.024887085, -0.02029419, 0.006324768, -0.013175964, 0.0033874512, -0.062561035, -0.0008773804, -0.008834839, 0...
http://arxiv.org/abs/2309.14779v1
Exploring Small Language Models with Prompt-Learning Paradigm for Efficient Domain-Specific Text Classification
Hengyu Luo, Peng Liu, Stefan Esping
cs.CL, cs.AI, cs.LG
Domain-specific text classification faces the challenge of scarce labeled data due to the high cost of manual labeling. Prompt-learning, known for its efficiency in few-shot scenarios, is proposed as an alternative to traditional fine-tuning methods. And besides, although large language models (LLMs) have gained promin...
http://arxiv.org/pdf/2309.14779v1
2023-09-26T09:24:46
2023-09-26T09:24:46
{ "title": [ -0.032989502, -0.0044059753, -0.04916382, -0.04031372, -0.00655365, -0.01272583, -0.048950195, -0.0049552917, 0.0012350082, 0.022842407, -0.015388489, 0.005783081, -0.0011253357, -0.027404785, -0.03805542, -0.012901306, -0.021621704, ...
http://arxiv.org/abs/2309.14774v1
BLIP-Adapter: Parameter-Efficient Transfer Learning for Mobile Screenshot Captioning
Ching-Yu Chiang, I-Hua Chang, Shih-Wei Liao
cs.LG, cs.CL, cs.CV, cs.HC
This study aims to explore efficient tuning methods for the screenshot captioning task. Recently, image captioning has seen significant advancements, but research in captioning tasks for mobile screens remains relatively scarce. Current datasets and use cases describing user behaviors within product screenshots are not...
http://arxiv.org/pdf/2309.14774v1
2023-09-26T09:16:44
2023-09-26T09:16:44
{ "title": [ 0.0053482056, -0.00008803606, -0.0051727295, -0.021774292, -0.0061187744, -0.022201538, 0.00013816357, -0.011940002, 0.00044345856, 0.051361084, -0.03213501, 0.0345459, -0.0029563904, 0.0015640259, -0.029037476, -0.0031986237, -0.0213775...
http://arxiv.org/abs/2309.14771v1
Boosting In-Context Learning with Factual Knowledge
Jianing Wang, Chengyu Wang, Chuanqi Tan, Jun Huang, Ming Gao
cs.CL, cs.AI
In-Context Learning (ICL) over Large language models (LLMs) aims at solving previously unseen tasks by conditioning on a few training examples, eliminating the need for parameter updates and achieving competitive performance. In this paper, we demonstrate that factual knowledge is imperative for the performance of ICL ...
http://arxiv.org/pdf/2309.14771v1
2023-09-26T09:06:39
2023-09-26T09:06:39
{ "title": [ 0.010734558, 0.0042648315, -0.0541687, -0.06689453, 0.024139404, -0.030075073, -0.014457703, -0.009933472, -0.0002632141, 0.02708435, -0.028579712, 0.022384644, -0.021072388, -0.0073547363, -0.028503418, -0.005180359, -0.003358841, 0...
http://arxiv.org/abs/2309.14770v1
KERMIT: Knowledge Graph Completion of Enhanced Relation Modeling with Inverse Transformation
Haotian Li, Lingzhi Wang, Yuliang Wei, Richard Yi Da Xu, Bailing Wang
cs.CL
Knowledge graph completion is a task that revolves around filling in missing triples based on the information available in a knowledge graph. Among the current studies, text-based methods complete the task by utilizing textual descriptions of triples. However, this modeling approach may encounter limitations, particula...
http://arxiv.org/pdf/2309.14770v1
2023-09-26T09:03:25
2023-09-26T09:03:25
{ "title": [ -0.025863647, -0.02130127, -0.056884766, -0.09790039, 0.0009074211, 0.023620605, -0.07727051, 0.0039634705, 0.019165039, 0.012336731, -0.029342651, 0.02027893, 0.019226074, -0.004047394, -0.03164673, 0.013046265, -0.013328552, 0.0493...
http://arxiv.org/abs/2309.14763v1
ConPET: Continual Parameter-Efficient Tuning for Large Language Models
Chenyang Song, Xu Han, Zheni Zeng, Kuai Li, Chen Chen, Zhiyuan Liu, Maosong Sun, Tao Yang
cs.CL, I.2.7
Continual learning necessitates the continual adaptation of models to newly emerging tasks while minimizing the catastrophic forgetting of old ones. This is extremely challenging for large language models (LLMs) with vanilla full-parameter tuning due to high computation costs, memory consumption, and forgetting issue. ...
http://arxiv.org/pdf/2309.14763v1
2023-09-26T08:52:04
2023-09-26T08:52:04
{ "title": [ -0.028839111, -0.02609253, -0.030975342, -0.10101318, -0.006134033, 0.016799927, -0.08459473, 0.026016235, 0.013900757, 0.02545166, -0.012252808, 0.0034885406, 0.020736694, -0.018432617, -0.038513184, 0.019241333, -0.014625549, 0.051...
http://arxiv.org/abs/2309.14760v1
Program Repair with Minimal Edits Using CodeT5
Atsushi Shirafuji, Md. Mostafizer Rahman, Md Faizul Ibne Amin, Yutaka Watanobe
cs.CL, cs.AI, cs.SE
Programmers often struggle to identify and fix bugs in their programs. In recent years, many language models (LMs) have been proposed to fix erroneous programs and support error recovery. However, the LMs tend to generate solutions that differ from the original input programs. This leads to potential comprehension diff...
http://arxiv.org/pdf/2309.14760v1
2023-09-26T08:45:05
2023-09-26T08:45:05
{ "title": [ 0.02243042, -0.0038661957, -0.027191162, -0.077819824, -0.02470398, -0.01902771, -0.05368042, 0.027282715, -0.030639648, 0.052886963, 0.0030021667, -0.009063721, 0.031219482, -0.04095459, 0.0034885406, -0.0010986328, 0.003200531, -0....
http://arxiv.org/abs/2309.14735v2
Legal Question-Answering in the Indian Context: Efficacy, Challenges, and Potential of Modern AI Models
Shubham Kumar Nigam, Shubham Kumar Mishra, Ayush Kumar Mishra, Noel Shallum, Arnab Bhattacharya
cs.CL, cs.AI
Legal QA platforms bear the promise to metamorphose the manner in which legal experts engage with jurisprudential documents. In this exposition, we embark on a comparative exploration of contemporary AI frameworks, gauging their adeptness in catering to the unique demands of the Indian legal milieu, with a keen emphasi...
http://arxiv.org/pdf/2309.14735v2
2023-10-16T04:40:18
2023-09-26T07:56:55
{ "title": [ -0.037200928, -0.015113831, -0.031707764, -0.024032593, -0.0071640015, 0.003929138, -0.020828247, -0.00080156326, 0.012168884, 0.015556335, -0.047668457, 0.009246826, 0.015037537, -0.014122009, -0.033081055, 0.0046043396, -0.02229309, ...
http://arxiv.org/abs/2309.14726v1
PLMM: Personal Large Models on Mobile Devices
Yuanhao Gong
cs.CV, cs.AI, cs.CE, cs.CL, cs.LG
Inspired by Federated Learning, in this paper, we propose personal large models that are distilled from traditional large language models but more adaptive to local users' personal information such as education background and hobbies. We classify the large language models into three levels: the personal level, expert l...
http://arxiv.org/pdf/2309.14726v1
2023-09-26T07:36:20
2023-09-26T07:36:20
{ "title": [ -0.039916992, -0.013084412, -0.025177002, -0.03857422, -0.016525269, 0.009017944, -0.05230713, -0.022277832, -0.012611389, 0.022964478, 0.005947113, 0.0031738281, -0.007698059, -0.04611206, -0.00869751, 0.011917114, 0.012168884, 0.00...
http://arxiv.org/abs/2309.14717v2
QA-LoRA: Quantization-Aware Low-Rank Adaptation of Large Language Models
Yuhui Xu, Lingxi Xie, Xiaotao Gu, Xin Chen, Heng Chang, Hengheng Zhang, Zhengsu Chen, Xiaopeng Zhang, Qi Tian
cs.LG, cs.CL
Recently years have witnessed a rapid development of large language models (LLMs). Despite the strong ability in many language-understanding tasks, the heavy computational burden largely restricts the application of LLMs especially when one needs to deploy them onto edge devices. In this paper, we propose a quantizatio...
http://arxiv.org/pdf/2309.14717v2
2023-10-09T07:39:04
2023-09-26T07:22:23
{ "title": [ -0.047851562, -0.01259613, -0.052246094, -0.025527954, -0.013557434, -0.010734558, -0.03805542, -0.0040397644, 0.0037765503, 0.05859375, 0.004714966, -0.009902954, 0.020721436, -0.020095825, -0.049102783, 0.012207031, 0.0011091232, 0...
http://arxiv.org/abs/2309.14683v1
A Simple Text to Video Model via Transformer
Gang Chen
cs.CV, cs.CL, 68T10, I.2.6
We present a general and simple text to video model based on Transformer. Since both text and video are sequential data, we encode both texts and images into the same hidden space, which are further fed into Transformer to capture the temporal consistency and then decoder to generate either text or images. Considering ...
http://arxiv.org/pdf/2309.14683v1
2023-09-26T05:26:30
2023-09-26T05:26:30
{ "title": [ 0.008041382, -0.011024475, -0.011726379, -0.015205383, -0.047912598, -0.005874634, -0.022994995, -0.011192322, 0.008041382, 0.04937744, -0.061157227, -0.010688782, 0.044921875, -0.023422241, 0.00035643578, -0.02015686, 0.03427124, 0....