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/2310.13247v1
Anomaly Detection of Command Shell Sessions based on DistilBERT: Unsupervised and Supervised Approaches
Zefang Liu, John Buford
cs.CL, cs.CR
Anomaly detection in command shell sessions is a critical aspect of computer security. Recent advances in deep learning and natural language processing, particularly transformer-based models, have shown great promise for addressing complex security challenges. In this paper, we implement a comprehensive approach to det...
http://arxiv.org/pdf/2310.13247v1
2023-10-20T03:04:32
2023-10-20T03:04:32
{ "title": [ 0.01878357, 0.016403198, -0.03729248, -0.062286377, 0.021743774, -0.015945435, -0.02507019, -0.015930176, 0.0049858093, 0.0049819946, -0.01725769, -0.000060737133, -0.008613586, -0.032348633, -0.013114929, -0.016662598, -0.028961182, ...
http://arxiv.org/abs/2310.13243v1
Open-source Large Language Models are Strong Zero-shot Query Likelihood Models for Document Ranking
Shengyao Zhuang, Bing Liu, Bevan Koopman, Guido Zuccon
cs.IR, cs.CL
In the field of information retrieval, Query Likelihood Models (QLMs) rank documents based on the probability of generating the query given the content of a document. Recently, advanced large language models (LLMs) have emerged as effective QLMs, showcasing promising ranking capabilities. This paper focuses on investig...
http://arxiv.org/pdf/2310.13243v1
2023-10-20T02:54:42
2023-10-20T02:54:42
{ "title": [ 0.010902405, 0.003829956, -0.005470276, -0.032958984, -0.031677246, -0.0018043518, -0.055999756, -0.01828003, -0.011810303, 0.045043945, -0.005771637, -0.0057296753, 0.029953003, -0.03152466, -0.007129669, -0.003479004, 0.0028743744, ...
http://arxiv.org/abs/2310.13231v1
Multi-level Contrastive Learning for Script-based Character Understanding
Dawei Li, Hengyuan Zhang, Yanran Li, Shiping Yang
cs.CL, cs.AI, cs.LG
In this work, we tackle the scenario of understanding characters in scripts, which aims to learn the characters' personalities and identities from their utterances. We begin by analyzing several challenges in this scenario, and then propose a multi-level contrastive learning framework to capture characters' global info...
http://arxiv.org/pdf/2310.13231v1
2023-10-20T02:40:52
2023-10-20T02:40:52
{ "title": [ -0.008613586, -0.0146102905, -0.0032863617, -0.054626465, -0.010147095, -0.0059394836, -0.019470215, 0.009101868, 0.03869629, 0.019119263, -0.030517578, 0.01083374, -0.0057029724, -0.010398865, -0.047424316, -0.015449524, -0.0032405853, ...
http://arxiv.org/abs/2310.13228v1
The Less the Merrier? Investigating Language Representation in Multilingual Models
Hellina Hailu Nigatu, Atnafu Lambebo Tonja, Jugal Kalita
cs.CL
Multilingual Language Models offer a way to incorporate multiple languages in one model and utilize cross-language transfer learning to improve performance for different Natural Language Processing (NLP) tasks. Despite progress in multilingual models, not all languages are supported as well, particularly in low-resourc...
http://arxiv.org/pdf/2310.13228v1
2023-10-20T02:26:34
2023-10-20T02:26:34
{ "title": [ -0.024337769, -0.0036296844, -0.033111572, -0.046447754, 0.00434494, 0.0001103282, -0.05783081, -0.0090408325, 0.0055389404, 0.040374756, -0.0033817291, 0.004585266, 0.01737976, -0.011810303, -0.043670654, 0.01499176, -0.021820068, 0...
http://arxiv.org/abs/2310.13227v1
ToolChain*: Efficient Action Space Navigation in Large Language Models with A* Search
Yuchen Zhuang, Xiang Chen, Tong Yu, Saayan Mitra, Victor Bursztyn, Ryan A. Rossi, Somdeb Sarkhel, Chao Zhang
cs.CL, cs.AI, cs.LG
Large language models (LLMs) have demonstrated powerful decision-making and planning capabilities in solving complicated real-world problems. LLM-based autonomous agents can interact with diverse tools (e.g., functional APIs) and generate solution plans that execute a series of API function calls in a step-by-step mann...
http://arxiv.org/pdf/2310.13227v1
2023-10-20T02:24:35
2023-10-20T02:24:35
{ "title": [ -0.033081055, -0.02848816, -0.027267456, -0.07196045, -0.014465332, 0.013458252, -0.055389404, 0.007446289, 0.020141602, 0.0340271, -0.023345947, 0.016464233, 0.016311646, -0.012054443, -0.0340271, 0.014045715, -0.004337311, 0.016555...
http://arxiv.org/abs/2310.13226v1
Enhancing Zero-Shot Crypto Sentiment with Fine-tuned Language Model and Prompt Engineering
Rahman S M Wahidur, Ishmam Tashdeed, Manjit Kaur, Heung-No-Lee
cs.CL
Blockchain technology has revolutionized the financial landscape, with cryptocurrencies gaining widespread adoption for their decentralized and transparent nature. As the sentiment expressed on social media platforms can significantly influence cryptocurrency discussions and market movements, sentiment analysis has eme...
http://arxiv.org/pdf/2310.13226v1
2023-10-20T02:15:51
2023-10-20T02:15:51
{ "title": [ -0.0030937195, -0.019195557, -0.036193848, -0.00051546097, -0.027755737, -0.03010559, -0.0008368492, 0.0008878708, 0.019577026, 0.039215088, -0.0052833557, -0.020706177, 0.025543213, -0.03665161, -0.05770874, 0.007850647, -0.014862061, ...
http://arxiv.org/abs/2310.13213v1
MultiCoNER v2: a Large Multilingual dataset for Fine-grained and Noisy Named Entity Recognition
Besnik Fetahu, Zhiyu Chen, Sudipta Kar, Oleg Rokhlenko, Shervin Malmasi
cs.CL, cs.AI
We present MULTICONER V2, a dataset for fine-grained Named Entity Recognition covering 33 entity classes across 12 languages, in both monolingual and multilingual settings. This dataset aims to tackle the following practical challenges in NER: (i) effective handling of fine-grained classes that include complex entities...
http://arxiv.org/pdf/2310.13213v1
2023-10-20T01:14:46
2023-10-20T01:14:46
{ "title": [ -0.022659302, -0.0020103455, -0.051696777, -0.07543945, 0.028747559, 0.003944397, -0.053100586, 0.0070648193, 0.011146545, 0.031219482, -0.036254883, 0.013320923, 0.011459351, 0.0017061234, -0.029922485, 0.0064582825, -0.0041236877, ...
http://arxiv.org/abs/2310.13206v1
Primacy Effect of ChatGPT
Yiwei Wang, Yujun Cai, Muhao Chen, Yuxuan Liang, Bryan Hooi
cs.CL, cs.AI
Instruction-tuned large language models (LLMs), such as ChatGPT, have led to promising zero-shot performance in discriminative natural language understanding (NLU) tasks. This involves querying the LLM using a prompt containing the question, and the candidate labels to choose from. The question-answering capabilities o...
http://arxiv.org/pdf/2310.13206v1
2023-10-20T00:37:28
2023-10-20T00:37:28
{ "title": [ 0.029083252, -0.01184845, -0.024246216, -0.023284912, -0.040893555, -0.032989502, -0.028244019, -0.024154663, -0.015670776, 0.037139893, -0.01701355, 0.021652222, -0.02319336, -0.028411865, 0.0042762756, 0.0042304993, -0.007873535, 0...
http://arxiv.org/abs/2310.18330v1
Towards Detecting Contextual Real-Time Toxicity for In-Game Chat
Zachary Yang, Nicolas Grenan-Godbout, Reihaneh Rabbany
cs.CL, cs.AI, cs.LG
Real-time toxicity detection in online environments poses a significant challenge, due to the increasing prevalence of social media and gaming platforms. We introduce ToxBuster, a simple and scalable model that reliably detects toxic content in real-time for a line of chat by including chat history and metadata. ToxBus...
http://arxiv.org/pdf/2310.18330v1
2023-10-20T00:29:57
2023-10-20T00:29:57
{ "title": [ 0.03253174, -0.007621765, -0.010612488, -0.06439209, 0.016204834, -0.01133728, -0.0036811829, -0.014640808, 0.007888794, 0.045959473, -0.023132324, 0.014091492, 0.0064582825, -0.010818481, -0.055267334, -0.0053901672, -0.015701294, 0...
http://arxiv.org/abs/2310.13196v1
NameGuess: Column Name Expansion for Tabular Data
Jiani Zhang, Zhengyuan Shen, Balasubramaniam Srinivasan, Shen Wang, Huzefa Rangwala, George Karypis
cs.CL, cs.DB, cs.LG
Recent advances in large language models have revolutionized many sectors, including the database industry. One common challenge when dealing with large volumes of tabular data is the pervasive use of abbreviated column names, which can negatively impact performance on various data search, access, and understanding tas...
http://arxiv.org/pdf/2310.13196v1
2023-10-19T23:11:37
2023-10-19T23:11:37
{ "title": [ -0.043304443, -0.015525818, -0.1003418, -0.031921387, -0.01902771, -0.031982422, -0.04345703, 0.009056091, -0.015113831, 0.051818848, -0.05291748, 0.023590088, 0.039520264, -0.020492554, -0.00944519, -0.020309448, 0.026977539, -0.010...
http://arxiv.org/abs/2310.13191v1
Towards Robust Pruning: An Adaptive Knowledge-Retention Pruning Strategy for Language Models
Jianwei Li, Qi Lei, Wei Cheng, Dongkuan Xu
cs.CL, cs.AI
The pruning objective has recently extended beyond accuracy and sparsity to robustness in language models. Despite this, existing methods struggle to enhance robustness against adversarial attacks when continually increasing model sparsity and require a retraining process. As humans step into the era of large language ...
http://arxiv.org/pdf/2310.13191v1
2023-10-19T23:02:29
2023-10-19T23:02:29
{ "title": [ -0.032928467, -0.01134491, -0.036346436, -0.06317139, -0.010864258, 0.017364502, -0.07659912, 0.022354126, 0.041931152, 0.03994751, -0.01335907, 0.0046081543, 0.037628174, -0.009719849, -0.07196045, 0.019821167, -0.0013570786, 0.0289...
http://arxiv.org/abs/2310.13189v2
Fast and Accurate Factual Inconsistency Detection Over Long Documents
Barrett Martin Lattimer, Patrick Chen, Xinyuan Zhang, Yi Yang
cs.CL, cs.AI
Generative AI models exhibit remarkable potential; however, hallucinations across various tasks present a significant challenge, particularly for longer inputs that current approaches struggle to address effectively. We introduce SCALE (Source Chunking Approach for Large-scale inconsistency Evaluation), a task-agnostic...
http://arxiv.org/pdf/2310.13189v2
2023-10-23T03:51:33
2023-10-19T22:55:39
{ "title": [ -0.018234253, 0.009422302, -0.052246094, -0.05429077, 0.016555786, -0.014373779, -0.02796936, -0.008201599, -0.0005726814, 0.0340271, -0.03378296, -0.0013237, 0.016784668, -0.0038070679, -0.036834717, -0.008163452, -0.02229309, 0.061...
http://arxiv.org/abs/2310.13183v1
Breaking through Deterministic Barriers: Randomized Pruning Mask Generation and Selection
Jianwei Li, Weizhi Gao, Qi Lei, Dongkuan Xu
cs.CV, cs.CL
It is widely acknowledged that large and sparse models have higher accuracy than small and dense models under the same model size constraints. This motivates us to train a large model and then remove its redundant neurons or weights by pruning. Most existing works pruned the networks in a deterministic way, the perform...
http://arxiv.org/pdf/2310.13183v1
2023-10-19T22:32:51
2023-10-19T22:32:51
{ "title": [ -0.04788208, 0.0010786057, -0.042297363, -0.06542969, -0.0054969788, 0.026046753, -0.064697266, -0.0027923584, -0.019546509, 0.010887146, 0.0051574707, 0.01852417, -0.01663208, -0.041015625, -0.014221191, -0.0023479462, 0.013031006, ...
http://arxiv.org/abs/2310.13146v1
CLIFT: Analysing Natural Distribution Shift on Question Answering Models in Clinical Domain
Ankit Pal
cs.CL, cs.AI, cs.CY, cs.DC, cs.LG
This paper introduces a new testbed CLIFT (Clinical Shift) for the clinical domain Question-answering task. The testbed includes 7.5k high-quality question answering samples to provide a diverse and reliable benchmark. We performed a comprehensive experimental study and evaluated several QA deep-learning models under t...
http://arxiv.org/pdf/2310.13146v1
2023-10-19T20:43:11
2023-10-19T20:43:11
{ "title": [ -0.0017204285, -0.003145218, -0.043548584, -0.044708252, 0.015396118, -0.008476257, -0.014183044, 0.0010700226, 0.019577026, 0.02961731, -0.010131836, -0.0031223297, 0.0071372986, -0.013557434, -0.058898926, 0.00073194504, -0.010047913, ...
http://arxiv.org/abs/2310.13132v2
Better to Ask in English: Cross-Lingual Evaluation of Large Language Models for Healthcare Queries
Yiqiao Jin, Mohit Chandra, Gaurav Verma, Yibo Hu, Munmun De Choudhury, Srijan Kumar
cs.CL, cs.AI
Large language models (LLMs) are transforming the ways the general public accesses and consumes information. Their influence is particularly pronounced in pivotal sectors like healthcare, where lay individuals are increasingly appropriating LLMs as conversational agents for everyday queries. While LLMs demonstrate impr...
http://arxiv.org/pdf/2310.13132v2
2023-10-23T17:47:47
2023-10-19T20:02:40
{ "title": [ -0.012176514, -0.011238098, -0.036102295, -0.061157227, 0.021102905, -0.008033752, -0.044677734, -0.012123108, -0.006511688, 0.024856567, -0.015197754, 0.009811401, -0.00045776367, -0.015853882, -0.03552246, -0.0004901886, 0.005203247, ...
http://arxiv.org/abs/2310.13127v1
Auto-Instruct: Automatic Instruction Generation and Ranking for Black-Box Language Models
Zhihan Zhang, Shuohang Wang, Wenhao Yu, Yichong Xu, Dan Iter, Qingkai Zeng, Yang Liu, Chenguang Zhu, Meng Jiang
cs.CL
Large language models (LLMs) can perform a wide range of tasks by following natural language instructions, without the necessity of task-specific fine-tuning. Unfortunately, the performance of LLMs is greatly influenced by the quality of these instructions, and manually writing effective instructions for each task is a...
http://arxiv.org/pdf/2310.13127v1
2023-10-19T19:52:55
2023-10-19T19:52:55
{ "title": [ -0.03314209, -0.02420044, -0.027862549, -0.06427002, -0.011413574, 0.016494751, -0.08013916, -0.006614685, -0.004753113, 0.022644043, -0.032348633, 0.031341553, -0.01424408, 0.0071144104, -0.03552246, 0.0007224083, -0.016418457, 0.02...
http://arxiv.org/abs/2310.13106v1
Unsupervised Candidate Answer Extraction through Differentiable Masker-Reconstructor Model
Zhuoer Wang, Yicheng Wang, Ziwei Zhu, James Caverlee
cs.CL
Question generation is a widely used data augmentation approach with extensive applications, and extracting qualified candidate answers from context passages is a critical step for most question generation systems. However, existing methods for candidate answer extraction are reliant on linguistic rules or annotated da...
http://arxiv.org/pdf/2310.13106v1
2023-10-19T19:07:08
2023-10-19T19:07:08
{ "title": [ -0.011665344, 0.025604248, -0.024765015, -0.025039673, -0.0025196075, 0.0059127808, -0.026123047, -0.042114258, 0.039916992, -0.010910034, -0.008682251, -0.010627747, -0.014533997, -0.028747559, -0.020965576, -0.0007944107, 0.0034103394,...
http://arxiv.org/abs/2310.13099v1
No offence, Bert -- I insult only humans! Multiple addressees sentence-level attack on toxicity detection neural network
Sergey Berezin, Reza Farahbakhsh, Noel Crespi
cs.CL, cs.AI, cs.LG
We introduce a simple yet efficient sentence-level attack on black-box toxicity detector models. By adding several positive words or sentences to the end of a hateful message, we are able to change the prediction of a neural network and pass the toxicity detection system check. This approach is shown to be working on s...
http://arxiv.org/pdf/2310.13099v1
2023-10-19T18:56:50
2023-10-19T18:56:50
{ "title": [ 0.032073975, -0.033721924, -0.0099487305, -0.019729614, -0.040039062, -0.036865234, 0.0051994324, 0.014549255, -0.030517578, 0.027252197, -0.01576233, 0.025787354, 0.027130127, -0.022415161, 0.011734009, -0.031204224, -0.032043457, 0...
http://arxiv.org/abs/2310.13092v1
Do Language Models Learn about Legal Entity Types during Pretraining?
Claire Barale, Michael Rovatsos, Nehal Bhuta
cs.CL
Language Models (LMs) have proven their ability to acquire diverse linguistic knowledge during the pretraining phase, potentially serving as a valuable source of incidental supervision for downstream tasks. However, there has been limited research conducted on the retrieval of domain-specific knowledge, and specificall...
http://arxiv.org/pdf/2310.13092v1
2023-10-19T18:47:21
2023-10-19T18:47:21
{ "title": [ -0.026290894, -0.0028076172, -0.045013428, -0.028945923, -0.032165527, -0.008079529, -0.05545044, 0.012428284, 0.0049476624, 0.0134887695, -0.032226562, 0.010154724, 0.038208008, -0.04272461, 0.05041504, 0.009384155, -0.0104904175, 0...
http://arxiv.org/abs/2310.13080v1
From Multilingual Complexity to Emotional Clarity: Leveraging Commonsense to Unveil Emotions in Code-Mixed Dialogues
Shivani Kumar, Ramaneswaran S, Md Shad Akhtar, Tanmoy Chakraborty
cs.CL, cs.AI
Understanding emotions during conversation is a fundamental aspect of human communication, driving NLP research for Emotion Recognition in Conversation (ERC). While considerable research has focused on discerning emotions of individual speakers in monolingual dialogues, understanding the emotional dynamics in code-mixe...
http://arxiv.org/pdf/2310.13080v1
2023-10-19T18:17:00
2023-10-19T18:17:00
{ "title": [ -0.020767212, -0.03955078, -0.016281128, -0.04168701, -0.021881104, -0.0047302246, -0.044006348, 0.012268066, 0.034179688, 0.034057617, 0.005882263, 0.0009217262, 0.011375427, -0.041412354, -0.058380127, 0.0075035095, -0.0047836304, ...
http://arxiv.org/abs/2310.13068v1
GARI: Graph Attention for Relative Isomorphism of Arabic Word Embeddings
Muhammad Asif Ali, Maha Alshmrani, Jianbin Qin, Yan Hu, Di Wang
cs.CL
Bilingual Lexical Induction (BLI) is a core challenge in NLP, it relies on the relative isomorphism of individual embedding spaces. Existing attempts aimed at controlling the relative isomorphism of different embedding spaces fail to incorporate the impact of semantically related words in the model training objective. ...
http://arxiv.org/pdf/2310.13068v1
2023-10-19T18:08:22
2023-10-19T18:08:22
{ "title": [ -0.016647339, -0.011444092, -0.015686035, -0.059295654, -0.015701294, 0.005378723, -0.051635742, 0.0018835068, -0.008728027, 0.03866577, -0.010276794, 0.011672974, 0.0006685257, -0.026031494, -0.028060913, -0.01008606, -0.007873535, ...
http://arxiv.org/abs/2310.12973v1
Frozen Transformers in Language Models Are Effective Visual Encoder Layers
Ziqi Pang, Ziyang Xie, Yunze Man, Yu-Xiong Wang
cs.CV, cs.AI, cs.CL, cs.LG
This paper reveals that large language models (LLMs), despite being trained solely on textual data, are surprisingly strong encoders for purely visual tasks in the absence of language. Even more intriguingly, this can be achieved by a simple yet previously overlooked strategy -- employing a frozen transformer block fro...
http://arxiv.org/pdf/2310.12973v1
2023-10-19T17:59:05
2023-10-19T17:59:05
{ "title": [ 0.0335083, 0.00008517504, -0.0064430237, -0.031951904, -0.037231445, -0.022079468, -0.016464233, -0.024902344, 0.013023376, 0.04626465, -0.028533936, 0.00021207333, 0.050048828, -0.033111572, -0.007221222, 0.01889038, 0.0078125, 0.05...
http://arxiv.org/abs/2310.12971v1
CLAIR: Evaluating Image Captions with Large Language Models
David Chan, Suzanne Petryk, Joseph E. Gonzalez, Trevor Darrell, John Canny
cs.CV, cs.AI, cs.CL
The evaluation of machine-generated image captions poses an interesting yet persistent challenge. Effective evaluation measures must consider numerous dimensions of similarity, including semantic relevance, visual structure, object interactions, caption diversity, and specificity. Existing highly-engineered measures at...
http://arxiv.org/pdf/2310.12971v1
2023-10-19T17:59:01
2023-10-19T17:59:01
{ "title": [ -0.0099105835, -0.0035495758, -0.005016327, -0.028457642, -0.009132385, -0.011192322, -0.028518677, -0.032440186, 0.029922485, 0.03805542, -0.034606934, 0.015686035, 0.00045990944, 0.0029506683, -0.05831909, 0.01197052, -0.0124435425, ...
http://arxiv.org/abs/2310.12963v2
AutoMix: Automatically Mixing Language Models
Aman Madaan, Pranjal Aggarwal, Ankit Anand, Srividya Pranavi Potharaju, Swaroop Mishra, Pei Zhou, Aditya Gupta, Dheeraj Rajagopal, Karthik Kappaganthu, Yiming Yang, Shyam Upadhyay, Mausam, Manaal Faruqui
cs.CL, cs.AI
Large language models (LLMs) are now available in various sizes and configurations from cloud API providers. While this diversity offers a broad spectrum of choices, effectively leveraging the options to optimize computational cost and performance remains challenging. In this work, we present AutoMix, an approach that ...
http://arxiv.org/pdf/2310.12963v2
2023-11-15T18:23:40
2023-10-19T17:57:39
{ "title": [ -0.0435791, -0.0068244934, -0.050811768, -0.06100464, 0.014503479, 0.013725281, -0.067871094, -0.0039405823, -0.010437012, 0.045654297, -0.005279541, 0.0051193237, 0.02029419, -0.018035889, -0.03805542, 0.013809204, -0.015197754, 0.0...
http://arxiv.org/abs/2310.12962v1
An Emulator for Fine-Tuning Large Language Models using Small Language Models
Eric Mitchell, Rafael Rafailov, Archit Sharma, Chelsea Finn, Christopher D. Manning
cs.CL, cs.AI, cs.LG
Widely used language models (LMs) are typically built by scaling up a two-stage training pipeline: a pre-training stage that uses a very large, diverse dataset of text and a fine-tuning (sometimes, 'alignment') stage that uses targeted examples or other specifications of desired behaviors. While it has been hypothesize...
http://arxiv.org/pdf/2310.12962v1
2023-10-19T17:57:16
2023-10-19T17:57:16
{ "title": [ -0.045196533, -0.023422241, -0.03894043, -0.06768799, -0.007843018, 0.010261536, -0.066101074, 0.020446777, -0.0030574799, 0.039733887, -0.013191223, -0.017532349, 0.020080566, -0.021087646, -0.057037354, 0.011566162, -0.0068626404, ...
http://arxiv.org/abs/2310.12960v1
SEGO: Sequential Subgoal Optimization for Mathematical Problem-Solving
Xueliang Zhao, Xinting Huang, Wei Bi, Lingpeng Kong
cs.CL
Large Language Models (LLMs) have driven substantial progress in artificial intelligence in recent years, exhibiting impressive capabilities across a wide range of tasks, including mathematical problem-solving. Inspired by the success of subgoal-based methods, we propose a novel framework called \textbf{SE}quential sub...
http://arxiv.org/pdf/2310.12960v1
2023-10-19T17:56:40
2023-10-19T17:56:40
{ "title": [ -0.032928467, -0.0020217896, -0.013427734, -0.039276123, 0.0088272095, 0.012039185, -0.020401001, -0.04296875, -0.0046043396, 0.024230957, 0.005176544, 0.02949524, -0.01449585, -0.020874023, -0.02456665, 0.00674057, 0.008636475, -0.0...
http://arxiv.org/abs/2310.12942v2
On the Representational Capacity of Recurrent Neural Language Models
Franz Nowak, Anej Svete, Li Du, Ryan Cotterell
cs.CL, cs.LG
This work investigates the computational expressivity of language models (LMs) based on recurrent neural networks (RNNs). Siegelmann and Sontag (1992) famously showed that RNNs with rational weights and hidden states and unbounded computation time are Turing complete. However, LMs define weightings over strings in addi...
http://arxiv.org/pdf/2310.12942v2
2023-10-23T10:24:26
2023-10-19T17:39:47
{ "title": [ -0.0013055801, 0.0013132095, -0.022781372, -0.06161499, 0.010368347, 0.0036773682, -0.04119873, 0.0022621155, 0.016021729, 0.04345703, -0.0069274902, -0.006465912, -0.0011892319, -0.014259338, -0.05908203, 0.025466919, -0.0075035095, ...
http://arxiv.org/abs/2310.12936v2
A Predictive Factor Analysis of Social Biases and Task-Performance in Pretrained Masked Language Models
Yi Zhou, Jose Camacho-Collados, Danushka Bollegala
cs.CL
Various types of social biases have been reported with pretrained Masked Language Models (MLMs) in prior work. However, multiple underlying factors are associated with an MLM such as its model size, size of the training data, training objectives, the domain from which pretraining data is sampled, tokenization, and lang...
http://arxiv.org/pdf/2310.12936v2
2023-10-22T14:19:01
2023-10-19T17:33:33
{ "title": [ -0.02708435, 0.0018415451, -0.06173706, -0.03515625, -0.00881958, -0.031463623, -0.050842285, -0.0043411255, -0.010482788, 0.023162842, 0.005519867, -0.00730896, 0.033172607, -0.04586792, -0.016601562, -0.0103302, -0.026535034, 0.040...
http://arxiv.org/abs/2310.12902v1
Experimental Narratives: A Comparison of Human Crowdsourced Storytelling and AI Storytelling
Nina Begus
cs.CL, cs.AI
The paper proposes a framework that combines behavioral and computational experiments employing fictional prompts as a novel tool for investigating cultural artifacts and social biases in storytelling both by humans and generative AI. The study analyzes 250 stories authored by crowdworkers in June 2019 and 80 stories g...
http://arxiv.org/pdf/2310.12902v1
2023-10-19T16:54:38
2023-10-19T16:54:38
{ "title": [ -0.045135498, -0.016082764, -0.04714966, -0.0110321045, -0.021499634, -0.015457153, 0.003604889, -0.0014467239, -0.02859497, 0.039367676, -0.014762878, 0.019088745, -0.0023326874, -0.009048462, -0.023529053, -0.008621216, -0.022766113, ...
http://arxiv.org/abs/2310.12892v1
A Systematic Study of Performance Disparities in Multilingual Task-Oriented Dialogue Systems
Songbo Hu, Han Zhou, Moy Yuan, Milan Gritta, Guchun Zhang, Ignacio Iacobacci, Anna Korhonen, Ivan Vulić
cs.CL
Achieving robust language technologies that can perform well across the world's many languages is a central goal of multilingual NLP. In this work, we take stock of and empirically analyse task performance disparities that exist between multilingual task-oriented dialogue (ToD) systems. We first define new quantitative...
http://arxiv.org/pdf/2310.12892v1
2023-10-19T16:41:44
2023-10-19T16:41:44
{ "title": [ -0.022277832, -0.0127334595, -0.047576904, -0.050079346, 0.015411377, -0.017547607, -0.036987305, 0.0236969, 0.0010585785, 0.041168213, -0.003736496, -0.009986877, 0.004383087, -0.017974854, -0.06100464, -0.009216309, -0.015838623, 0...
http://arxiv.org/abs/2310.12874v2
StoryAnalogy: Deriving Story-level Analogies from Large Language Models to Unlock Analogical Understanding
Cheng Jiayang, Lin Qiu, Tsz Ho Chan, Tianqing Fang, Weiqi Wang, Chunkit Chan, Dongyu Ru, Qipeng Guo, Hongming Zhang, Yangqiu Song, Yue Zhang, Zheng Zhang
cs.CL
Analogy-making between narratives is crucial for human reasoning. In this paper, we evaluate the ability to identify and generate analogies by constructing a first-of-its-kind large-scale story-level analogy corpus, \textsc{StoryAnalogy}, which contains 24K story pairs from diverse domains with human annotations on two...
http://arxiv.org/pdf/2310.12874v2
2023-10-23T11:58:24
2023-10-19T16:29:23
{ "title": [ -0.017349243, -0.027755737, -0.00920105, -0.046813965, -0.0206604, -0.0016403198, -0.0289917, -0.020629883, 0.0049057007, 0.0256958, -0.009635925, 0.0067443848, 0.014831543, -0.022140503, -0.02659607, 0.008125305, -0.008636475, 0.029...
http://arxiv.org/abs/2310.12864v1
The Locality and Symmetry of Positional Encodings
Lihu Chen, Gaël Varoquaux, Fabian M. Suchanek
cs.CL
Positional Encodings (PEs) are used to inject word-order information into transformer-based language models. While they can significantly enhance the quality of sentence representations, their specific contribution to language models is not fully understood, especially given recent findings that various positional enco...
http://arxiv.org/pdf/2310.12864v1
2023-10-19T16:15:15
2023-10-19T16:15:15
{ "title": [ -0.011299133, 0.00022029877, -0.015281677, -0.05545044, 0.027145386, 0.012878418, -0.0074157715, -0.027938843, 0.037078857, 0.057739258, -0.0109939575, -0.0051956177, 0.011779785, -0.004547119, -0.060821533, 0.036865234, 0.0031337738, ...
http://arxiv.org/abs/2310.12860v2
Probing LLMs for hate speech detection: strengths and vulnerabilities
Sarthak Roy, Ashish Harshavardhan, Animesh Mukherjee, Punyajoy Saha
cs.CL, cs.CY
Recently efforts have been made by social media platforms as well as researchers to detect hateful or toxic language using large language models. However, none of these works aim to use explanation, additional context and victim community information in the detection process. We utilise different prompt variation, inpu...
http://arxiv.org/pdf/2310.12860v2
2023-10-28T05:07:31
2023-10-19T16:11:02
{ "title": [ 0.0015115738, 0.006767273, -0.027404785, -0.026428223, 0.0085372925, -0.021209717, -0.00009471178, -0.02053833, -0.015625, 0.021652222, -0.02696228, 0.028900146, -0.011772156, -0.015487671, -0.04147339, -0.01158905, -0.023117065, 0.0...
http://arxiv.org/abs/2310.12851v1
EmoDiarize: Speaker Diarization and Emotion Identification from Speech Signals using Convolutional Neural Networks
Hanan Hamza, Fiza Gafoor, Fathima Sithara, Gayathri Anil, V. S. Anoop
cs.SD, cs.CL, eess.AS
In the era of advanced artificial intelligence and human-computer interaction, identifying emotions in spoken language is paramount. This research explores the integration of deep learning techniques in speech emotion recognition, offering a comprehensive solution to the challenges associated with speaker diarization a...
http://arxiv.org/pdf/2310.12851v1
2023-10-19T16:02:53
2023-10-19T16:02:53
{ "title": [ 0.007850647, -0.0035495758, -0.042785645, -0.045776367, 0.0030784607, -0.023712158, -0.017822266, -0.006427765, -0.009880066, 0.052978516, -0.03225708, -9.536743e-7, 0.04434204, -0.02041626, -0.026641846, -0.007507324, 0.010414124, 0...
http://arxiv.org/abs/2310.12836v1
Knowledge-Augmented Language Model Verification
Jinheon Baek, Soyeong Jeong, Minki Kang, Jong C. Park, Sung Ju Hwang
cs.CL, cs.LG
Recent Language Models (LMs) have shown impressive capabilities in generating texts with the knowledge internalized in parameters. Yet, LMs often generate the factually incorrect responses to the given queries, since their knowledge may be inaccurate, incomplete, and outdated. To address this problem, previous works pr...
http://arxiv.org/pdf/2310.12836v1
2023-10-19T15:40:00
2023-10-19T15:40:00
{ "title": [ -0.038146973, -0.009338379, -0.048675537, -0.10437012, 0.0029964447, 0.019302368, -0.1015625, 0.027313232, -0.0006995201, 0.03540039, -0.012504578, -0.0007829666, 0.01876831, -0.015670776, -0.037872314, 0.011428833, -0.02281189, 0.04...
http://arxiv.org/abs/2310.12823v2
AgentTuning: Enabling Generalized Agent Abilities for LLMs
Aohan Zeng, Mingdao Liu, Rui Lu, Bowen Wang, Xiao Liu, Yuxiao Dong, Jie Tang
cs.CL, cs.AI, cs.LG
Open large language models (LLMs) with great performance in various tasks have significantly advanced the development of LLMs. However, they are far inferior to commercial models such as ChatGPT and GPT-4 when acting as agents to tackle complex tasks in the real world. These agent tasks employ LLMs as the central contr...
http://arxiv.org/pdf/2310.12823v2
2023-10-22T16:19:16
2023-10-19T15:19:53
{ "title": [ -0.02784729, -0.008110046, -0.036712646, -0.050994873, -0.007827759, -0.002254486, -0.026443481, 0.0005335808, -0.0023384094, 0.031280518, -0.046783447, 0.014320374, 0.0104599, -0.010108948, -0.029586792, 0.0033168793, -0.013847351, ...
http://arxiv.org/abs/2310.12821v3
GestureGPT: Zero-shot Interactive Gesture Understanding and Grounding with Large Language Model Agents
Xin Zeng, Xiaoyu Wang, Tengxiang Zhang, Chun Yu, Shengdong Zhao, Yiqiang Chen
cs.CL, cs.HC
Current gesture recognition systems primarily focus on identifying gestures within a predefined set, leaving a gap in connecting these gestures to interactive GUI elements or system functions (e.g., linking a 'thumb-up' gesture to a 'like' button). We introduce GestureGPT, a novel zero-shot gesture understanding and gr...
http://arxiv.org/pdf/2310.12821v3
2023-10-30T03:04:07
2023-10-19T15:17:34
{ "title": [ -0.009437561, -0.012298584, -0.020767212, -0.05831909, -0.005973816, 0.0013532639, -0.036895752, -0.007587433, 0.023239136, 0.018096924, -0.030441284, -0.0029087067, 0.0020256042, -0.015975952, -0.03842163, -0.0007953644, -0.005832672, ...
http://arxiv.org/abs/2310.12818v1
Boosting Inference Efficiency: Unleashing the Power of Parameter-Shared Pre-trained Language Models
Weize Chen, Xiaoyue Xu, Xu Han, Yankai Lin, Ruobing Xie, Zhiyuan Liu, Maosong Sun, Jie Zhou
cs.CL, cs.AI, cs.LG
Parameter-shared pre-trained language models (PLMs) have emerged as a successful approach in resource-constrained environments, enabling substantial reductions in model storage and memory costs without significant performance compromise. However, it is important to note that parameter sharing does not alleviate computa...
http://arxiv.org/pdf/2310.12818v1
2023-10-19T15:13:58
2023-10-19T15:13:58
{ "title": [ -0.011772156, -0.017700195, -0.030792236, -0.09051514, 0.0014848709, 0.004863739, -0.07537842, -0.0051116943, -0.0053520203, 0.020690918, -0.00774765, 0.013259888, 0.003786087, -0.022338867, -0.025390625, 0.019805908, -0.0049705505, ...
http://arxiv.org/abs/2310.12815v1
Prompt Injection Attacks and Defenses in LLM-Integrated Applications
Yupei Liu, Yuqi Jia, Runpeng Geng, Jinyuan Jia, Neil Zhenqiang Gong
cs.CR, cs.AI, cs.CL, cs.LG
Large Language Models (LLMs) are increasingly deployed as the backend for a variety of real-world applications called LLM-Integrated Applications. Multiple recent works showed that LLM-Integrated Applications are vulnerable to prompt injection attacks, in which an attacker injects malicious instruction/data into the in...
http://arxiv.org/pdf/2310.12815v1
2023-10-19T15:12:09
2023-10-19T15:12:09
{ "title": [ -0.029647827, 0.006881714, -0.036254883, -0.0602417, 0.0021209717, 0.008468628, -0.051116943, -0.00028061867, -0.06048584, 0.0045318604, 0.00806427, 0.018966675, -0.059387207, -0.048736572, -0.0003876686, -0.017837524, -0.013885498, ...
http://arxiv.org/abs/2310.12808v1
Model Merging by Uncertainty-Based Gradient Matching
Nico Daheim, Thomas Möllenhoff, Edoardo Maria Ponti, Iryna Gurevych, Mohammad Emtiyaz Khan
cs.LG, cs.AI, cs.CL
Models trained on different datasets can be merged by a weighted-averaging of their parameters, but why does it work and when can it fail? Here, we connect the inaccuracy of weighted-averaging to mismatches in the gradients and propose a new uncertainty-based scheme to improve the performance by reducing the mismatch. ...
http://arxiv.org/pdf/2310.12808v1
2023-10-19T15:02:45
2023-10-19T15:02:45
{ "title": [ -0.031799316, 0.024734497, -0.05090332, -0.05999756, 0.017868042, 0.016281128, -0.072021484, -0.038238525, -0.006614685, 0.018127441, -0.02784729, 0.0004081726, -0.005203247, -0.023895264, -0.011947632, -0.010063171, 0.0084991455, 0....
http://arxiv.org/abs/2310.12803v1
Causal-structure Driven Augmentations for Text OOD Generalization
Amir Feder, Yoav Wald, Claudia Shi, Suchi Saria, David Blei
cs.LG, cs.CL
The reliance of text classifiers on spurious correlations can lead to poor generalization at deployment, raising concerns about their use in safety-critical domains such as healthcare. In this work, we propose to use counterfactual data augmentation, guided by knowledge of the causal structure of the data, to simulate ...
http://arxiv.org/pdf/2310.12803v1
2023-10-19T14:59:25
2023-10-19T14:59:25
{ "title": [ 0.017166138, -0.0049209595, 0.0037841797, -0.09375, -0.006816864, 0.030380249, -0.045806885, -0.0033626556, -0.008392334, 0.023391724, -0.03967285, 0.051361084, 0.0069274902, 0.0022964478, -0.0028514862, 0.0029830933, -0.018630981, 0...
http://arxiv.org/abs/2310.12798v2
MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter
Zhiyuan Liu, Sihang Li, Yanchen Luo, Hao Fei, Yixin Cao, Kenji Kawaguchi, Xiang Wang, Tat-Seng Chua
cs.CL, cs.MM
Language Models (LMs) have demonstrated impressive molecule understanding ability on various 1D text-related tasks. However, they inherently lack 2D graph perception - a critical ability of human professionals in comprehending molecules' topological structures. To bridge this gap, we propose MolCA: Molecular Graph-Lang...
http://arxiv.org/pdf/2310.12798v2
2023-11-07T08:56:36
2023-10-19T14:52:58
{ "title": [ -0.08605957, -0.030197144, -0.07080078, -0.054351807, -0.043426514, 0.027374268, -0.10913086, 0.029937744, 0.003967285, 0.017700195, -0.0009469986, 0.0011529922, 0.018051147, -0.038391113, -0.0021972656, 0.015686035, -0.00440979, 0.0...
http://arxiv.org/abs/2310.12794v1
Are Structural Concepts Universal in Transformer Language Models? Towards Interpretable Cross-Lingual Generalization
Ningyu Xu, Qi Zhang, Jingting Ye, Menghan Zhang, Xuanjing Huang
cs.CL
Large language models (LLMs) have exhibited considerable cross-lingual generalization abilities, whereby they implicitly transfer knowledge across languages. However, the transfer is not equally successful for all languages, especially for low-resource ones, which poses an ongoing challenge. It is unclear whether we ha...
http://arxiv.org/pdf/2310.12794v1
2023-10-19T14:50:51
2023-10-19T14:50:51
{ "title": [ -0.0096588135, 0.0024852753, -0.0119018555, -0.07128906, -0.00014555454, 0.011627197, -0.078125, -0.013870239, 0.0104522705, 0.04006958, -0.018737793, 0.016159058, 0.030471802, -0.0027484894, -0.04168701, 0.014892578, -0.023925781, 0...
http://arxiv.org/abs/2310.12778v1
Label-Aware Automatic Verbalizer for Few-Shot Text Classification
Thanakorn Thaminkaew, Piyawat Lertvittayakumjorn, Peerapon Vateekul
cs.CL, cs.LG
Prompt-based learning has shown its effectiveness in few-shot text classification. One important factor in its success is a verbalizer, which translates output from a language model into a predicted class. Notably, the simplest and widely acknowledged verbalizer employs manual labels to represent the classes. However, ...
http://arxiv.org/pdf/2310.12778v1
2023-10-19T14:30:07
2023-10-19T14:30:07
{ "title": [ -0.004146576, 0.004386902, -0.024139404, -0.049621582, -0.004753113, 0.0073013306, -0.022827148, -0.0027484894, 0.0057144165, 0.02645874, -0.033111572, 0.020095825, -0.019195557, -0.0017185211, -0.05126953, -0.022033691, -0.012367249, ...
http://arxiv.org/abs/2310.12774v1
Survival of the Most Influential Prompts: Efficient Black-Box Prompt Search via Clustering and Pruning
Han Zhou, Xingchen Wan, Ivan Vulić, Anna Korhonen
cs.CL, cs.AI, cs.LG
Prompt-based learning has been an effective paradigm for large pretrained language models (LLM), enabling few-shot or even zero-shot learning. Black-box prompt search has received growing interest recently for its distinctive properties of gradient-free optimization, proven particularly useful and powerful for model-as...
http://arxiv.org/pdf/2310.12774v1
2023-10-19T14:25:06
2023-10-19T14:25:06
{ "title": [ 0.014030457, -0.0151901245, -0.010276794, -0.06185913, 0.008659363, -0.003982544, -0.022521973, -0.025390625, 0.009025574, 0.017150879, -0.027359009, 0.009017944, -0.0063171387, -0.0053138733, -0.048553467, -0.008377075, -0.0118637085, ...
http://arxiv.org/abs/2310.12766v1
Transformer-based Entity Legal Form Classification
Alexander Arimond, Mauro Molteni, Dominik Jany, Zornitsa Manolova, Damian Borth, Andreas G. F. Hoepner
cs.CL, cs.LG
We propose the application of Transformer-based language models for classifying entity legal forms from raw legal entity names. Specifically, we employ various BERT variants and compare their performance against multiple traditional baselines. Our evaluation encompasses a substantial subset of freely available Legal En...
http://arxiv.org/pdf/2310.12766v1
2023-10-19T14:11:43
2023-10-19T14:11:43
{ "title": [ 0.009140015, 0.03289795, -0.078063965, -0.03945923, 0.0069122314, -0.04522705, -0.026016235, 0.0041999817, -0.008308411, 0.037475586, -0.029663086, -0.010910034, 0.06384277, -0.048339844, 0.026794434, 0.009971619, -0.018875122, 0.072...
http://arxiv.org/abs/2310.12751v1
Character-level Chinese Backpack Language Models
Hao Sun, John Hewitt
cs.CL
The Backpack is a Transformer alternative shown to improve interpretability in English language modeling by decomposing predictions into a weighted sum of token sense components. However, Backpacks' reliance on token-defined meaning raises questions as to their potential for languages other than English, a language for...
http://arxiv.org/pdf/2310.12751v1
2023-10-19T13:54:57
2023-10-19T13:54:57
{ "title": [ -0.03353882, -0.016586304, -0.018432617, 0.00004529953, -0.0680542, -0.006412506, -0.060943604, -0.0034332275, -0.026794434, 0.037384033, -0.020095825, -0.026321411, 0.015396118, -0.044158936, 0.008659363, -0.023666382, 0.012641907, ...
http://arxiv.org/abs/2310.12727v1
Representing and Computing Uncertainty in Phonological Reconstruction
Johann-Mattis List, Nathan W. Hill, Robert Forkel, Frederic Blum
cs.CL
Despite the inherently fuzzy nature of reconstructions in historical linguistics, most scholars do not represent their uncertainty when proposing proto-forms. With the increasing success of recently proposed approaches to automating certain aspects of the traditional comparative method, the formal representation of pro...
http://arxiv.org/pdf/2310.12727v1
2023-10-19T13:27:42
2023-10-19T13:27:42
{ "title": [ -0.03756714, -0.0053100586, -0.024551392, -0.050842285, -0.023986816, 0.0051193237, -0.04977417, 0.01197052, -0.0058631897, 0.030395508, -0.024810791, 0.013916016, 0.0074920654, -0.024139404, -0.02722168, -0.0047569275, -0.008766174, ...
http://arxiv.org/abs/2310.13032v3
Quality-Diversity through AI Feedback
Herbie Bradley, Andrew Dai, Hannah Teufel, Jenny Zhang, Koen Oostermeijer, Marco Bellagente, Jeff Clune, Kenneth Stanley, Grégory Schott, Joel Lehman
cs.CL, cs.AI, cs.LG, cs.NE
In many text-generation problems, users may prefer not only a single response, but a diverse range of high-quality outputs from which to choose. Quality-diversity (QD) search algorithms aim at such outcomes, by continually improving and diversifying a population of candidates. However, the applicability of QD to qualit...
http://arxiv.org/pdf/2310.13032v3
2023-11-08T17:06:30
2023-10-19T12:13:58
{ "title": [ -0.020050049, 0.02230835, -0.040985107, -0.021911621, -0.01171875, -0.024612427, -0.023605347, -0.039215088, -0.008255005, 0.04031372, -0.031311035, -0.004180908, -0.003288269, -0.03338623, -0.021972656, -0.0048942566, 0.009407043, 0...
http://arxiv.org/abs/2310.12664v1
Is ChatGPT a Financial Expert? Evaluating Language Models on Financial Natural Language Processing
Yue Guo, Zian Xu, Yi Yang
cs.CL
The emergence of Large Language Models (LLMs), such as ChatGPT, has revolutionized general natural language preprocessing (NLP) tasks. However, their expertise in the financial domain lacks a comprehensive evaluation. To assess the ability of LLMs to solve financial NLP tasks, we present FinLMEval, a framework for Fina...
http://arxiv.org/pdf/2310.12664v1
2023-10-19T11:43:15
2023-10-19T11:43:15
{ "title": [ -0.027053833, -0.00058841705, -0.033813477, -0.0074386597, -0.021331787, -0.010063171, -0.020507812, -0.018066406, -0.031951904, 0.04522705, -0.01689148, 0.0005531311, 0.017288208, -0.013092041, -0.039916992, -0.0033550262, -0.023208618,...
http://arxiv.org/abs/2310.13031v1
A Use Case: Reformulating Query Rewriting as a Statistical Machine Translation Problem
Abdullah Can Algan, Emre Yürekli, Aykut Çayır
cs.CL, cs.AI, cs.LG
One of the most important challenges for modern search engines is to retrieve relevant web content based on user queries. In order to achieve this challenge, search engines have a module to rewrite user queries. That is why modern web search engines utilize some statistical and neural models used in the natural languag...
http://arxiv.org/pdf/2310.13031v1
2023-10-19T11:37:14
2023-10-19T11:37:14
{ "title": [ -0.024719238, -0.032196045, -0.017684937, -0.06451416, -0.009376526, 0.017623901, -0.061950684, 0.01133728, 0.0020484924, 0.0385437, -0.010971069, 0.014282227, 0.014892578, -0.002828598, -0.03933716, 0.006187439, 0.0046920776, -0.002...
http://arxiv.org/abs/2310.12648v2
Towards Real-World Streaming Speech Translation for Code-Switched Speech
Belen Alastruey, Matthias Sperber, Christian Gollan, Dominic Telaar, Tim Ng, Aashish Agarwal
cs.CL
Code-switching (CS), i.e. mixing different languages in a single sentence, is a common phenomenon in communication and can be challenging in many Natural Language Processing (NLP) settings. Previous studies on CS speech have shown promising results for end-to-end speech translation (ST), but have been limited to offlin...
http://arxiv.org/pdf/2310.12648v2
2023-10-23T11:47:53
2023-10-19T11:15:02
{ "title": [ -0.021575928, -0.016036987, -0.016647339, -0.059661865, -0.0036239624, 0.0040245056, -0.041137695, 0.024017334, -0.02130127, 0.058135986, -0.0023345947, -0.0031433105, 0.029754639, -0.018737793, -0.046813965, 0.011352539, -0.006137848, ...
http://arxiv.org/abs/2310.12640v1
Non-Autoregressive Sentence Ordering
Yi Bin, Wenhao Shi, Bin Ji, Jipeng Zhang, Yujuan Ding, Yang Yang
cs.CL
Existing sentence ordering approaches generally employ encoder-decoder frameworks with the pointer net to recover the coherence by recurrently predicting each sentence step-by-step. Such an autoregressive manner only leverages unilateral dependencies during decoding and cannot fully explore the semantic dependency betw...
http://arxiv.org/pdf/2310.12640v1
2023-10-19T10:57:51
2023-10-19T10:57:51
{ "title": [ 0.007133484, 0.014915466, -0.04953003, -0.066345215, -0.004688263, -0.027679443, -0.05105591, -0.011520386, -0.0058174133, 0.051940918, -0.020599365, 0.0026359558, 0.013725281, -0.010810852, -0.036987305, 0.0062179565, -0.021987915, ...
http://arxiv.org/abs/2310.12620v1
Predict the Future from the Past? On the Temporal Data Distribution Shift in Financial Sentiment Classifications
Yue Guo, Chenxi Hu, Yi Yang
cs.CL
Temporal data distribution shift is prevalent in the financial text. How can a financial sentiment analysis system be trained in a volatile market environment that can accurately infer sentiment and be robust to temporal data distribution shifts? In this paper, we conduct an empirical study on the financial sentiment a...
http://arxiv.org/pdf/2310.12620v1
2023-10-19T09:59:52
2023-10-19T09:59:52
{ "title": [ 0.014533997, 0.012397766, -0.059783936, -0.05105591, 0.033935547, -0.03692627, 0.009552002, -0.013191223, 0.01134491, 0.032043457, -0.043273926, -0.01586914, -0.005756378, -0.00045847893, -0.076416016, -0.03366089, -0.00258255, 0.042...
http://arxiv.org/abs/2310.12611v1
Identifying and Adapting Transformer-Components Responsible for Gender Bias in an English Language Model
Abhijith Chintam, Rahel Beloch, Willem Zuidema, Michael Hanna, Oskar van der Wal
cs.CL, cs.AI
Language models (LMs) exhibit and amplify many types of undesirable biases learned from the training data, including gender bias. However, we lack tools for effectively and efficiently changing this behavior without hurting general language modeling performance. In this paper, we study three methods for identifying cau...
http://arxiv.org/pdf/2310.12611v1
2023-10-19T09:39:21
2023-10-19T09:39:21
{ "title": [ -0.021484375, 0.0070114136, -0.042755127, -0.008430481, -0.007411957, -0.029876709, -0.015838623, -0.015602112, -0.0154418945, 0.040496826, -0.016418457, -0.014328003, 0.03933716, -0.020721436, -0.038604736, -0.008773804, -0.037872314, ...
http://arxiv.org/abs/2310.12585v1
Time-Aware Representation Learning for Time-Sensitive Question Answering
Jungbin Son, Alice Oh
cs.CL, cs.AI, cs.LG
Time is one of the crucial factors in real-world question answering (QA) problems. However, language models have difficulty understanding the relationships between time specifiers, such as 'after' and 'before', and numbers, since existing QA datasets do not include sufficient time expressions. To address this issue, we...
http://arxiv.org/pdf/2310.12585v1
2023-10-19T08:48:45
2023-10-19T08:48:45
{ "title": [ 0.02142334, 0.007068634, -0.038848877, -0.06994629, 0.012245178, -0.02935791, -0.017211914, 0.019897461, 0.021835327, 0.02104187, -0.023406982, -0.010986328, -0.00995636, -0.016952515, -0.06921387, -0.014945984, -0.00063943863, 0.018...
http://arxiv.org/abs/2310.12580v2
Pretraining Language Models with Text-Attributed Heterogeneous Graphs
Tao Zou, Le Yu, Yifei Huang, Leilei Sun, Bowen Du
cs.CL, cs.AI
In many real-world scenarios (e.g., academic networks, social platforms), different types of entities are not only associated with texts but also connected by various relationships, which can be abstracted as Text-Attributed Heterogeneous Graphs (TAHGs). Current pretraining tasks for Language Models (LMs) primarily foc...
http://arxiv.org/pdf/2310.12580v2
2023-10-23T01:46:04
2023-10-19T08:41:21
{ "title": [ -0.013153076, -0.004184723, -0.04171753, -0.059814453, -0.01737976, 0.004360199, -0.064575195, 0.0057296753, -0.013008118, 0.026672363, -0.0062294006, 0.024551392, -0.010009766, -0.02458191, -0.015319824, -0.0017652512, -0.017730713, ...
http://arxiv.org/abs/2310.12575v1
Multilingual estimation of political-party positioning: From label aggregation to long-input Transformers
Dmitry Nikolaev, Tanise Ceron, Sebastian Padó
cs.CL
Scaling analysis is a technique in computational political science that assigns a political actor (e.g. politician or party) a score on a predefined scale based on a (typically long) body of text (e.g. a parliamentary speech or an election manifesto). For example, political scientists have often used the left--right sc...
http://arxiv.org/pdf/2310.12575v1
2023-10-19T08:34:48
2023-10-19T08:34:48
{ "title": [ 0.010055542, 0.003921509, -0.008453369, -0.038116455, -0.0019159317, -0.011131287, -0.023788452, -0.038604736, -0.008773804, 0.030960083, -0.009124756, 0.009521484, -0.016113281, -0.027374268, -0.021957397, -0.0016860962, -0.0009713173, ...
http://arxiv.org/abs/2310.12558v1
Large Language Models Help Humans Verify Truthfulness -- Except When They Are Convincingly Wrong
Chenglei Si, Navita Goyal, Sherry Tongshuang Wu, Chen Zhao, Shi Feng, Hal Daumé III, Jordan Boyd-Graber
cs.CL, cs.HC
Large Language Models (LLMs) are increasingly used for accessing information on the web. Their truthfulness and factuality are thus of great interest. To help users make the right decisions about the information they're getting, LLMs should not only provide but also help users fact-check information. In this paper, we ...
http://arxiv.org/pdf/2310.12558v1
2023-10-19T08:09:58
2023-10-19T08:09:58
{ "title": [ 0.0011663437, -0.011489868, -0.026535034, -0.0007686615, -0.048736572, -0.040039062, -0.0049972534, -0.030441284, -0.018569946, 0.045196533, -0.057647705, -0.004196167, 0.01776123, -0.00806427, -0.029510498, -0.048950195, -0.024215698, ...
http://arxiv.org/abs/2310.12557v1
DepWiGNN: A Depth-wise Graph Neural Network for Multi-hop Spatial Reasoning in Text
Shuaiyi Li, Yang Deng, Wai Lam
cs.CL, cs.AI
Spatial reasoning in text plays a crucial role in various real-world applications. Existing approaches for spatial reasoning typically infer spatial relations from pure text, which overlook the gap between natural language and symbolic structures. Graph neural networks (GNNs) have showcased exceptional proficiency in i...
http://arxiv.org/pdf/2310.12557v1
2023-10-19T08:07:22
2023-10-19T08:07:22
{ "title": [ -0.0118637085, -0.038879395, -0.014831543, -0.03149414, -0.032104492, 0.0051956177, -0.024551392, 0.016921997, 0.025985718, 0.02947998, -0.00932312, -0.027664185, 0.003709793, -0.02154541, -0.054779053, -0.012145996, -0.0012483597, 0...
http://arxiv.org/abs/2310.12541v2
Large Language Model for Multi-objective Evolutionary Optimization
Fei Liu, Xi Lin, Zhenkun Wang, Shunyu Yao, Xialiang Tong, Mingxuan Yuan, Qingfu Zhang
cs.NE, cs.AI, cs.CL, cs.ET
Multiobjective evolutionary algorithms (MOEAs) are major methods for solving multiobjective optimization problems (MOPs). Many MOEAs have been proposed in the past decades, of which the search operators need a carefully handcrafted design with domain knowledge. Recently, some attempts have been made to replace the manu...
http://arxiv.org/pdf/2310.12541v2
2023-10-25T10:11:12
2023-10-19T07:46:54
{ "title": [ -0.049835205, -0.005836487, -0.0124053955, -0.014404297, 0.0016460419, 0.0012750626, -0.034240723, -0.028213501, -0.0019836426, 0.043670654, 0.023544312, 0.008415222, 0.012268066, -0.03363037, -0.033172607, 0.033447266, -0.016540527, ...
http://arxiv.org/abs/2310.13028v1
Reliable Academic Conference Question Answering: A Study Based on Large Language Model
Zhiwei Huang, Long Jin, Junjie Wang, Mingchen Tu, Yin Hua, Zhiqiang Liu, Jiawei Meng, Huajun Chen, Wen Zhang
cs.CL, cs.AI
The rapid growth of computer science has led to a proliferation of research presented at academic conferences, fostering global scholarly communication. Researchers consistently seek accurate, current information about these events at all stages. This data surge necessitates an intelligent question-answering system to ...
http://arxiv.org/pdf/2310.13028v1
2023-10-19T07:39:07
2023-10-19T07:39:07
{ "title": [ -0.020309448, 0.0054779053, -0.039520264, -0.06439209, 0.0073661804, 0.007499695, -0.055847168, -0.0027637482, 0.019882202, 0.034942627, -0.0025253296, -0.018005371, 0.014472961, -0.014183044, -0.0635376, -0.0022583008, -0.0113220215, ...
http://arxiv.org/abs/2310.12537v1
Product Attribute Value Extraction using Large Language Models
Alexander Brinkmann, Roee Shraga, Christian Bizer
cs.CL
E-commerce applications such as faceted product search or product comparison are based on structured product descriptions like attribute/value pairs. The vendors on e-commerce platforms do not provide structured product descriptions but describe offers using titles or descriptions. To process such offers, it is necessa...
http://arxiv.org/pdf/2310.12537v1
2023-10-19T07:39:00
2023-10-19T07:39:00
{ "title": [ -0.052001953, -0.012649536, -0.04888916, -0.052734375, -0.001411438, 0.0048217773, -0.05807495, -0.0038032532, 0.01424408, 0.027130127, -0.006046295, -0.02520752, 0.0013942719, -0.023208618, -0.046142578, 0.0071983337, -0.013404846, ...
http://arxiv.org/abs/2310.12531v2
ICU: Conquering Language Barriers in Vision-and-Language Modeling by Dividing the Tasks into Image Captioning and Language Understanding
Guojun Wu
cs.CL
Most multilingual vision-and-language (V&L) research aims to accomplish multilingual and multimodal capabilities within one model. However, the scarcity of multilingual captions for images has hindered the development. To overcome this obstacle, we propose ICU, Image Caption Understanding, which divides a V&L task into...
http://arxiv.org/pdf/2310.12531v2
2023-10-20T03:35:11
2023-10-19T07:11:48
{ "title": [ 0.011451721, -0.029205322, -0.012481689, -0.024475098, -0.037628174, -0.01751709, -0.023895264, -0.00831604, 0.020141602, 0.017318726, -0.04675293, -0.0033721924, 0.016555786, -0.02418518, -0.008834839, -0.031402588, -0.0031719208, 0...
http://arxiv.org/abs/2310.12522v1
Named Entity Recognition for Monitoring Plant Health Threats in Tweets: a ChouBERT Approach
Shufan Jiang, Rafael Angarita, Stéphane Cormier, Francis Rousseaux
cs.CL, cs.LG
An important application scenario of precision agriculture is detecting and measuring crop health threats using sensors and data analysis techniques. However, the textual data are still under-explored among the existing solutions due to the lack of labelled data and fine-grained semantic resources. Recent research sugg...
http://arxiv.org/pdf/2310.12522v1
2023-10-19T06:54:55
2023-10-19T06:54:55
{ "title": [ 0.008964539, 0.010192871, -0.018035889, -0.0826416, 0.014854431, 0.006515503, -0.037902832, 0.01058197, -0.016983032, 0.022918701, -0.045654297, 0.046203613, -0.014709473, -0.01348114, -0.0135269165, -0.01802063, -0.014717102, 0.0342...
http://arxiv.org/abs/2310.13025v1
Powerset multi-class cross entropy loss for neural speaker diarization
Alexis Plaquet, Hervé Bredin
cs.SD, cs.AI, cs.CL, cs.NE, eess.AS
Since its introduction in 2019, the whole end-to-end neural diarization (EEND) line of work has been addressing speaker diarization as a frame-wise multi-label classification problem with permutation-invariant training. Despite EEND showing great promise, a few recent works took a step back and studied the possible com...
http://arxiv.org/pdf/2310.13025v1
2023-10-19T06:51:43
2023-10-19T06:51:43
{ "title": [ -0.017456055, 0.00002771616, -0.03338623, -0.020965576, -0.010047913, -0.0049972534, -0.011352539, -0.0055236816, -0.008651733, 0.06402588, -0.006137848, 0.01928711, 0.032287598, -0.011993408, -0.026779175, 0.022003174, 0.0049972534, ...
http://arxiv.org/abs/2310.12520v1
Lost in Translation: When GPT-4V(ision) Can't See Eye to Eye with Text. A Vision-Language-Consistency Analysis of VLLMs and Beyond
Xiang Zhang, Senyu Li, Zijun Wu, Ning Shi
cs.CL, cs.CV
Recent advancements in multimodal techniques open exciting possibilities for models excelling in diverse tasks involving text, audio, and image processing. Models like GPT-4V, blending computer vision and language modeling, excel in complex text and image tasks. Numerous prior research endeavors have diligently examine...
http://arxiv.org/pdf/2310.12520v1
2023-10-19T06:45:11
2023-10-19T06:45:11
{ "title": [ -0.046936035, -0.007381439, -0.0011510849, 0.047546387, -0.04324341, 0.0074043274, -0.0074691772, -0.033172607, 0.040649414, 0.031097412, -0.012435913, -0.033355713, 0.0037269592, -0.03125, -0.0546875, -0.02458191, 0.0011129379, 0.02...
http://arxiv.org/abs/2310.12516v1
Automatic Hallucination Assessment for Aligned Large Language Models via Transferable Adversarial Attacks
Xiaodong Yu, Hao Cheng, Xiaodong Liu, Dan Roth, Jianfeng Gao
cs.CL, cs.AI, cs.LG
Although remarkable progress has been achieved in preventing large language model (LLM) hallucinations using instruction tuning and retrieval augmentation, it remains challenging to measure the reliability of LLMs using human-crafted evaluation data which is not available for many tasks and domains and could suffer fro...
http://arxiv.org/pdf/2310.12516v1
2023-10-19T06:37:32
2023-10-19T06:37:32
{ "title": [ -0.00031161308, 0.0033683777, -0.023483276, -0.03463745, -0.01473999, 0.00258255, -0.030426025, -0.015808105, -0.0014476776, 0.04248047, 0.006603241, 0.010986328, 0.012107849, -0.024795532, -0.045013428, 0.009048462, -0.00756073, 0.0...
http://arxiv.org/abs/2310.13024v1
Towards Anytime Fine-tuning: Continually Pre-trained Language Models with Hypernetwork Prompt
Gangwei Jiang, Caigao Jiang, Siqiao Xue, James Y. Zhang, Jun Zhou, Defu Lian, Ying Wei
cs.CL, cs.AI, cs.LG
Continual pre-training has been urgent for adapting a pre-trained model to a multitude of domains and tasks in the fast-evolving world. In practice, a continually pre-trained model is expected to demonstrate not only greater capacity when fine-tuned on pre-trained domains but also a non-decreasing performance on unseen...
http://arxiv.org/pdf/2310.13024v1
2023-10-19T06:34:40
2023-10-19T06:34:40
{ "title": [ -0.0034637451, -0.017730713, -0.042938232, -0.077941895, 0.0027503967, -0.005531311, -0.0552063, 0.025314331, 0.0036373138, 0.038909912, -0.013328552, 0.00026273727, 0.021408081, -0.018997192, -0.047210693, 0.010253906, 0.0025978088, ...
http://arxiv.org/abs/2310.13023v1
GraphGPT: Graph Instruction Tuning for Large Language Models
Jiabin Tang, Yuhao Yang, Wei Wei, Lei Shi, Lixin Su, Suqi Cheng, Dawei Yin, Chao Huang
cs.CL, cs.AI
Graph Neural Networks (GNNs) have advanced graph structure understanding via recursive information exchange and aggregation among graph nodes. To improve model robustness, self-supervised learning (SSL) has emerged as a promising approach for data augmentation. However, existing methods for generating pre-trained graph...
http://arxiv.org/pdf/2310.13023v1
2023-10-19T06:17:46
2023-10-19T06:17:46
{ "title": [ -0.06781006, -0.036865234, -0.0362854, -0.0619812, -0.02923584, 0.021148682, -0.08721924, 0.02470398, -0.0048561096, 0.025039673, -0.021575928, 0.017684937, -0.003768921, -0.014556885, -0.028289795, 0.00932312, -0.009986877, 0.043762...
http://arxiv.org/abs/2310.12505v1
Attack Prompt Generation for Red Teaming and Defending Large Language Models
Boyi Deng, Wenjie Wang, Fuli Feng, Yang Deng, Qifan Wang, Xiangnan He
cs.CL, cs.CR, cs.LG
Large language models (LLMs) are susceptible to red teaming attacks, which can induce LLMs to generate harmful content. Previous research constructs attack prompts via manual or automatic methods, which have their own limitations on construction cost and quality. To address these issues, we propose an integrated approa...
http://arxiv.org/pdf/2310.12505v1
2023-10-19T06:15:05
2023-10-19T06:15:05
{ "title": [ -0.024047852, -0.01083374, -0.031829834, -0.055908203, -0.004043579, -0.0013751984, -0.039764404, 0.0063934326, -0.043762207, 0.013916016, 0.0048332214, 0.013877869, -0.027755737, -0.030776978, -0.013092041, -0.01927185, -0.00945282, ...
http://arxiv.org/abs/2310.12490v1
Co$^2$PT: Mitigating Bias in Pre-trained Language Models through Counterfactual Contrastive Prompt Tuning
Xiangjue Dong, Ziwei Zhu, Zhuoer Wang, Maria Teleki, James Caverlee
cs.CL
Pre-trained Language Models are widely used in many important real-world applications. However, recent studies show that these models can encode social biases from large pre-training corpora and even amplify biases in downstream applications. To address this challenge, we propose Co$^2$PT, an efficient and effective de...
http://arxiv.org/pdf/2310.12490v1
2023-10-19T05:49:58
2023-10-19T05:49:58
{ "title": [ 0.012878418, -0.0031223297, -0.013702393, -0.046661377, -0.020706177, -0.014183044, -0.058502197, -0.029190063, -0.0066604614, 0.03262329, -0.034301758, 0.0051727295, 0.018615723, -0.025268555, -0.019638062, -0.011627197, -0.011299133, ...
http://arxiv.org/abs/2310.12489v2
MedAI Dialog Corpus (MEDIC): Zero-Shot Classification of Doctor and AI Responses in Health Consultations
Olumide E. Ojo, Olaronke O. Adebanji, Alexander Gelbukh, Hiram Calvo, Anna Feldman
cs.CL
Zero-shot classification enables text to be classified into classes not seen during training. In this research, we investigate the effectiveness of pre-trained language models to accurately classify responses from Doctors and AI in health consultations through zero-shot learning. Our study aims to determine whether the...
http://arxiv.org/pdf/2310.12489v2
2023-10-20T12:21:39
2023-10-19T05:48:28
{ "title": [ 0.011428833, -0.01486969, -0.041046143, -0.024047852, -0.03050232, -0.03933716, -0.0211792, -0.007972717, 0.0025691986, 0.045898438, -0.0051651, 0.016357422, 0.019897461, -0.03729248, -0.030059814, 0.00023508072, 0.020751953, 0.02136...
http://arxiv.org/abs/2310.12481v1
Not All Countries Celebrate Thanksgiving: On the Cultural Dominance in Large Language Models
Wenxuan Wang, Wenxiang Jiao, Jingyuan Huang, Ruyi Dai, Jen-tse Huang, Zhaopeng Tu, Michael R. Lyu
cs.CL, cs.AI
In this paper, we identify a cultural dominance issue within large language models (LLMs) due to the predominant use of English data in model training (e.g. ChatGPT). LLMs often provide inappropriate English-culture-related answers that are not relevant to the expected culture when users ask in non-English languages. T...
http://arxiv.org/pdf/2310.12481v1
2023-10-19T05:38:23
2023-10-19T05:38:23
{ "title": [ 0.014892578, -0.005207062, -0.02394104, -0.048095703, -0.0060920715, -0.03543091, -0.021530151, -0.00055217743, -0.031799316, 0.008979797, -0.019500732, -0.010894775, 0.016677856, -0.04788208, 0.02154541, -0.02432251, -0.0015354156, ...
http://arxiv.org/abs/2310.12477v1
An Exploration of In-Context Learning for Speech Language Model
Ming-Hao Hsu, Kai-Wei Chang, Shang-Wen Li, Hung-yi Lee
eess.AS, cs.AI, cs.CL
Ever since the development of GPT-3 in the natural language processing (NLP) field, in-context learning (ICL) has played an important role in utilizing large language models (LLMs). By presenting the LM utterance-label demonstrations at the input, the LM can accomplish few-shot learning without relying on gradient desc...
http://arxiv.org/pdf/2310.12477v1
2023-10-19T05:31:45
2023-10-19T05:31:45
{ "title": [ 0.001408577, -0.0033512115, -0.04071045, -0.061401367, 0.0102005005, -0.01374054, -0.03326416, 0.0181427, -0.008262634, 0.04837036, -0.03768921, 0.015129089, 0.0115737915, -0.005760193, -0.036193848, -0.0027198792, -0.015274048, 0.04...
http://arxiv.org/abs/2310.12467v2
Contrastive Learning for Inference in Dialogue
Etsuko Ishii, Yan Xu, Bryan Wilie, Ziwei Ji, Holy Lovenia, Willy Chung, Pascale Fung
cs.CL
Inference, especially those derived from inductive processes, is a crucial component in our conversation to complement the information implicitly or explicitly conveyed by a speaker. While recent large language models show remarkable advances in inference tasks, their performance in inductive reasoning, where not all i...
http://arxiv.org/pdf/2310.12467v2
2023-11-13T04:18:58
2023-10-19T04:49:36
{ "title": [ -0.0011320114, -0.011894226, -0.021560669, -0.06866455, 0.0042419434, -0.004764557, -0.021453857, 0.0027885437, 0.020874023, 0.0073280334, -0.032592773, -0.0029144287, -0.027450562, -0.015945435, -0.046691895, -0.015701294, -0.020950317,...
http://arxiv.org/abs/2310.12462v1
Unmasking Transformers: A Theoretical Approach to Data Recovery via Attention Weights
Yichuan Deng, Zhao Song, Shenghao Xie, Chiwun Yang
cs.LG, cs.CL, stat.ML
In the realm of deep learning, transformers have emerged as a dominant architecture, particularly in natural language processing tasks. However, with their widespread adoption, concerns regarding the security and privacy of the data processed by these models have arisen. In this paper, we address a pivotal question: Ca...
http://arxiv.org/pdf/2310.12462v1
2023-10-19T04:41:01
2023-10-19T04:41:01
{ "title": [ 0.0020980835, 0.008636475, -0.04864502, -0.06604004, 0.037017822, -0.011383057, -0.015396118, -0.0072669983, 0.008148193, 0.027114868, -0.012619019, 0.002073288, -0.0014753342, -0.0073547363, -0.04498291, 0.0018939972, 0.009780884, 0...
http://arxiv.org/abs/2310.12454v1
Rethinking the Construction of Effective Metrics for Understanding the Mechanisms of Pretrained Language Models
You Li, Jinhui Yin, Yuming Lin
cs.CL, cs.AI
Pretrained language models are expected to effectively map input text to a set of vectors while preserving the inherent relationships within the text. Consequently, designing a white-box model to compute metrics that reflect the presence of specific internal relations in these vectors has become a common approach for p...
http://arxiv.org/pdf/2310.12454v1
2023-10-19T04:16:40
2023-10-19T04:16:40
{ "title": [ -0.026641846, -0.015899658, -0.014015198, -0.037841797, -0.01902771, -0.0051345825, -0.043273926, -0.0093688965, -0.019210815, 0.032958984, 0.014457703, 0.009895325, 0.014549255, -0.032043457, -0.028839111, 0.017791748, -0.0008163452, ...
http://arxiv.org/abs/2310.12450v2
A Read-and-Select Framework for Zero-shot Entity Linking
Zhenran Xu, Yulin Chen, Baotian Hu, Min Zhang
cs.CL
Zero-shot entity linking (EL) aims at aligning entity mentions to unseen entities to challenge the generalization ability. Previous methods largely focus on the candidate retrieval stage and ignore the essential candidate ranking stage, which disambiguates among entities and makes the final linking prediction. In this ...
http://arxiv.org/pdf/2310.12450v2
2023-10-29T14:58:54
2023-10-19T04:08:10
{ "title": [ 0.020355225, -0.010932922, -0.03527832, -0.08520508, 0.019729614, -0.01574707, -0.024475098, 0.0016498566, 0.020950317, 0.040374756, -0.008369446, 0.011253357, 0.016052246, -0.018463135, -0.037872314, 0.014785767, 0.0006790161, 0.018...
http://arxiv.org/abs/2310.12444v1
Revisiting Sparse Retrieval for Few-shot Entity Linking
Yulin Chen, Zhenran Xu, Baotian Hu, Min Zhang
cs.CL
Entity linking aims to link ambiguous mentions to their corresponding entities in a knowledge base. One of the key challenges comes from insufficient labeled data for specific domains. Although dense retrievers have achieved excellent performance on several benchmarks, their performance decreases significantly when onl...
http://arxiv.org/pdf/2310.12444v1
2023-10-19T03:51:10
2023-10-19T03:51:10
{ "title": [ -0.004447937, -0.019683838, -0.03955078, -0.055847168, 0.008621216, -0.003774643, -0.016952515, 0.011756897, 0.016296387, 0.033416748, -0.0005517006, 0.0027809143, 0.020065308, -0.020263672, -0.03656006, 0.0073280334, 0.012260437, -0...
http://arxiv.org/abs/2310.12443v1
Know Where to Go: Make LLM a Relevant, Responsible, and Trustworthy Searcher
Xiang Shi, Jiawei Liu, Yinpeng Liu, Qikai Cheng, Wei Lu
cs.IR, cs.AI, cs.CL
The advent of Large Language Models (LLMs) has shown the potential to improve relevance and provide direct answers in web searches. However, challenges arise in validating the reliability of generated results and the credibility of contributing sources, due to the limitations of traditional information retrieval algori...
http://arxiv.org/pdf/2310.12443v1
2023-10-19T03:49:36
2023-10-19T03:49:36
{ "title": [ 0.02293396, 0.008155823, -0.030776978, -0.05407715, -0.007095337, -0.017242432, -0.032958984, -0.0019283295, -0.033447266, 0.0035934448, 0.006614685, 0.021408081, -0.032562256, -0.06958008, 0.035369873, -0.045837402, -0.011650085, 0....
http://arxiv.org/abs/2310.12442v1
Efficient Long-Range Transformers: You Need to Attend More, but Not Necessarily at Every Layer
Qingru Zhang, Dhananjay Ram, Cole Hawkins, Sheng Zha, Tuo Zhao
cs.CL, cs.LG
Pretrained transformer models have demonstrated remarkable performance across various natural language processing tasks. These models leverage the attention mechanism to capture long- and short-range dependencies in the sequence. However, the (full) attention mechanism incurs high computational cost - quadratic in the ...
http://arxiv.org/pdf/2310.12442v1
2023-10-19T03:32:05
2023-10-19T03:32:05
{ "title": [ 0.032287598, 0.022445679, -0.025268555, -0.033721924, 0.027893066, -0.032806396, 0.029541016, -0.027557373, -0.03729248, 0.021453857, -0.024734497, 0.018310547, -0.0079422, -0.032226562, 0.010475159, -0.036315918, 0.013931274, 0.0368...
http://arxiv.org/abs/2310.12439v1
PoisonPrompt: Backdoor Attack on Prompt-based Large Language Models
Hongwei Yao, Jian Lou, Zhan Qin
cs.CL, cs.AI, I.2.7, I.2.7
Prompts have significantly improved the performance of pretrained Large Language Models (LLMs) on various downstream tasks recently, making them increasingly indispensable for a diverse range of LLM application scenarios. However, the backdoor vulnerability, a serious security threat that can maliciously alter the vict...
http://arxiv.org/pdf/2310.12439v1
2023-10-19T03:25:28
2023-10-19T03:25:28
{ "title": [ -0.029541016, -0.015357971, -0.040283203, -0.056243896, -0.00033307076, -0.003458023, -0.047851562, 0.00025558472, -0.022781372, 0.018157959, -0.00548172, 0.008018494, -0.0026168823, -0.029510498, -0.016937256, -0.0069351196, -0.01322937...
http://arxiv.org/abs/2310.12430v1
DocXChain: A Powerful Open-Source Toolchain for Document Parsing and Beyond
Cong Yao
cs.CV, cs.CL
In this report, we introduce DocXChain, a powerful open-source toolchain for document parsing, which is designed and developed to automatically convert the rich information embodied in unstructured documents, such as text, tables and charts, into structured representations that are readable and manipulable by machines....
http://arxiv.org/pdf/2310.12430v1
2023-10-19T02:49:09
2023-10-19T02:49:09
{ "title": [ -0.04296875, -0.018066406, -0.06829834, -0.076171875, -0.0035171509, 0.001080513, -0.07458496, 0.005794525, -0.005214691, 0.03704834, -0.042541504, 0.020126343, 0.020965576, -0.008735657, -0.027496338, -0.008560181, 0.0037002563, 0.0...
http://arxiv.org/abs/2310.12426v1
MAF: Multi-Aspect Feedback for Improving Reasoning in Large Language Models
Deepak Nathani, David Wang, Liangming Pan, William Yang Wang
cs.CL, cs.AI
Language Models (LMs) have shown impressive performance in various natural language tasks. However, when it comes to natural language reasoning, LMs still face challenges such as hallucination, generating incorrect intermediate reasoning steps, and making mathematical errors. Recent research has focused on enhancing LM...
http://arxiv.org/pdf/2310.12426v1
2023-10-19T02:32:39
2023-10-19T02:32:39
{ "title": [ -0.026107788, -0.029525757, -0.03036499, -0.089782715, -0.012298584, 0.018661499, -0.07873535, 0.026947021, 0.027618408, 0.030700684, 0.00038313866, -0.0075912476, 0.033050537, -0.0206604, -0.046020508, 0.026489258, -0.015975952, 0.0...
http://arxiv.org/abs/2310.13022v1
Uncertainty-aware Parameter-Efficient Self-training for Semi-supervised Language Understanding
Jianing Wang, Qiushi Sun, Nuo Chen, Chengyu Wang, Jun Huang, Ming Gao, Xiang Li
cs.LG, cs.AI, cs.CL
The recent success of large pre-trained language models (PLMs) heavily hinges on massive labeled data, which typically produces inferior performance in low-resource scenarios. To remedy this dilemma, we study self-training as one of the predominant semi-supervised learning (SSL) approaches, which utilizes large-scale u...
http://arxiv.org/pdf/2310.13022v1
2023-10-19T02:18:29
2023-10-19T02:18:29
{ "title": [ -0.012268066, 0.013549805, -0.031982422, -0.08001709, -0.0061035156, 0.0038719177, -0.074645996, 0.013366699, -0.0017614365, 0.0032444, -0.003824234, -0.0104904175, -0.010536194, -0.03173828, -0.028396606, -0.012016296, 0.002538681, ...
http://arxiv.org/abs/2310.12418v1
The Shifted and The Overlooked: A Task-oriented Investigation of User-GPT Interactions
Siru Ouyang, Shuohang Wang, Yang Liu, Ming Zhong, Yizhu Jiao, Dan Iter, Reid Pryzant, Chenguang Zhu, Heng Ji, Jiawei Han
cs.CL
Recent progress in Large Language Models (LLMs) has produced models that exhibit remarkable performance across a variety of NLP tasks. However, it remains unclear whether the existing focus of NLP research accurately captures the genuine requirements of human users. This paper provides a comprehensive analysis of the d...
http://arxiv.org/pdf/2310.12418v1
2023-10-19T02:12:17
2023-10-19T02:12:17
{ "title": [ 0.0016832352, 0.00617218, -0.019866943, 0.000749588, 0.0034942627, -0.038482666, 0.0070877075, -0.05014038, -0.0141067505, 0.013084412, -0.025543213, 0.0037765503, -0.020614624, -0.023361206, -0.043914795, -0.011482239, -0.025924683, ...
http://arxiv.org/abs/2310.12406v1
FinEntity: Entity-level Sentiment Classification for Financial Texts
Yixuan Tang, Yi Yang, Allen H Huang, Andy Tam, Justin Z Tang
cs.CL
In the financial domain, conducting entity-level sentiment analysis is crucial for accurately assessing the sentiment directed toward a specific financial entity. To our knowledge, no publicly available dataset currently exists for this purpose. In this work, we introduce an entity-level sentiment classification datase...
http://arxiv.org/pdf/2310.12406v1
2023-10-19T01:38:40
2023-10-19T01:38:40
{ "title": [ -0.008155823, -0.004470825, -0.04458618, -0.052093506, 0.015731812, -0.012084961, -0.014274597, -0.0028057098, 0.032958984, 0.040527344, -0.043914795, 0.004371643, 0.01701355, 0.010002136, -0.07952881, -0.0036907196, -0.02368164, 0.0...
http://arxiv.org/abs/2310.12404v1
Loop Copilot: Conducting AI Ensembles for Music Generation and Iterative Editing
Yixiao Zhang, Akira Maezawa, Gus Xia, Kazuhiko Yamamoto, Simon Dixon
cs.SD, cs.CL, cs.HC, cs.LG, eess.AS
Creating music is iterative, requiring varied methods at each stage. However, existing AI music systems fall short in orchestrating multiple subsystems for diverse needs. To address this gap, we introduce Loop Copilot, a novel system that enables users to generate and iteratively refine music through an interactive, mu...
http://arxiv.org/pdf/2310.12404v1
2023-10-19T01:20:12
2023-10-19T01:20:12
{ "title": [ -0.02760315, -0.01739502, -0.05407715, -0.03390503, -0.019042969, -0.021209717, -0.019256592, 0.008018494, -0.0030460358, 0.044708252, -0.019699097, 0.012069702, 0.03753662, -0.019088745, -0.026519775, 0.007247925, -0.019561768, 0.03...
http://arxiv.org/abs/2310.12379v1
Solving Hard Analogy Questions with Relation Embedding Chains
Nitesh Kumar, Steven Schockaert
cs.CL, cs.AI
Modelling how concepts are related is a central topic in Lexical Semantics. A common strategy is to rely on knowledge graphs (KGs) such as ConceptNet, and to model the relation between two concepts as a set of paths. However, KGs are limited to a fixed set of relation types, and they are incomplete and often noisy. Ano...
http://arxiv.org/pdf/2310.12379v1
2023-10-18T23:13:22
2023-10-18T23:13:22
{ "title": [ -0.0043945312, -0.019622803, -0.01725769, -0.047973633, -0.005794525, 0.00605011, -0.010597229, -0.011199951, 0.04763794, 0.0008339882, -0.02949524, 0.012252808, -0.020629883, -0.014923096, -0.032409668, 0.008491516, -0.00095033646, ...
http://arxiv.org/abs/2310.12362v1
REMARK-LLM: A Robust and Efficient Watermarking Framework for Generative Large Language Models
Ruisi Zhang, Shehzeen Samarah Hussain, Paarth Neekhara, Farinaz Koushanfar
cs.CR, cs.CL
We present REMARK-LLM, a novel efficient, and robust watermarking framework designed for texts generated by large language models (LLMs). Synthesizing human-like content using LLMs necessitates vast computational resources and extensive datasets, encapsulating critical intellectual property (IP). However, the generated...
http://arxiv.org/pdf/2310.12362v1
2023-10-18T22:14:37
2023-10-18T22:14:37
{ "title": [ -0.012329102, 0.004383087, -0.047546387, -0.07293701, 0.017333984, -0.0033416748, -0.059783936, -0.009506226, -0.0035533905, 0.031799316, -0.010772705, 0.02180481, 0.017181396, -0.02659607, -0.022323608, 0.011352539, -0.027145386, 0....
http://arxiv.org/abs/2310.12360v1
GRI: Graph-based Relative Isomorphism of Word Embedding Spaces
Muhammad Asif Ali, Yan Hu, Jianbin Qin, Di Wang
cs.CL
Automated construction of bilingual dictionaries using monolingual embedding spaces is a core challenge in machine translation. The end performance of these dictionaries relies upon the geometric similarity of individual spaces, i.e., their degree of isomorphism. Existing attempts aimed at controlling the relative isom...
http://arxiv.org/pdf/2310.12360v1
2023-10-18T22:10:47
2023-10-18T22:10:47
{ "title": [ -0.03555298, -0.011054993, -0.02809143, -0.033721924, -0.01399231, 0.012306213, -0.05722046, -0.023147583, -0.005783081, 0.032043457, -0.03363037, 0.020217896, 0.009544373, -0.01751709, -0.014205933, 0.0020942688, -0.012023926, 0.025...
http://arxiv.org/abs/2310.12352v1
knn-seq: Efficient, Extensible kNN-MT Framework
Hiroyuki Deguchi, Hayate Hirano, Tomoki Hoshino, Yuto Nishida, Justin Vasselli, Taro Watanabe
cs.CL
k-nearest-neighbor machine translation (kNN-MT) boosts the translation quality of a pre-trained neural machine translation (NMT) model by utilizing translation examples during decoding. Translation examples are stored in a vector database, called a datastore, which contains one entry for each target token from the para...
http://arxiv.org/pdf/2310.12352v1
2023-10-18T21:56:04
2023-10-18T21:56:04
{ "title": [ -0.041015625, -0.035827637, -0.051513672, -0.07940674, -0.0040893555, 0.018981934, -0.07067871, 0.001627922, 0.0036296844, 0.0418396, 0.0031204224, 0.012367249, 0.036468506, -0.032043457, -0.026870728, 0.036499023, -0.0018177032, 0.0...
http://arxiv.org/abs/2310.12344v1
LACMA: Language-Aligning Contrastive Learning with Meta-Actions for Embodied Instruction Following
Cheng-Fu Yang, Yen-Chun Chen, Jianwei Yang, Xiyang Dai, Lu Yuan, Yu-Chiang Frank Wang, Kai-Wei Chang
cs.CL, cs.CV
End-to-end Transformers have demonstrated an impressive success rate for Embodied Instruction Following when the environment has been seen in training. However, they tend to struggle when deployed in an unseen environment. This lack of generalizability is due to the agent's insensitivity to subtle changes in natural la...
http://arxiv.org/pdf/2310.12344v1
2023-10-18T21:43:07
2023-10-18T21:43:07
{ "title": [ -0.029296875, -0.018432617, -0.019073486, -0.0024719238, -0.017028809, -0.003025055, -0.017623901, -0.034362793, -0.017501831, 0.036224365, -0.014076233, 0.040008545, -0.010475159, -0.011756897, -0.01889038, -0.0017147064, -0.017242432, ...
http://arxiv.org/abs/2310.12342v2
Eliminating Reasoning via Inferring with Planning: A New Framework to Guide LLMs' Non-linear Thinking
Yongqi Tong, Yifan Wang, Dawei Li, Sizhe Wang, Zi Lin, Simeng Han, Jingbo Shang
cs.CL, cs.AI
Chain-of-Thought(CoT) prompting and its variants explore equipping large language models (LLMs) with high-level reasoning abilities by emulating human-like linear cognition and logic. However, the human mind is complicated and mixed with both linear and nonlinear thinking. In this work, we propose \textbf{I}nferential ...
http://arxiv.org/pdf/2310.12342v2
2023-11-15T00:59:54
2023-10-18T21:42:16
{ "title": [ -0.009811401, -0.02029419, -0.041381836, -0.034423828, -0.042877197, -0.016296387, -0.0440979, -0.014945984, -0.033569336, -0.009803772, -0.044891357, 0.019897461, -0.041259766, -0.030075073, 0.0034561157, -0.049835205, -0.008544922, ...
http://arxiv.org/abs/2310.12318v1
The Sentiment Problem: A Critical Survey towards Deconstructing Sentiment Analysis
Pranav Narayanan Venkit, Mukund Srinath, Sanjana Gautam, Saranya Venkatraman, Vipul Gupta, Rebecca J. Passonneau, Shomir Wilson
cs.CL, cs.AI, cs.CY, cs.HC
We conduct an inquiry into the sociotechnical aspects of sentiment analysis (SA) by critically examining 189 peer-reviewed papers on their applications, models, and datasets. Our investigation stems from the recognition that SA has become an integral component of diverse sociotechnical systems, exerting influence on bo...
http://arxiv.org/pdf/2310.12318v1
2023-10-18T20:42:44
2023-10-18T20:42:44
{ "title": [ 0.0025634766, -0.009918213, -0.011566162, -0.03616333, 0.020706177, 0.0017633438, 0.009109497, -0.026611328, 0.012779236, 0.02909851, -0.028884888, -0.0016708374, -0.021621704, -0.0064086914, -0.074157715, -0.008460999, -0.012924194, ...
http://arxiv.org/abs/2310.12303v1
Document-Level Language Models for Machine Translation
Frithjof Petrick, Christian Herold, Pavel Petrushkov, Shahram Khadivi, Hermann Ney
cs.CL, cs.AI, cs.LG
Despite the known limitations, most machine translation systems today still operate on the sentence-level. One reason for this is, that most parallel training data is only sentence-level aligned, without document-level meta information available. In this work, we set out to build context-aware translation systems utili...
http://arxiv.org/pdf/2310.12303v1
2023-10-18T20:10:07
2023-10-18T20:10:07
{ "title": [ -0.04321289, -0.02104187, -0.012382507, -0.058380127, -0.01399231, 0.022567749, -0.064331055, 0.006855011, 0.015960693, 0.045074463, -0.009017944, -0.00869751, 0.03173828, -0.017120361, -0.049468994, 0.024475098, -0.005455017, 0.0267...
http://arxiv.org/abs/2310.12300v1
Measuring Pointwise $\mathcal{V}$-Usable Information In-Context-ly
Sheng Lu, Shan Chen, Yingya Li, Danielle Bitterman, Guergana Savova, Iryna Gurevych
cs.CL
In-context learning (ICL) is a new learning paradigm that has gained popularity along with the development of large language models. In this work, we adapt a recently proposed hardness metric, pointwise $\mathcal{V}$-usable information (PVI), to an in-context version (in-context PVI). Compared to the original PVI, in-c...
http://arxiv.org/pdf/2310.12300v1
2023-10-18T20:07:41
2023-10-18T20:07:41
{ "title": [ -0.014175415, -0.009849548, -0.049224854, -0.036987305, -0.0017328262, -0.026947021, -0.019180298, -0.03062439, -0.012191772, 0.022628784, -0.06536865, 0.0022697449, 0.0071754456, -0.027862549, 0.007045746, -0.012184143, 0.018676758, ...
http://arxiv.org/abs/2310.12274v1
An Image is Worth Multiple Words: Learning Object Level Concepts using Multi-Concept Prompt Learning
Chen Jin, Ryutaro Tanno, Amrutha Saseendran, Tom Diethe, Philip Teare
cs.CV, cs.AI, cs.CL, cs.GR, cs.LG
Textural Inversion, a prompt learning method, learns a singular embedding for a new "word" to represent image style and appearance, allowing it to be integrated into natural language sentences to generate novel synthesised images. However, identifying and integrating multiple object-level concepts within one scene pose...
http://arxiv.org/pdf/2310.12274v1
2023-10-18T19:18:19
2023-10-18T19:18:19
{ "title": [ 0.028747559, -0.017547607, -0.030593872, -0.019897461, -0.010025024, -0.049987793, 0.000562191, -0.02168274, -0.0032730103, -0.0015554428, -0.041503906, 0.026351929, -0.007987976, -0.033721924, 0.009773254, -0.015541077, -0.0017824173, ...
http://arxiv.org/abs/2310.12236v1
Direct Neural Machine Translation with Task-level Mixture of Experts models
Isidora Chara Tourni, Subhajit Naskar
cs.CL
Direct neural machine translation (direct NMT) is a type of NMT system that translates text between two non-English languages. Direct NMT systems often face limitations due to the scarcity of parallel data between non-English language pairs. Several approaches have been proposed to address this limitation, such as mult...
http://arxiv.org/pdf/2310.12236v1
2023-10-18T18:19:45
2023-10-18T18:19:45
{ "title": [ -0.02859497, 0.007385254, -0.023895264, -0.062316895, 0.0027122498, 0.010437012, -0.04876709, 0.0027236938, -0.002527237, 0.055419922, 0.014434814, 0.004196167, 0.014450073, -0.021530151, -0.05432129, 0.028442383, 0.0044021606, 0.009...