id stringlengths 33 34 | doc_url stringlengths 33 34 | title stringlengths 10 211 | publication_date timestamp[s] | update_date timestamp[s] | authors stringlengths 5 20.1k | category_primary stringclasses 78
values | category_all stringlengths 5 138 | abstract stringlengths 123 3.09k | embeddings listlengths 768 768 |
|---|---|---|---|---|---|---|---|---|---|
http://arxiv.org/abs/2406.14774v1 | http://arxiv.org/pdf/2406.14774v1 | Evaluating Numerical Reasoning in Text-to-Image Models | 2024-06-20T22:56:31 | 2024-06-20T22:56:31 | Ivana Kajić, Olivia Wiles, Isabela Albuquerque, Matthias Bauer, Su Wang, Jordi Pont-Tuset, Aida Nematzadeh | cs.LG | cs.LG, cs.CL, cs.CV | Text-to-image generative models are capable of producing high-quality images
that often faithfully depict concepts described using natural language. In this
work, we comprehensively evaluate a range of text-to-image models on numerical
reasoning tasks of varying difficulty, and show that even the most advanced
models h... | [
-0.6450158358,
-0.7507883906,
0.8648600578,
0.4987634718,
-0.5082787871,
-0.2995869815,
-0.1535764486,
-0.3653519452,
-0.4006504714,
0.4557495415,
-0.2786026001,
-0.4806334972,
-0.5920608044,
0.4440529644,
-0.4473797977,
0.8108893633,
0.2331992686,
0.0088767512,
-0.8570514321,
... |
http://arxiv.org/abs/2406.14765v1 | http://arxiv.org/pdf/2406.14765v1 | ChatGPT as Research Scientist: Probing GPT's Capabilities as a Research Librarian, Research Ethicist, Data Generator and Data Predictor | 2024-06-20T22:30:06 | 2024-06-20T22:30:06 | Steven A. Lehr, Aylin Caliskan, Suneragiri Liyanage, Mahzarin R. Banaji | cs.AI | cs.AI, cs.CL, cs.CY, cs.IR, cs.LG, I.2.7; K.4.0; K.4.1; K.4.2 | How good a research scientist is ChatGPT? We systematically probed the
capabilities of GPT-3.5 and GPT-4 across four central components of the
scientific process: as a Research Librarian, Research Ethicist, Data Generator,
and Novel Data Predictor, using psychological science as a testing field. In
Study 1 (Research Li... | [
-0.3720831275,
-0.6669790745,
0.7121431828,
-0.1644846648,
0.0430342816,
0.4192367494,
-0.1057561487,
-0.4077023566,
0.0613600835,
0.1318705231,
-0.1802058816,
0.0907119438,
-0.6809720397,
0.2368048131,
-0.5453727245,
1.3516677618,
0.3000825047,
0.2707598805,
-0.0150159895,
0.3... |
http://arxiv.org/abs/2406.14764v1 | http://arxiv.org/pdf/2406.14764v1 | RE-AdaptIR: Improving Information Retrieval through Reverse Engineered Adaptation | 2024-06-20T22:28:11 | 2024-06-20T22:28:11 | William Fleshman, Benjamin Van Durme | cs.IR | cs.IR, cs.AI, cs.CL, cs.LG | Large language models (LLMs) fine-tuned for text-retrieval have demonstrated
state-of-the-art results across several information retrieval (IR) benchmarks.
However, supervised training for improving these models requires numerous
labeled examples, which are generally unavailable or expensive to acquire. In
this work, w... | [
0.068849735,
-0.5640432239,
0.4041002989,
-0.121188201,
-0.2681658268,
0.1149803326,
-0.3222028315,
-0.4345251918,
0.1457082629,
0.8131992817,
-0.3531134129,
-0.4758424461,
-0.5383353829,
0.2124954015,
-0.53101331,
0.8944633007,
-0.0522114001,
0.0243971907,
0.0073029394,
-0.236... |
http://arxiv.org/abs/2406.14763v1 | http://arxiv.org/pdf/2406.14763v1 | A Learn-Then-Reason Model Towards Generalization in Knowledge Base Question Answering | 2024-06-20T22:22:41 | 2024-06-20T22:22:41 | Lingxi Zhang, Jing Zhang, Yanling Wang, Cuiping Li, Hong Chen | cs.CL | cs.CL, cs.AI | Large-scale knowledge bases (KBs) like Freebase and Wikidata house millions
of structured knowledge. Knowledge Base Question Answering (KBQA) provides a
user-friendly way to access these valuable KBs via asking natural language
questions. In order to improve the generalization capabilities of KBQA models,
extensive res... | [
-0.4158180058,
-0.8664929867,
0.7855358124,
-0.0876777917,
-0.1357204914,
-0.0666005611,
-0.3271337152,
-0.3219361603,
-0.063629128,
0.4866551459,
-0.5351132154,
-0.5703454018,
-0.4920509458,
-0.1767789572,
-0.1527283788,
0.7485285997,
-0.0629614443,
0.1375015676,
-0.2630025148,
... |
http://arxiv.org/abs/2406.14760v1 | http://arxiv.org/pdf/2406.14760v1 | An LLM Feature-based Framework for Dialogue Constructiveness Assessment | 2024-06-20T22:10:52 | 2024-06-20T22:10:52 | Lexin Zhou, Youmna Farag, Andreas Vlachos | cs.CL | cs.CL, cs.AI, cs.LG | Research on dialogue constructiveness assessment focuses on (i) analysing
conversational factors that influence individuals to take specific actions, win
debates, change their perspectives or broaden their open-mindedness and (ii)
predicting constructive outcomes following dialogues for such use cases. These
objectives... | [
-0.5576332211,
-0.5552437901,
0.527757585,
0.2481209934,
-0.2382860333,
0.2841172516,
-0.4175410867,
-0.4130576849,
0.4265156984,
0.4084275067,
-0.6165342927,
-0.2687928677,
-0.6423543096,
0.0605660416,
-0.2505522966,
0.909090519,
0.0722560063,
0.2814161181,
-0.2414756566,
-0.3... |
http://arxiv.org/abs/2406.15516v1 | http://arxiv.org/pdf/2406.15516v1 | System Description for the Displace Speaker Diarization Challenge 2023 | 2024-06-20T21:40:02 | 2024-06-20T21:40:02 | Ali Aliyev | cs.CL | cs.CL, cs.SD, eess.AS | This paper describes our solution for the Diarization of Speaker and Language
in Conversational Environments Challenge (Displace 2023). We used a combination
of VAD for finding segfments with speech, Resnet architecture based CNN for
feature extraction from these segments, and spectral clustering for features
clusterin... | [
-0.8857351542,
-0.9151104093,
0.1420563906,
0.2360845953,
-0.0845999196,
0.5479671359,
-0.0801388696,
-0.3699557185,
0.1296833605,
0.4850788414,
-0.5610865355,
-0.5307276845,
-0.5485543609,
0.0552944317,
-0.5871953368,
0.6004202962,
0.4576636851,
-0.0629544333,
-0.3315938413,
0... |
http://arxiv.org/abs/2406.14747v1 | http://arxiv.org/pdf/2406.14747v1 | An Adapter-Based Unified Model for Multiple Spoken Language Processing Tasks | 2024-06-20T21:39:04 | 2024-06-20T21:39:04 | Varsha Suresh, Salah Aït-Mokhtar, Caroline Brun, Ioan Calapodescu | cs.CL | cs.CL, cs.AI | Self-supervised learning models have revolutionized the field of speech
processing. However, the process of fine-tuning these models on downstream
tasks requires substantial computational resources, particularly when dealing
with multiple speech-processing tasks. In this paper, we explore the potential
of adapter-based... | [
-0.4174527526,
-0.5168088078,
0.2518521249,
0.1743980497,
0.0721223503,
-0.5735746026,
-0.3642814755,
-0.4020485282,
-0.3598380983,
0.6346770525,
-0.3774125874,
-0.0921099484,
-0.4784859121,
0.14034006,
-0.233103618,
0.6877509356,
0.3084050119,
-0.0675115511,
0.1895399541,
-0.3... |
http://arxiv.org/abs/2406.14745v2 | http://arxiv.org/pdf/2406.14745v2 | Relation Extraction with Fine-Tuned Large Language Models in Retrieval Augmented Generation Frameworks | 2024-06-20T21:27:57 | 2024-06-24T06:57:05 | Sefika Efeoglu, Adrian Paschke | cs.CL | cs.CL, cs.AI | Information Extraction (IE) is crucial for converting unstructured data into
structured formats like Knowledge Graphs (KGs). A key task within IE is
Relation Extraction (RE), which identifies relationships between entities in
text. Various RE methods exist, including supervised, unsupervised, weakly
supervised, and rul... | [
-0.4153825939,
-1.0681920052,
0.7264672518,
0.0205142759,
-0.2672357261,
0.0129583245,
-0.2679311037,
-0.6341534853,
-0.0593046211,
0.6601557136,
-0.4219642282,
-0.4802074134,
-0.5849565268,
0.3623486459,
-0.3533078134,
0.9598116279,
-0.1011070013,
-0.2230763435,
-0.3343122303,
... |
http://arxiv.org/abs/2406.14739v1 | http://arxiv.org/pdf/2406.14739v1 | Learning to Retrieve Iteratively for In-Context Learning | 2024-06-20T21:07:55 | 2024-06-20T21:07:55 | Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Benjamin Van Durme | cs.CL | cs.CL | We introduce iterative retrieval, a novel framework that empowers retrievers
to make iterative decisions through policy optimization. Finding an optimal
portfolio of retrieved items is a combinatorial optimization problem, generally
considered NP-hard. This approach provides a learned approximation to such a
solution, ... | [
-0.0755194277,
-0.3705627322,
0.5985168815,
0.3413606882,
-0.1538895667,
0.1710382104,
-0.5041173697,
0.0947569907,
-0.0881901607,
0.4084868729,
-0.6025852561,
-0.349008888,
-0.488471508,
0.004743577,
-0.4565256238,
0.9960143566,
-0.1037834808,
-0.0172876604,
-0.1249284148,
0.0... |
http://arxiv.org/abs/2406.14737v1 | http://arxiv.org/pdf/2406.14737v1 | Dissecting the Ullman Variations with a SCALPEL: Why do LLMs fail at Trivial Alterations to the False Belief Task? | 2024-06-20T21:02:30 | 2024-06-20T21:02:30 | Zhiqiang Pi, Annapurna Vadaparty, Benjamin K. Bergen, Cameron R. Jones | cs.CL | cs.CL | Recent empirical results have sparked a debate about whether or not Large
Language Models (LLMs) are capable of Theory of Mind (ToM). While some have
found LLMs to be successful on ToM evaluations such as the False Belief task
(Kosinski, 2023), others have argued that LLMs solve these tasks by exploiting
spurious corre... | [
-0.5237378478,
-0.724345088,
1.0636142492,
-0.0354739949,
-0.0669045374,
0.0199073181,
-0.2703857422,
-0.1506845355,
0.3142354488,
0.3061075807,
-0.6121825576,
-0.2387031019,
-0.6427532434,
-0.0360729732,
-0.5671992302,
0.9539136887,
-0.1117615849,
-0.1444717646,
0.0614192896,
... |
http://arxiv.org/abs/2406.14732v1 | http://arxiv.org/pdf/2406.14732v1 | TTQA-RS- A break-down prompting approach for Multi-hop Table-Text Question Answering with Reasoning and Summarization | 2024-06-20T20:55:38 | 2024-06-20T20:55:38 | Jayetri Bardhan, Bushi Xiao, Daisy Zhe Wang | cs.CL | cs.CL, cs.IR | Question answering (QA) over tables and text has gained much popularity over
the years. Multi-hop table-text QA requires multiple hops between the table and
text, making it a challenging QA task. Although several works have attempted to
solve the table-text QA task, most involve training the models and requiring
labele... | [
-0.3660227656,
-0.9275000691,
0.8415390849,
0.1000968441,
-0.3346714079,
0.108962521,
0.344574213,
-0.1294775754,
0.1462926865,
0.3712124527,
-0.6740628481,
-0.3242608309,
-0.2945003808,
0.6209068894,
-0.3789978623,
0.9311336875,
-0.1647614837,
0.065527536,
-0.1965636164,
0.033... |
http://arxiv.org/abs/2406.14721v1 | http://arxiv.org/pdf/2406.14721v1 | 1+1>2: Can Large Language Models Serve as Cross-Lingual Knowledge Aggregators? | 2024-06-20T20:32:53 | 2024-06-20T20:32:53 | Yue Huang, Chenrui Fan, Yuan Li, Siyuan Wu, Tianyi Zhou, Xiangliang Zhang, Lichao Sun | cs.CL | cs.CL | Large Language Models (LLMs) have garnered significant attention due to their
remarkable ability to process information across various languages. Despite
their capabilities, they exhibit inconsistencies in handling identical queries
in different languages, presenting challenges for further advancement. This
paper intro... | [
-0.5444226265,
-0.7000589371,
0.480787009,
0.0824538991,
-0.2505363822,
0.4194275737,
-0.4213835895,
-0.8836154342,
0.2670155764,
0.5149832964,
-0.1974890381,
-0.2434736043,
-0.7378919721,
0.2119788677,
-0.3393386304,
0.4357160926,
0.1833263338,
0.3870224357,
-0.2169232965,
-0.... |
http://arxiv.org/abs/2406.14711v2 | http://arxiv.org/pdf/2406.14711v2 | MultiAgent Collaboration Attack: Investigating Adversarial Attacks in Large Language Model Collaborations via Debate | 2024-06-20T20:09:37 | 2024-06-26T16:05:20 | Alfonso Amayuelas, Xianjun Yang, Antonis Antoniades, Wenyue Hua, Liangming Pan, William Wang | cs.CL | cs.CL, cs.AI, cs.MA | Large Language Models (LLMs) have shown exceptional results on current
benchmarks when working individually. The advancement in their capabilities,
along with a reduction in parameter size and inference times, has facilitated
the use of these models as agents, enabling interactions among multiple models
to execute comp... | [
-0.6175351739,
-1.0194767714,
0.6510981917,
0.1486884356,
0.1452267617,
0.3341750801,
-0.2405457646,
-0.4059101343,
0.0893152729,
0.6629477143,
-0.4153360426,
-0.2025851756,
-0.8973324895,
0.1154995412,
-0.5113272667,
0.7698726058,
0.1186516508,
0.301371485,
-0.1049080342,
-0.2... |
http://arxiv.org/abs/2406.14709v1 | http://arxiv.org/pdf/2406.14709v1 | Factual Dialogue Summarization via Learning from Large Language Models | 2024-06-20T20:03:37 | 2024-06-20T20:03:37 | Rongxin Zhu, Jey Han Lau, Jianzhong Qi | cs.CL | cs.CL, F.2.2; I.2.7 | Factual consistency is an important quality in dialogue summarization. Large
language model (LLM)-based automatic text summarization models generate more
factually consistent summaries compared to those by smaller pretrained language
models, but they face deployment challenges in real-world applications due to
privacy ... | [
0.0030430728,
-0.5933214426,
0.8533447981,
-0.0538865775,
-0.3999324143,
0.3026916981,
-0.1497174054,
-0.3155017495,
0.1225503087,
0.4909613132,
-0.4410046041,
-0.6583306193,
-0.7784115672,
0.2521185577,
-0.2979163826,
1.2537385225,
0.0711430088,
0.0066673611,
-0.1541333944,
-0... |
http://arxiv.org/abs/2406.14703v1 | http://arxiv.org/pdf/2406.14703v1 | Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics | 2024-06-20T19:50:56 | 2024-06-20T19:50:56 | Seungbeen Lee, Seungwon Lim, Seungju Han, Giyeong Oh, Hyungjoo Chae, Jiwan Chung, Minju Kim, Beong-woo Kwak, Yeonsoo Lee, Dongha Lee, Jinyoung Yeo, Youngjae Yu | cs.CL | cs.CL, cs.AI | The idea of personality in descriptive psychology, traditionally defined
through observable behavior, has now been extended to Large Language Models
(LLMs) to better understand their behavior. This raises a question: do LLMs
exhibit distinct and consistent personality traits, similar to humans? Existing
self-assessment... | [
-0.5001643896,
-0.5976584554,
0.7243365645,
0.5545961857,
-0.0219038092,
0.3581477702,
-0.2474932671,
-0.6224115491,
0.4709462523,
0.3533344567,
-0.7309138179,
-0.2048334479,
-0.2856808007,
0.3323387206,
-0.0967532247,
0.9622027278,
0.1909131259,
0.33162871,
-0.1741205454,
-0.5... |
http://arxiv.org/abs/2406.14701v1 | http://arxiv.org/pdf/2406.14701v1 | Speech Prefix-Tuning with RNNT Loss for Improving LLM Predictions | 2024-06-20T19:50:49 | 2024-06-20T19:50:49 | Murali Karthick Baskar, Andrew Rosenberg, Bhuvana Ramabhadran, Neeraj Gaur, Zhong Meng | cs.AI | cs.AI, cs.CL, cs.SD, eess.AS | In this paper, we focus on addressing the constraints faced when applying
LLMs to ASR. Recent works utilize prefixLM-type models, which directly apply
speech as a prefix to LLMs for ASR. We have found that optimizing speech
prefixes leads to better ASR performance and propose applying RNNT loss to
perform speech prefix... | [
-0.3554490805,
-0.5434411168,
0.1880983114,
0.000558822,
-0.5604220033,
0.2168839276,
-0.6738179326,
-0.3645718694,
0.263056159,
0.4089761078,
-0.798301518,
-0.2604552507,
-0.3819678724,
0.4341053963,
-0.1054696664,
0.5876138806,
0.1055052653,
0.4157918096,
-0.1010599062,
0.075... |
http://arxiv.org/abs/2406.14695v1 | http://arxiv.org/pdf/2406.14695v1 | Depth $F_1$: Improving Evaluation of Cross-Domain Text Classification by Measuring Semantic Generalizability | 2024-06-20T19:35:17 | 2024-06-20T19:35:17 | Parker Seegmiller, Joseph Gatto, Sarah Masud Preum | cs.CL | cs.CL, cs.AI | Recent evaluations of cross-domain text classification models aim to measure
the ability of a model to obtain domain-invariant performance in a target
domain given labeled samples in a source domain. The primary strategy for this
evaluation relies on assumed differences between source domain samples and
target domain s... | [
-0.4545004368,
-0.7363836169,
0.355707109,
0.2743334472,
-0.3407861292,
0.3328892589,
-0.0504004695,
-0.6256193519,
-0.1750665009,
0.2956876755,
-0.5658706427,
-0.9448228478,
-1.0472648144,
0.0598980002,
-0.8243933916,
0.8843878508,
-0.3108980656,
-0.1250593513,
-0.7440040708,
... |
http://arxiv.org/abs/2406.14686v1 | http://arxiv.org/pdf/2406.14686v1 | A Contrastive Learning Approach to Mitigate Bias in Speech Models | 2024-06-20T19:20:00 | 2024-06-20T19:20:00 | Alkis Koudounas, Flavio Giobergia, Eliana Pastor, Elena Baralis | cs.CL | cs.CL, cs.LG, eess.AS | Speech models may be affected by performance imbalance in different
population subgroups, raising concerns about fair treatment across these
groups. Prior attempts to mitigate unfairness either focus on user-defined
subgroups, potentially overlooking other affected subgroups, or do not
explicitly improve the internal r... | [
-0.3472152948,
-0.3065893054,
0.079567723,
-0.0083326744,
-0.1878452152,
0.0408015177,
0.0661411807,
-0.2400531173,
-0.2564811409,
0.4044173956,
-0.535620451,
-0.3307448924,
-0.6484724283,
-0.4037046731,
-0.1372715384,
0.9291855097,
0.3202360868,
0.8445875049,
-0.208833009,
-0.... |
http://arxiv.org/abs/2406.14683v1 | http://arxiv.org/pdf/2406.14683v1 | TAGLAS: An atlas of text-attributed graph datasets in the era of large graph and language models | 2024-06-20T19:11:35 | 2024-06-20T19:11:35 | Jiarui Feng, Hao Liu, Lecheng Kong, Yixin Chen, Muhan Zhang | cs.LG | cs.LG, cs.CL | In this report, we present TAGLAS, an atlas of text-attributed graph (TAG)
datasets and benchmarks. TAGs are graphs with node and edge features
represented in text, which have recently gained wide applicability in training
graph-language or graph foundation models. In TAGLAS, we collect and integrate
more than 23 TAG d... | [
-0.0528903268,
-0.2351441979,
0.1743533164,
0.1200610399,
-0.0994243398,
0.39293769,
-0.2059383392,
-0.6295877695,
0.1133940294,
0.4854662418,
-0.1248872355,
-0.7108180523,
-0.4989356995,
0.0936454013,
-0.5023328066,
1.3864445686,
-0.2329947203,
0.0531116761,
-0.4216741323,
-0.... |
http://arxiv.org/abs/2406.14680v1 | http://arxiv.org/pdf/2406.14680v1 | Dravidian language family through Universal Dependencies lens | 2024-06-20T18:59:46 | 2024-06-20T18:59:46 | Taraka Rama, Sowmya Vajjala | cs.CL | cs.CL | The Universal Dependencies (UD) project aims to create a cross-linguistically
consistent dependency annotation for multiple languages, to facilitate
multilingual NLP. It currently supports 114 languages. Dravidian languages are
spoken by over 200 million people across the word, and yet there are only two
languages from... | [
-0.4681417644,
-0.0852641389,
0.3560789526,
0.3277853429,
0.0796496943,
0.3253219128,
-0.1123961881,
-0.8275851607,
0.2565945387,
0.4536575377,
-0.4356467426,
-0.13924326,
-0.3692952096,
0.3537276387,
0.1227000207,
0.6227709651,
-0.213435635,
0.1168697998,
0.2793641388,
-0.8380... |
http://arxiv.org/abs/2406.14678v1 | http://arxiv.org/pdf/2406.14678v1 | Bidirectional Transformer Representations of (Spanish) Ambiguous Words in Context: A New Lexical Resource and Empirical Analysis | 2024-06-20T18:58:11 | 2024-06-20T18:58:11 | Pamela D. Rivière, Anne L. Beatty-Martínez, Sean Trott | cs.CL | cs.CL | Lexical ambiguity -- where a single wordform takes on distinct,
context-dependent meanings -- serves as a useful tool to compare across
different large language models' (LLMs') ability to form distinct,
contextualized representations of the same stimulus. Few studies have
systematically compared LLMs' contextualized wo... | [
-0.3871052265,
-0.5057547092,
0.4298923016,
0.6731954813,
-0.3172177672,
0.0927130952,
-0.3534923196,
-0.5913058519,
0.3109814823,
0.0900079533,
-0.6919232011,
-0.1469106376,
-0.7829942107,
0.4965358078,
-0.3903742135,
0.5347675681,
-0.0806029961,
0.0805204809,
-0.1878678054,
-... |
http://arxiv.org/abs/2406.14673v1 | http://arxiv.org/pdf/2406.14673v1 | Insights into LLM Long-Context Failures: When Transformers Know but Don't Tell | 2024-06-20T18:50:44 | 2024-06-20T18:50:44 | Taiming Lu, Muhan Gao, Kuai Yu, Adam Byerly, Daniel Khashabi | cs.CL | cs.CL | Large Language Models (LLMs) exhibit positional bias, struggling to utilize
information from the middle or end of long contexts. Our study explores LLMs'
long-context reasoning by probing their hidden representations. We find that
while LLMs encode the position of target information, they often fail to
leverage this in... | [
-0.4668393433,
-0.5947821736,
1.1685771942,
0.210178405,
0.0147449933,
0.1338380128,
-0.2619515657,
-0.5219240189,
-0.063211903,
0.4236453176,
-0.7806671858,
-0.2212174386,
-0.8478266597,
0.0280605052,
-0.6419131756,
0.7318317294,
-0.1252246648,
-0.2143172622,
-0.1121446043,
-0... |
http://arxiv.org/abs/2406.14670v1 | http://arxiv.org/pdf/2406.14670v1 | Exploring Design Choices for Building Language-Specific LLMs | 2024-06-20T18:47:43 | 2024-06-20T18:47:43 | Atula Tejaswi, Nilesh Gupta, Eunsol Choi | cs.CL | cs.CL, cs.AI, cs.LG | Despite rapid progress in large language models (LLMs), their performance on
a vast majority of languages remain unsatisfactory. In this paper, we study
building language-specific LLMs by adapting monolingual and multilingual LLMs.
We conduct systematic experiments on how design choices (base model selection,
vocabular... | [
-0.4786574841,
-0.6929725409,
0.3638459444,
0.3695619106,
-0.4317836463,
0.1455033273,
-0.7292798758,
-0.5581233501,
0.1673375219,
0.4216771722,
-0.210147664,
-0.4279489517,
-0.4078389406,
0.2944944203,
0.0613748021,
0.7913987041,
-0.2823356092,
0.0259297434,
-0.0575280003,
-0.... |
http://arxiv.org/abs/2406.15513v1 | http://arxiv.org/pdf/2406.15513v1 | PKU-SafeRLHF: A Safety Alignment Preference Dataset for Llama Family Models | 2024-06-20T18:37:36 | 2024-06-20T18:37:36 | Jiaming Ji, Donghai Hong, Borong Zhang, Boyuan Chen, Josef Dai, Boren Zheng, Tianyi Qiu, Boxun Li, Yaodong Yang | cs.AI | cs.AI, cs.CL | In this work, we introduce the PKU-SafeRLHF dataset, designed to promote
research on safety alignment in large language models (LLMs). As a sibling
project to SafeRLHF and BeaverTails, we separate annotations of helpfulness and
harmlessness for question-answering pairs, providing distinct perspectives on
these coupled ... | [
-0.1205624267,
-0.6214584708,
0.2529472411,
0.1665946245,
-0.26765275,
-0.0785699338,
0.0536674969,
-0.5767305493,
0.1616705656,
0.5657168627,
-0.3882918954,
-0.5711954236,
-0.6188187003,
-0.0420601256,
-0.1736606359,
1.3129462004,
0.0368797518,
-0.3177995384,
-0.0656333566,
0.... |
http://arxiv.org/abs/2406.14666v1 | http://arxiv.org/pdf/2406.14666v1 | Co-training for Low Resource Scientific Natural Language Inference | 2024-06-20T18:35:47 | 2024-06-20T18:35:47 | Mobashir Sadat, Cornelia Caragea | cs.CL | cs.CL | Scientific Natural Language Inference (NLI) is the task of predicting the
semantic relation between a pair of sentences extracted from research articles.
The automatic annotation method based on distant supervision for the training
set of SciNLI (Sadat and Caragea, 2022b), the first and most popular dataset
for this ta... | [
-0.2898049951,
-0.5428587794,
0.1381731629,
0.1991876066,
-0.2441305071,
-0.0153933847,
-0.3654188514,
-0.7340219617,
0.4040914774,
0.1059693024,
-0.6017801166,
-0.5217056274,
-0.4709730446,
0.3862951994,
0.0119667882,
1.2275824547,
0.0760178417,
0.2512931824,
-0.3165827096,
-0... |
http://arxiv.org/abs/2406.14657v1 | http://arxiv.org/pdf/2406.14657v1 | OpenDebateEvidence: A Massive-Scale Argument Mining and Summarization Dataset | 2024-06-20T18:22:59 | 2024-06-20T18:22:59 | Allen Roush, Yusuf Shabazz, Arvind Balaji, Peter Zhang, Stefano Mezza, Markus Zhang, Sanjay Basu, Sriram Vishwanath, Mehdi Fatemi, Ravid Schwartz-Ziv | cs.CL | cs.CL, cs.AI, cs.LG | We introduce OpenDebateEvidence, a comprehensive dataset for argument mining
and summarization sourced from the American Competitive Debate community. This
dataset includes over 3.5 million documents with rich metadata, making it one
of the most extensive collections of debate evidence. OpenDebateEvidence
captures the ... | [
-0.586399138,
-0.9008769989,
0.276104182,
0.0068114735,
-0.2289499342,
-0.1060597003,
-0.1954174191,
0.2375586182,
0.1162870079,
0.4690715373,
-0.1758793741,
-0.5804313421,
-0.3976027369,
0.005703235,
-0.2278537601,
1.1455819607,
-0.2764148414,
0.0940972641,
-0.7263225913,
-0.4... |
http://arxiv.org/abs/2406.14654v1 | http://arxiv.org/pdf/2406.14654v1 | Major Entity Identification: A Generalizable Alternative to Coreference Resolution | 2024-06-20T18:17:58 | 2024-06-20T18:17:58 | Kawshik Manikantan, Shubham Toshniwal, Makarand Tapaswi, Vineet Gandhi | cs.CL | cs.CL, cs.AI, cs.LG, I.2.7 | The limited generalization of coreference resolution (CR) models has been a
major bottleneck in the task's broad application. Prior work has identified
annotation differences, especially for mention detection, as one of the main
reasons for the generalization gap and proposed using additional annotated
target domain da... | [
-0.5596628785,
-0.7912008166,
0.5523093939,
-0.1395845413,
-0.3319064975,
-0.0235598888,
-0.7029735446,
-0.7003625035,
0.2461697757,
0.3363848925,
-0.6247876883,
-0.6334124804,
-0.9179551601,
0.2626031935,
-0.308850348,
0.985817194,
0.1146253273,
-0.1355320811,
-0.2810549736,
-... |
http://arxiv.org/abs/2406.19415v1 | http://arxiv.org/pdf/2406.19415v1 | An Analysis of Multilingual FActScore | 2024-06-20T18:09:40 | 2024-06-20T18:09:40 | Kim Trong Vu, Michael Krumdick, Varshini Reddy, Franck Dernoncourt, Viet Dac Lai | cs.CL | cs.CL | FActScore has gained popularity as a metric to estimate the factuality of
long-form texts generated by Large Language Models (LLMs) in English. However,
there has not been any work in studying the behavior of FActScore in other
languages. This paper studies the limitations of each component in the
four-component pipeli... | [
-0.2455511093,
-0.7109746933,
0.8233038187,
0.247468397,
-0.19838202,
0.0995400771,
-0.5278069377,
-0.6205058694,
0.2398235798,
0.3352908492,
-0.4609607458,
-0.6282635331,
-0.6203220487,
0.3576308489,
-0.1076375768,
1.3609472513,
0.0042950343,
-0.0150653375,
-0.651029408,
-0.43... |
http://arxiv.org/abs/2406.14644v1 | http://arxiv.org/pdf/2406.14644v1 | Unveiling the Spectrum of Data Contamination in Language Models: A Survey from Detection to Remediation | 2024-06-20T18:07:26 | 2024-06-20T18:07:26 | Chunyuan Deng, Yilun Zhao, Yuzhao Heng, Yitong Li, Jiannan Cao, Xiangru Tang, Arman Cohan | cs.CL | cs.CL | Data contamination has garnered increased attention in the era of large
language models (LLMs) due to the reliance on extensive internet-derived
training corpora. The issue of training corpus overlap with evaluation
benchmarks--referred to as contamination--has been the focus of significant
recent research. This body o... | [
-0.5234677792,
-1.0544872284,
0.5885725021,
-0.1641660035,
0.0748075843,
0.4574871957,
-0.1724233776,
-0.6986294985,
-0.0197782628,
1.0961581469,
-0.2661701739,
-0.426810801,
-0.7464258671,
-0.0959860981,
-0.9524288177,
0.9090952277,
-0.1084311157,
-0.3221201003,
-0.4489751458,
... |
http://arxiv.org/abs/2406.14643v1 | http://arxiv.org/pdf/2406.14643v1 | Holistic Evaluation for Interleaved Text-and-Image Generation | 2024-06-20T18:07:19 | 2024-06-20T18:07:19 | Minqian Liu, Zhiyang Xu, Zihao Lin, Trevor Ashby, Joy Rimchala, Jiaxin Zhang, Lifu Huang | cs.CV | cs.CV, cs.AI, cs.CL | Interleaved text-and-image generation has been an intriguing research
direction, where the models are required to generate both images and text
pieces in an arbitrary order. Despite the emerging advancements in interleaved
generation, the progress in its evaluation still significantly lags behind.
Existing evaluation b... | [
-0.5711271167,
-0.5367170572,
0.3858139217,
0.6458963156,
-0.5222363472,
0.2750462294,
-0.2225728631,
-0.4785775244,
-0.5438292623,
0.0044373819,
-0.4917402267,
-0.3741366267,
-0.620341599,
0.0062191552,
-0.4210162461,
0.9646821618,
-0.002324322,
0.1414617449,
-0.6451676488,
-0... |
http://arxiv.org/abs/2406.14629v1 | http://arxiv.org/pdf/2406.14629v1 | Can LLMs Learn by Teaching? A Preliminary Study | 2024-06-20T18:00:17 | 2024-06-20T18:00:17 | Xuefei Ning, Zifu Wang, Shiyao Li, Zinan Lin, Peiran Yao, Tianyu Fu, Matthew B. Blaschko, Guohao Dai, Huazhong Yang, Yu Wang | cs.CL | cs.CL, cs.AI | Teaching to improve student models (e.g., knowledge distillation) is an
extensively studied methodology in LLMs. However, for humans, teaching not only
improves students but also improves teachers. We ask: Can LLMs also learn by
teaching (LbT)? If yes, we can potentially unlock the possibility of
continuously advancing... | [
-0.1971277297,
-0.6686341166,
0.6291469932,
-0.2504683733,
-0.2464007884,
0.3620225489,
-0.2268089801,
-0.3831445873,
-0.0968151093,
0.0623678565,
-0.5436737537,
-0.2082861066,
-0.4620209634,
-0.1309098601,
-0.2811815441,
1.0122095346,
0.1140205041,
0.7364276648,
-0.4275926054,
... |
http://arxiv.org/abs/2406.14563v1 | http://arxiv.org/pdf/2406.14563v1 | Model Merging and Safety Alignment: One Bad Model Spoils the Bunch | 2024-06-20T17:59:58 | 2024-06-20T17:59:58 | Hasan Abed Al Kader Hammoud, Umberto Michieli, Fabio Pizzati, Philip Torr, Adel Bibi, Bernard Ghanem, Mete Ozay | cs.CL | cs.CL, cs.AI, cs.LG | Merging Large Language Models (LLMs) is a cost-effective technique for
combining multiple expert LLMs into a single versatile model, retaining the
expertise of the original ones. However, current approaches often overlook the
importance of safety alignment during merging, leading to highly misaligned
models. This work ... | [
-0.2944256365,
-0.4192869067,
0.6374003291,
-0.1141440347,
-0.1712766886,
0.2785697877,
0.1399750262,
-0.7859007716,
-0.2375563383,
0.6715629697,
-0.4310202599,
-0.2304538041,
-0.6074368954,
-0.4020476937,
-0.7208327651,
1.0338314772,
-0.1242022067,
0.0217581559,
-0.4551791847,
... |
http://arxiv.org/abs/2406.14562v1 | http://arxiv.org/pdf/2406.14562v1 | Whiteboard-of-Thought: Thinking Step-by-Step Across Modalities | 2024-06-20T17:59:45 | 2024-06-20T17:59:45 | Sachit Menon, Richard Zemel, Carl Vondrick | cs.CL | cs.CL, cs.AI, cs.CV | When presented with questions involving visual thinking, humans naturally
switch reasoning modalities, often forming mental images or drawing visual
aids. Large language models have shown promising results in arithmetic and
symbolic reasoning by expressing intermediate reasoning in text as a chain of
thought, yet strug... | [
-0.4010312557,
-1.0947127342,
1.1639533043,
0.2090794593,
-0.4170559645,
0.098089911,
-0.1606775075,
-0.3599009812,
-0.0147849917,
0.1937240511,
-0.7165887952,
-0.3885970414,
-0.6000438333,
-0.0017061025,
-0.0542714968,
0.6165799499,
-0.386734575,
0.0695403814,
-0.1228155345,
-... |
http://arxiv.org/abs/2406.14561v1 | http://arxiv.org/pdf/2406.14561v1 | How to Compute the Probability of a Word | 2024-06-20T17:59:42 | 2024-06-20T17:59:42 | Tiago Pimentel, Clara Meister | cs.CL | cs.CL | Language models (LMs) estimate the probability distribution over sequences of
natural language; these distributions are crucial for computing perplexity and
surprisal in linguistics research. While we are usually concerned with
measuring these values for words, most LMs operate over subwords. Despite
seemingly straight... | [
-0.0593256913,
-0.6803171039,
0.7700092793,
0.0053703347,
-0.3775993586,
-0.1519058943,
0.1546871364,
-0.4113385379,
0.0104264226,
0.1586455107,
-0.3415819407,
-0.3040645719,
-0.7699462771,
-0.2098847777,
-0.3501437306,
0.8496171236,
0.1056139395,
0.0932323188,
-0.0916357189,
-... |
http://arxiv.org/abs/2406.14553v1 | http://arxiv.org/pdf/2406.14553v1 | xCOMET-lite: Bridging the Gap Between Efficiency and Quality in Learned MT Evaluation Metrics | 2024-06-20T17:58:34 | 2024-06-20T17:58:34 | Daniil Larionov, Mikhail Seleznyov, Vasiliy Viskov, Alexander Panchenko, Steffen Eger | cs.CL | cs.CL | State-of-the-art trainable machine translation evaluation metrics like xCOMET
achieve high correlation with human judgment but rely on large encoders (up to
10.7B parameters), making them computationally expensive and inaccessible to
researchers with limited resources. To address this issue, we investigate
whether the ... | [
-0.3287221789,
-0.0986426175,
0.4408054352,
-0.2022265643,
-0.6189002395,
0.3667992055,
-0.6001640558,
-0.0948599651,
0.1870091408,
0.1953983903,
-0.5316206813,
-0.3413265347,
-0.8323444724,
0.1858699769,
-0.3451929688,
0.968044579,
-0.0584457293,
-0.056406565,
0.104222253,
-0.... |
http://arxiv.org/abs/2406.14550v1 | http://arxiv.org/pdf/2406.14550v1 | GraphReader: Building Graph-based Agent to Enhance Long-Context Abilities of Large Language Models | 2024-06-20T17:57:51 | 2024-06-20T17:57:51 | Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, Bo Zheng | cs.CL | cs.CL, cs.AI | Long-context capabilities are essential for large language models (LLMs) to
tackle complex and long-input tasks. Despite numerous efforts made to optimize
LLMs for long contexts, challenges persist in robustly processing long inputs.
In this paper, we introduce GraphReader, a graph-based agent system designed to
handle... | [
-0.369702369,
-1.0121406317,
0.9596708417,
0.1170672402,
-0.2803260684,
0.4143102169,
-0.3591278791,
-0.7835140228,
0.1267112046,
0.6870361567,
-0.3738771975,
-0.5738030076,
-0.6342861056,
0.3585852683,
-0.1771578789,
0.9029595256,
-0.2819212973,
-0.5668255091,
-0.1561932117,
-... |
http://arxiv.org/abs/2406.14546v1 | http://arxiv.org/pdf/2406.14546v1 | Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data | 2024-06-20T17:55:04 | 2024-06-20T17:55:04 | Johannes Treutlein, Dami Choi, Jan Betley, Cem Anil, Samuel Marks, Roger Baker Grosse, Owain Evans | cs.CL | cs.CL, cs.AI, cs.LG | One way to address safety risks from large language models (LLMs) is to
censor dangerous knowledge from their training data. While this removes the
explicit information, implicit information can remain scattered across various
training documents. Could an LLM infer the censored knowledge by piecing
together these impli... | [
-0.2280125767,
-0.9979028702,
0.9641753435,
-0.2238666266,
-0.1141119972,
0.2519948184,
-0.4275802076,
-0.5340943336,
-0.2372928411,
0.4228208661,
-0.453715235,
-0.5464199781,
-0.62133044,
-0.2922841311,
-0.3699759543,
0.8847393394,
-0.0999237448,
0.0563344657,
-0.4694107473,
-... |
http://arxiv.org/abs/2406.14545v1 | http://arxiv.org/pdf/2406.14545v1 | Unmasking Database Vulnerabilities: Zero-Knowledge Schema Inference Attacks in Text-to-SQL Systems | 2024-06-20T17:54:33 | 2024-06-20T17:54:33 | Đorđe Klisura, Anthony Rios | cs.CL | cs.CL | Relational databases are integral to modern information systems, serving as
the foundation for storing, querying, and managing data efficiently and
effectively. Advancements in large language modeling have led to the emergence
of text-to-SQL technologies, significantly enhancing the querying and
extracting of informati... | [
-0.1401407868,
-1.237604022,
0.6648259759,
-0.3037383854,
-0.1054423824,
0.0202107113,
0.1657869071,
-0.5334600806,
0.1263112873,
1.0245046616,
-0.5121329427,
-0.589240849,
-0.3051111102,
-0.1195356622,
-0.9627045989,
0.9544912577,
0.510496676,
-0.1950394958,
-0.470885843,
0.13... |
http://arxiv.org/abs/2406.14544v1 | http://arxiv.org/pdf/2406.14544v1 | Prism: A Framework for Decoupling and Assessing the Capabilities of VLMs | 2024-06-20T17:54:03 | 2024-06-20T17:54:03 | Yuxuan Qiao, Haodong Duan, Xinyu Fang, Junming Yang, Lin Chen, Songyang Zhang, Jiaqi Wang, Dahua Lin, Kai Chen | cs.CV | cs.CV, cs.CL | Vision Language Models (VLMs) demonstrate remarkable proficiency in
addressing a wide array of visual questions, which requires strong perception
and reasoning faculties. Assessing these two competencies independently is
crucial for model refinement, despite the inherent difficulty due to the
intertwined nature of seei... | [
-0.5633266568,
-0.5359130502,
0.5566396713,
0.4520348012,
-0.2664242983,
0.1281109899,
0.0367165692,
-0.6242189407,
-0.2708220184,
0.6534022093,
-0.4514026046,
-0.0396940224,
-0.2701174915,
0.0951257125,
-0.5143493414,
0.2221594453,
-0.061886806,
0.2984428704,
-0.4834390879,
-0... |
http://arxiv.org/abs/2406.14532v1 | http://arxiv.org/pdf/2406.14532v1 | RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math Reasoning by Eight-Fold | 2024-06-20T17:45:54 | 2024-06-20T17:45:54 | Amrith Setlur, Saurabh Garg, Xinyang Geng, Naman Garg, Virginia Smith, Aviral Kumar | cs.LG | cs.LG, cs.CL | Training on model-generated synthetic data is a promising approach for
finetuning LLMs, but it remains unclear when it helps or hurts. In this paper,
we investigate this question for math reasoning via an empirical study,
followed by building a conceptual understanding of our observations. First, we
find that while the... | [
-0.3957817852,
-0.8994311094,
0.5303574204,
0.0652082115,
-0.0914770737,
-0.0427649915,
-0.1258109957,
-0.2962869406,
-0.0098566171,
0.4018638432,
-0.6708365083,
-0.0844058096,
-0.4777904153,
0.1692996323,
-0.4015114307,
1.1626656055,
0.1317678243,
0.4832360446,
-0.2526940703,
... |
http://arxiv.org/abs/2406.14517v1 | http://arxiv.org/pdf/2406.14517v1 | PostMark: A Robust Blackbox Watermark for Large Language Models | 2024-06-20T17:27:14 | 2024-06-20T17:27:14 | Yapei Chang, Kalpesh Krishna, Amir Houmansadr, John Wieting, Mohit Iyyer | cs.LG | cs.LG, cs.AI, cs.CL, cs.CR | The most effective techniques to detect LLM-generated text rely on inserting
a detectable signature -- or watermark -- during the model's decoding process.
Most existing watermarking methods require access to the underlying LLM's
logits, which LLM API providers are loath to share due to fears of model
distillation. As ... | [
-0.166593641,
-0.896527946,
0.6802957654,
0.3287084401,
-0.8331777453,
0.0804546475,
0.132907182,
-0.3976993263,
0.106045939,
0.9110369682,
-0.2341537476,
-0.8169936538,
-0.9491047859,
-0.0261310805,
-0.6703933477,
0.7990868688,
-0.0859554484,
-0.2293722779,
-0.3310012221,
-0.2... |
http://arxiv.org/abs/2406.14511v1 | http://arxiv.org/pdf/2406.14511v1 | Investigating Mysteries of CoT-Augmented Distillation | 2024-06-20T17:15:46 | 2024-06-20T17:15:46 | Somin Wadhwa, Silvio Amir, Byron C. Wallace | cs.CL | cs.CL | Eliciting "chain of thought" (CoT) rationales -- sequences of token that
convey a "reasoning" process -- has been shown to consistently improve LLM
performance on tasks like question answering. More recent efforts have shown
that such rationales can also be used for model distillation: Including CoT
sequences (elicited... | [
-0.7570213675,
-0.7267580032,
0.8109852076,
-0.3082074523,
-0.6585268974,
0.4853331447,
-0.1135701761,
-0.4449217319,
-0.0674032569,
0.3718259931,
-0.3014905155,
-0.1380320638,
-0.7773672342,
0.272705704,
-0.7101175189,
1.1797581911,
0.0783156678,
0.3129128814,
0.0374668427,
0.... |
http://arxiv.org/abs/2406.14508v1 | http://arxiv.org/pdf/2406.14508v1 | Evidence of a log scaling law for political persuasion with large language models | 2024-06-20T17:12:38 | 2024-06-20T17:12:38 | Kobi Hackenburg, Ben M. Tappin, Paul Röttger, Scott Hale, Jonathan Bright, Helen Margetts | cs.CL | cs.CL, cs.AI, cs.CY, cs.HC | Large language models can now generate political messages as persuasive as
those written by humans, raising concerns about how far this persuasiveness may
continue to increase with model size. Here, we generate 720 persuasive messages
on 10 U.S. political issues from 24 language models spanning several orders of
magnit... | [
-0.0465905704,
-0.826302886,
0.7531260252,
0.2576625645,
-0.3152313828,
-0.1796668768,
-0.4653068781,
0.1223081276,
0.3345027864,
0.6468079686,
-0.3623337746,
-0.4755151868,
-0.5891946554,
-0.153154999,
-0.553352654,
1.1280300617,
0.547935605,
0.0356048085,
-0.2009623498,
0.236... |
http://arxiv.org/abs/2406.14504v1 | http://arxiv.org/pdf/2406.14504v1 | Translating Across Cultures: LLMs for Intralingual Cultural Adaptation | 2024-06-20T17:06:58 | 2024-06-20T17:06:58 | Pushpdeep Singh, Mayur Patidar, Lovekesh Vig | cs.CL | cs.CL | LLMs are increasingly being deployed for multilingual applications and have
demonstrated impressive translation capabilities between several low and high
resource languages. An aspect of translation that often gets overlooked is that
of cultural adaptation, or modifying source culture references to suit the
target cult... | [
-0.2646302879,
-0.0665003881,
0.3700321317,
0.5479305983,
-0.0944793373,
0.3995466828,
-0.5348998308,
-0.7805456519,
-0.0711416528,
0.3795765638,
-0.6267059445,
-0.4689175785,
-0.5738917589,
0.3249852955,
-0.0937717408,
0.7939697504,
-0.5137856603,
0.4900352359,
-0.4697178304,
... |
http://arxiv.org/abs/2406.14503v1 | http://arxiv.org/pdf/2406.14503v1 | Overview of the CAIL 2023 Argument Mining Track | 2024-06-20T17:06:13 | 2024-06-20T17:06:13 | Jingcong Liang, Junlong Wang, Xinyu Zhai, Yungui Zhuang, Yiyang Zheng, Xin Xu, Xiandong Ran, Xiaozheng Dong, Honghui Rong, Yanlun Liu, Hao Chen, Yuhan Wei, Donghai Li, Jiajie Peng, Xuanjing Huang, Chongde Shi, Yansong Feng, Yun Song, Zhongyu Wei | cs.CL | cs.CL | We give a detailed overview of the CAIL 2023 Argument Mining Track, one of
the Chinese AI and Law Challenge (CAIL) 2023 tracks. The main goal of the track
is to identify and extract interacting argument pairs in trial dialogs. It
mainly uses summarized judgment documents but can also refer to trial
recordings. The trac... | [
-0.3206846714,
-0.694848299,
0.6564654708,
0.2209648192,
-0.2433171719,
-0.1542426199,
0.2277664244,
-0.5279182196,
-0.2277339995,
0.7111725211,
-0.5248008966,
-0.5652172565,
-0.3854772151,
-0.3475972414,
-0.4766345322,
1.5644080639,
-0.1386252791,
0.4033729136,
-0.046379175,
-... |
http://arxiv.org/abs/2406.14500v1 | http://arxiv.org/pdf/2406.14500v1 | Improving Expert Radiology Report Summarization by Prompting Large Language Models with a Layperson Summary | 2024-06-20T17:01:55 | 2024-06-20T17:01:55 | Xingmeng Zhao, Tongnian Wang, Anthony Rios | cs.CL | cs.CL | Radiology report summarization (RRS) is crucial for patient care, requiring
concise "Impressions" from detailed "Findings." This paper introduces a novel
prompting strategy to enhance RRS by first generating a layperson summary. This
approach normalizes key observations and simplifies complex information using
non-expe... | [
-0.2626859844,
-0.2709698677,
0.6495258212,
0.0664748475,
-0.4754249156,
-0.1056898981,
-0.2608370185,
-0.4503058195,
0.2666321397,
0.4420461655,
-0.2452576309,
-0.4724617898,
-0.7436871529,
0.4304168224,
-0.1717743874,
1.0682002306,
0.1368656456,
0.3754402697,
-0.0121976295,
-... |
http://arxiv.org/abs/2406.14498v1 | http://arxiv.org/pdf/2406.14498v1 | LLaSA: Large Multimodal Agent for Human Activity Analysis Through Wearable Sensors | 2024-06-20T17:00:34 | 2024-06-20T17:00:34 | Sheikh Asif Imran, Mohammad Nur Hossain Khan, Subrata Biswas, Bashima Islam | cs.CL | cs.CL | Integrating inertial measurement units (IMUs) with large language models
(LLMs) advances multimodal AI by enhancing human activity understanding. We
introduce SensorCaps, a dataset of 26,288 IMU-derived activity narrations, and
OpenSQA, an instruction-following dataset with 257,562 question-answer pairs.
Combining LIMU... | [
0.1520767361,
-0.7036576867,
0.7744954228,
0.1329911947,
-0.1542935818,
0.3255427778,
0.0774345323,
-0.7210916877,
0.6012439728,
0.3084646761,
-0.7682960629,
-0.3183729053,
-0.5022959113,
-0.2294949293,
-0.3109695017,
0.6601032615,
-0.2187150717,
0.1477276236,
-0.3796522617,
-0... |
http://arxiv.org/abs/2406.14497v1 | http://arxiv.org/pdf/2406.14497v1 | CodeRAG-Bench: Can Retrieval Augment Code Generation? | 2024-06-20T16:59:52 | 2024-06-20T16:59:52 | Zora Zhiruo Wang, Akari Asai, Xinyan Velocity Yu, Frank F. Xu, Yiqing Xie, Graham Neubig, Daniel Fried | cs.SE | cs.SE, cs.CL | While language models (LMs) have proven remarkably adept at generating code,
many programs are challenging for LMs to generate using their parametric
knowledge alone. Providing external contexts such as library documentation can
facilitate generating accurate and functional code. Despite the success of
retrieval-augmen... | [
-0.3982562423,
-0.3506090641,
0.3615969419,
0.0335873552,
-0.1220335215,
0.0308616832,
-0.3835766912,
-0.5029317141,
-0.1589105725,
0.4173687696,
-0.14746584,
-0.0522864796,
-0.3334113657,
0.2183117867,
-0.7615166903,
1.0904538631,
0.0021062871,
0.1238878667,
-0.323869884,
-0.0... |
http://arxiv.org/abs/2406.14496v1 | http://arxiv.org/pdf/2406.14496v1 | African or European Swallow? Benchmarking Large Vision-Language Models for Fine-Grained Object Classification | 2024-06-20T16:59:39 | 2024-06-20T16:59:39 | Gregor Geigle, Radu Timofte, Goran Glavaš | cs.CV | cs.CV, cs.CL | Recent Large Vision-Language Models (LVLMs) demonstrate impressive abilities
on numerous image understanding and reasoning tasks. The task of fine-grained
object classification (e.g., distinction between \textit{animal species}),
however, has been probed insufficiently, despite its downstream importance. We
fill this e... | [
-0.5808039904,
-0.5703385472,
0.2005137056,
-0.0857441574,
-0.2433536798,
-0.2107439339,
-0.1874903888,
-0.990260303,
-0.0084288586,
0.4483483732,
-0.2491440624,
-0.3357458711,
-0.3860967159,
0.1677888781,
-0.3794786632,
0.4343608916,
-0.1808408648,
0.2594635189,
-0.2946866155,
... |
http://arxiv.org/abs/2406.14492v1 | http://arxiv.org/pdf/2406.14492v1 | Does Object Grounding Really Reduce Hallucination of Large Vision-Language Models? | 2024-06-20T16:56:11 | 2024-06-20T16:56:11 | Gregor Geigle, Radu Timofte, Goran Glavaš | cs.CV | cs.CV, cs.CL | Large vision-language models (LVLMs) have recently dramatically pushed the
state of the art in image captioning and many image understanding tasks (e.g.,
visual question answering). LVLMs, however, often \textit{hallucinate} and
produce captions that mention concepts that cannot be found in the image. These
hallucinati... | [
-0.4793384373,
-0.6702249646,
0.6206094027,
-0.1144624949,
-0.5369009972,
-0.4992612898,
-0.1981520504,
-0.5546724796,
-0.0958753377,
0.586673677,
-0.2316154689,
-0.6285219193,
-0.3518663049,
0.1037185341,
-0.4633166194,
0.6244410872,
0.0427817293,
0.1631742418,
-0.5863231421,
... |
http://arxiv.org/abs/2406.14491v1 | http://arxiv.org/pdf/2406.14491v1 | Instruction Pre-Training: Language Models are Supervised Multitask Learners | 2024-06-20T16:55:33 | 2024-06-20T16:55:33 | Daixuan Cheng, Yuxian Gu, Shaohan Huang, Junyu Bi, Minlie Huang, Furu Wei | cs.CL | cs.CL | Unsupervised multitask pre-training has been the critical method behind the
recent success of language models (LMs). However, supervised multitask learning
still holds significant promise, as scaling it in the post-training stage
trends towards better generalization. In this paper, we explore supervised
multitask pre-t... | [
-0.2676482499,
-0.5195580125,
0.6030635834,
0.2984903753,
-0.1719374806,
0.239773035,
-0.5429481268,
-0.2326011062,
0.1490617096,
0.6256184578,
-0.7390107512,
-0.5798043013,
-0.6544991732,
0.0682121813,
-0.3626525104,
1.1028017998,
0.0299016368,
0.1624295712,
-0.3209078312,
0.3... |
http://arxiv.org/abs/2406.14479v1 | http://arxiv.org/pdf/2406.14479v1 | On Layer-wise Representation Similarity: Application for Multi-Exit Models with a Single Classifier | 2024-06-20T16:41:09 | 2024-06-20T16:41:09 | Jiachen Jiang, Jinxin Zhou, Zhihui Zhu | cs.AI | cs.AI, cs.CL, cs.CV, cs.LG | Analyzing the similarity of internal representations within and across
different models has been an important technique for understanding the behavior
of deep neural networks. Most existing methods for analyzing the similarity
between representations of high dimensions, such as those based on Canonical
Correlation Anal... | [
-0.5074602962,
-0.1398407519,
0.5835838914,
-0.10183011,
-0.0848252773,
0.1243849918,
-0.2170965523,
-0.1988794357,
0.0748844892,
0.378644824,
-0.5183574557,
-0.454877466,
-0.4898479283,
-0.3922375143,
-0.552259326,
0.5027023554,
0.0010129318,
0.2292349637,
-0.3656940758,
-0.32... |
http://arxiv.org/abs/2406.14473v1 | http://arxiv.org/pdf/2406.14473v1 | Data-Centric AI in the Age of Large Language Models | 2024-06-20T16:34:07 | 2024-06-20T16:34:07 | Xinyi Xu, Zhaoxuan Wu, Rui Qiao, Arun Verma, Yao Shu, Jingtan Wang, Xinyuan Niu, Zhenfeng He, Jiangwei Chen, Zijian Zhou, Gregory Kang Ruey Lau, Hieu Dao, Lucas Agussurja, Rachael Hwee Ling Sim, Xiaoqiang Lin, Wenyang Hu, Zhongxiang Dai, Pang Wei Koh, Bryan Kian Hsiang Low | cs.LG | cs.LG, cs.CL | This position paper proposes a data-centric viewpoint of AI research,
focusing on large language models (LLMs). We start by making the key
observation that data is instrumental in the developmental (e.g., pretraining
and fine-tuning) and inferential stages (e.g., in-context learning) of LLMs,
and yet it receives dispro... | [
-0.3893717527,
-0.950443387,
0.6438502669,
-0.0243110377,
0.1865358949,
0.0434515625,
-0.4602331817,
-0.4704868197,
-0.027046252,
0.5117889047,
-0.361810565,
-0.466678679,
-0.5874335766,
-0.0283872876,
-0.6962249279,
0.8917846084,
0.0185938664,
0.1132917777,
-0.7674555182,
-0.0... |
http://arxiv.org/abs/2406.14464v1 | http://arxiv.org/pdf/2406.14464v1 | A Review of Common Online Speaker Diarization Methods | 2024-06-20T16:26:03 | 2024-06-20T16:26:03 | Roman Aperdannier, Sigurd Schacht, Alexander Piazza | cs.SD | cs.SD, cs.CL, eess.AS | Speaker diarization provides the answer to the question "who spoke when?" for
an audio file. This information can be used to complete audio transcripts for
further processing steps. Most speaker diarization systems assume that the
audio file is available as a whole. However, there are scenarios in which the
speaker lab... | [
-0.814971745,
-0.7951645851,
0.6192384958,
0.0938634425,
0.0751942471,
0.1480547935,
-0.2461653352,
-0.8749988675,
0.5321218371,
0.2928390205,
-0.513379395,
-0.1321599782,
-0.3864092529,
0.2076304108,
-0.4931854904,
0.6499242187,
0.3682336211,
0.1553783864,
-0.1504818201,
0.054... |
http://arxiv.org/abs/2406.14462v1 | http://arxiv.org/pdf/2406.14462v1 | Explicit and Implicit Large Language Model Personas Generate Opinions but Fail to Replicate Deeper Perceptions and Biases | 2024-06-20T16:24:07 | 2024-06-20T16:24:07 | Salvatore Giorgi, Tingting Liu, Ankit Aich, Kelsey Isman, Garrick Sherman, Zachary Fried, João Sedoc, Lyle H. Ungar, Brenda Curtis | cs.CL | cs.CL | Large language models (LLMs) are increasingly being used in human-centered
social scientific tasks, such as data annotation, synthetic data creation, and
engaging in dialog. However, these tasks are highly subjective and dependent on
human factors, such as one's environment, attitudes, beliefs, and lived
experiences. T... | [
-0.437574923,
-0.6710243225,
1.0550090075,
0.3425959647,
0.0487883911,
0.1776996255,
-0.3305619955,
-0.4319364727,
0.0023334117,
0.3863456249,
-0.5368804932,
-0.4173868895,
-0.3974941373,
0.2001765966,
-0.0439473428,
0.9645997882,
0.2219803929,
0.2217199206,
-0.295052141,
0.079... |
http://arxiv.org/abs/2406.14459v1 | http://arxiv.org/pdf/2406.14459v1 | Healing Powers of BERT: How Task-Specific Fine-Tuning Recovers Corrupted Language Models | 2024-06-20T16:18:04 | 2024-06-20T16:18:04 | Shijie Han, Zhenyu Zhang, Andrei Arsene Simion | cs.CL | cs.CL | Language models like BERT excel at sentence classification tasks due to
extensive pre-training on general data, but their robustness to parameter
corruption is unexplored. To understand this better, we look at what happens if
a language model is "broken", in the sense that some of its parameters are
corrupted and then ... | [
-0.1654001772,
-0.9104392529,
0.3512486517,
0.0083333729,
-0.1510783136,
-0.2011243552,
-0.3354183137,
-0.6294503808,
-0.0631892607,
1.0046838522,
-0.5034591556,
0.1078175604,
-0.8531867266,
-0.0676627159,
-0.499162972,
1.6736981869,
0.5845530629,
0.2085553408,
-0.0507412516,
-... |
http://arxiv.org/abs/2406.14434v1 | http://arxiv.org/pdf/2406.14434v1 | Towards Truthful Multilingual Large Language Models: Benchmarking and Alignment Strategies | 2024-06-20T15:59:07 | 2024-06-20T15:59:07 | Weihao Liu, Ning Wu, Wenbiao Ding, Shining Liang, Ming Gong, Dongmei Zhang | cs.CL | cs.CL | In the era of large language models (LLMs), building multilingual large
language models (MLLMs) that can serve users worldwide holds great
significance. However, existing research seldom focuses on the truthfulness of
MLLMs. Meanwhile, contemporary multilingual aligning technologies struggle to
balance massive language... | [
-0.2426051497,
-0.3608979285,
0.5331293941,
0.2637247443,
-0.0943004936,
0.3837548196,
-0.5474818349,
-0.5010556579,
-0.0394250751,
0.2087973058,
-0.4732411206,
-0.3508878052,
-0.3987360299,
0.1554526836,
-0.0134360706,
0.8488894701,
-0.0967782587,
0.4009893537,
-0.3878470957,
... |
http://arxiv.org/abs/2406.14425v2 | http://arxiv.org/pdf/2406.14425v2 | SynDARin: Synthesising Datasets for Automated Reasoning in Low-Resource Languages | 2024-06-20T15:49:28 | 2024-06-25T13:48:41 | Gayane Ghazaryan, Erik Arakelyan, Pasquale Minervini, Isabelle Augenstein | cs.CL | cs.CL, cs.AI, cs.LG | Question Answering (QA) datasets have been instrumental in developing and
evaluating Large Language Model (LLM) capabilities. However, such datasets are
scarce for languages other than English due to the cost and difficulties of
collection and manual annotation. This means that producing novel models and
measuring the ... | [
-0.4224584699,
-0.5803415179,
0.5462135077,
-0.0552563332,
-0.256383419,
0.0560338572,
-0.3490846157,
-0.075238511,
0.2291197926,
0.6286271214,
-0.6672593355,
-0.5994634032,
-0.1962889433,
0.6521607041,
-0.1761189103,
0.8548383117,
-0.2183985263,
0.2019540519,
-0.5851019621,
-0... |
http://arxiv.org/abs/2406.14408v2 | http://arxiv.org/pdf/2406.14408v2 | FVEL: Interactive Formal Verification Environment with Large Language Models via Theorem Proving | 2024-06-20T15:31:05 | 2024-06-21T02:51:41 | Xiaohan Lin, Qingxing Cao, Yinya Huang, Haiming Wang, Jianqiao Lu, Zhengying Liu, Linqi Song, Xiaodan Liang | cs.AI | cs.AI, cs.CL, cs.LG | Formal verification (FV) has witnessed growing significance with current
emerging program synthesis by the evolving large language models (LLMs).
However, current formal verification mainly resorts to symbolic verifiers or
hand-craft rules, resulting in limitations for extensive and flexible
verification. On the other ... | [
-0.2735084593,
-0.7858367562,
0.5759887695,
0.2666570544,
-0.0220395029,
0.0836618841,
-0.4172458947,
-0.5131527185,
-0.1554369777,
0.5980027914,
-0.43830809,
-0.6573216915,
-0.3498260379,
0.210991621,
-0.3425430059,
0.5067884326,
-0.5107522607,
-0.0914347619,
-0.2290051281,
0.... |
http://arxiv.org/abs/2406.14394v1 | http://arxiv.org/pdf/2406.14394v1 | SEC-QA: A Systematic Evaluation Corpus for Financial QA | 2024-06-20T15:12:41 | 2024-06-20T15:12:41 | Viet Dac Lai, Michael Krumdick, Charles Lovering, Varshini Reddy, Craig Schmidt, Chris Tanner | cs.CL | cs.CL | The financial domain frequently deals with large numbers of long documents
that are essential for daily operations. Significant effort is put towards
automating financial data analysis. However, a persistent challenge, not
limited to the finance domain, is the scarcity of datasets that accurately
reflect real-world tas... | [
-0.2276476026,
-0.6571093202,
0.4817710817,
0.2804053724,
-0.314576447,
0.6134259701,
-0.0360160284,
-0.4555546939,
0.1828801185,
0.4291362464,
-0.4459892809,
-0.5305046439,
-0.0478056893,
0.0960362628,
-0.6689232588,
0.879901886,
0.0509028845,
0.1806269139,
-0.2574430406,
-0.0... |
http://arxiv.org/abs/2406.14393v1 | http://arxiv.org/pdf/2406.14393v1 | Jailbreaking as a Reward Misspecification Problem | 2024-06-20T15:12:27 | 2024-06-20T15:12:27 | Zhihui Xie, Jiahui Gao, Lei Li, Zhenguo Li, Qi Liu, Lingpeng Kong | cs.LG | cs.LG, cs.CL | The widespread adoption of large language models (LLMs) has raised concerns
about their safety and reliability, particularly regarding their vulnerability
to adversarial attacks. In this paper, we propose a novel perspective that
attributes this vulnerability to reward misspecification during the alignment
process. We ... | [
-0.1247741729,
-0.4322532415,
0.5035387278,
0.2617559135,
0.1749400347,
0.4223007262,
-0.0709495246,
-0.2145933509,
0.5043019056,
0.3399965763,
-0.9383507967,
-0.4029030502,
-0.8186078668,
0.068498224,
-0.2416554242,
1.1317743063,
-0.2898983359,
-0.4745911658,
0.0449032448,
-0.... |
http://arxiv.org/abs/2406.14373v2 | http://arxiv.org/pdf/2406.14373v2 | Artificial Leviathan: Exploring Social Evolution of LLM Agents Through the Lens of Hobbesian Social Contract Theory | 2024-06-20T14:42:58 | 2024-07-01T22:06:13 | Gordon Dai, Weijia Zhang, Jinhan Li, Siqi Yang, Chidera Onochie lbe, Srihas Rao, Arthur Caetano, Misha Sra | cs.AI | cs.AI, cs.CL, cs.CY, cs.HC, cs.MA | The emergence of Large Language Models (LLMs) and advancements in Artificial
Intelligence (AI) offer an opportunity for computational social science
research at scale. Building upon prior explorations of LLM agent design, our
work introduces a simulated agent society where complex social relationships
dynamically form ... | [
-0.495958358,
-0.8892123103,
0.5320648551,
-0.1480183154,
-0.0325357765,
0.7882114649,
-0.0055043059,
-0.4268830121,
0.6796811223,
0.5278712511,
-0.3246999979,
0.0034146898,
-0.3374805152,
0.0781137198,
-0.2743085027,
0.9655722976,
0.2846416533,
-0.1989874542,
-0.2199027687,
-0... |
http://arxiv.org/abs/2406.16953v1 | http://arxiv.org/pdf/2406.16953v1 | Validation of a new, minimally-invasive, software smartphone device to predict sleep apnea and its severity: transversal study | 2024-06-20T14:36:15 | 2024-06-20T14:36:15 | Justine Frija, Juliette Millet, Emilie Bequignon, Ala Covali, Guillaume Cathelain, Josselin Houenou, Helene Benzaquen, Pierre Alexis Geoffroy, Emmanuel Bacry, Mathieu Grajoszex, Marie-Pia d Ortho | eess.SP | eess.SP, cs.CL | Obstructive sleep apnea (OSA) is frequent and responsible for cardiovascular
complications and excessive daytime sleepiness. It is underdiagnosed due to the
difficulty to access the gold standard for diagnosis, polysomnography (PSG).
Alternative methods using smartphone sensors could be useful to increase
diagnosis. Th... | [
-0.3977591097,
-0.455521971,
0.5046805143,
-0.1353164017,
-0.5093412399,
-0.2057684809,
-0.0045451568,
-0.5315634608,
0.5634877682,
0.13453269,
-0.8459163904,
-0.4930886328,
-0.4889946282,
-0.0114204269,
-0.1555864364,
0.5716177821,
-0.0762613565,
0.1777439862,
0.115492247,
-0.... |
http://arxiv.org/abs/2406.14358v1 | http://arxiv.org/pdf/2406.14358v1 | The neural correlates of logical-mathematical symbol systems processing resemble that of spatial cognition more than natural language processing | 2024-06-20T14:31:09 | 2024-06-20T14:31:09 | Yuannan Li, Shan Xu, Jia Liu | q-bio.NC | q-bio.NC, cs.AI, cs.CL | The ability to manipulate logical-mathematical symbols (LMS), encompassing
tasks such as calculation, reasoning, and programming, is a cognitive skill
arguably unique to humans. Considering the relatively recent emergence of this
ability in human evolutionary history, it has been suggested that LMS
processing may build... | [
-0.4800181985,
-0.9065832496,
0.7662774324,
0.1391373128,
0.2858374417,
0.0541753545,
-0.1839337349,
-0.682249248,
0.2892415822,
0.3197457194,
-0.6355392933,
-0.4154667854,
-0.5611774325,
0.1534656286,
-0.199540779,
0.4367360771,
0.0961418152,
0.0682342499,
-0.4159179926,
-0.32... |
http://arxiv.org/abs/2406.14336v1 | http://arxiv.org/pdf/2406.14336v1 | Exploring Spatial Representations in the Historical Lake District Texts with LLM-based Relation Extraction | 2024-06-20T14:04:59 | 2024-06-20T14:04:59 | Erum Haris, Anthony G. Cohn, John G. Stell | cs.CL | cs.CL, cs.AI | Navigating historical narratives poses a challenge in unveiling the spatial
intricacies of past landscapes. The proposed work addresses this challenge
within the context of the English Lake District, employing the Corpus of the
Lake District Writing. The method utilizes a generative pre-trained transformer
model to ext... | [
-0.5838069916,
-0.6509466171,
1.1215019226,
-0.1633659154,
0.0371132679,
-0.2428543717,
0.025805369,
-0.5698929429,
0.2885213196,
0.7527381182,
-0.7091315389,
-0.7139335871,
-0.585760951,
-0.1453204006,
-0.159924522,
0.5088813901,
-0.4132929444,
-0.0812592357,
0.0141704911,
-0.... |
http://arxiv.org/abs/2406.14335v1 | http://arxiv.org/pdf/2406.14335v1 | Self-supervised Interpretable Concept-based Models for Text Classification | 2024-06-20T14:04:53 | 2024-06-20T14:04:53 | Francesco De Santis, Philippe Bich, Gabriele Ciravegna, Pietro Barbiero, Danilo Giordano, Tania Cerquitelli | cs.CL | cs.CL, cs.AI | Despite their success, Large-Language Models (LLMs) still face criticism as
their lack of interpretability limits their controllability and reliability.
Traditional post-hoc interpretation methods, based on attention and
gradient-based analysis, offer limited insight into the model's decision-making
processes. In the i... | [
-0.0930077583,
-0.5424477458,
0.7045225501,
0.0565667525,
-0.50612396,
-0.0255696084,
-0.0240952019,
-0.6136721373,
0.2328970581,
0.8107245564,
-0.4581635892,
-0.7386092544,
-0.629743576,
0.0846428648,
-0.5792286396,
1.0686224699,
-0.3351197541,
-0.2785455585,
-0.4647252858,
0.... |
http://arxiv.org/abs/2406.14326v1 | http://arxiv.org/pdf/2406.14326v1 | medIKAL: Integrating Knowledge Graphs as Assistants of LLMs for Enhanced Clinical Diagnosis on EMRs | 2024-06-20T13:56:52 | 2024-06-20T13:56:52 | Mingyi Jia, Junwen Duan, Yan Song, Jianxin Wang | cs.CL | cs.CL | Electronic Medical Records (EMRs), while integral to modern healthcare,
present challenges for clinical reasoning and diagnosis due to their complexity
and information redundancy. To address this, we proposed medIKAL (Integrating
Knowledge Graphs as Assistants of LLMs), a framework that combines Large
Language Models (... | [
-0.3131534755,
-0.6383902431,
0.7326358557,
-0.1161354184,
-0.4706255496,
0.0006387245,
-0.0940383375,
-0.5381352901,
0.518601954,
0.3160037398,
-0.4196163118,
-0.8730064631,
-0.5686277151,
0.191049546,
-0.188235417,
0.9689808488,
-0.0748766735,
0.4317232668,
-0.386646539,
-0.1... |
http://arxiv.org/abs/2406.14322v2 | http://arxiv.org/pdf/2406.14322v2 | Mind the Privacy Unit! User-Level Differential Privacy for Language Model Fine-Tuning | 2024-06-20T13:54:32 | 2024-07-03T14:05:20 | Lynn Chua, Badih Ghazi, Yangsibo Huang, Pritish Kamath, Ravi Kumar, Daogao Liu, Pasin Manurangsi, Amer Sinha, Chiyuan Zhang | cs.CL | cs.CL, cs.CR, cs.LG | Large language models (LLMs) have emerged as powerful tools for tackling
complex tasks across diverse domains, but they also raise privacy concerns when
fine-tuned on sensitive data due to potential memorization. While differential
privacy (DP) offers a promising solution by ensuring models are 'almost
indistinguishabl... | [
-0.4448276162,
-0.660051167,
0.4833098948,
0.1995059252,
-0.1144546568,
0.2411993742,
-0.2132522017,
-0.4209703505,
-0.2930231094,
0.6421937346,
-0.5906703472,
-0.7583540678,
-0.3179380894,
0.0836388692,
-0.2910038531,
0.8839023113,
0.2512460053,
-0.2120859623,
-0.0875012204,
0... |
http://arxiv.org/abs/2406.14319v1 | http://arxiv.org/pdf/2406.14319v1 | LiveMind: Low-latency Large Language Models with Simultaneous Inference | 2024-06-20T13:52:30 | 2024-06-20T13:52:30 | Chuangtao Chen, Grace Li Zhang, Xunzhao Yin, Cheng Zhuo, Ulf Schlichtmann, Bing Li | cs.AI | cs.AI, cs.CL | In this paper, we introduce a novel low-latency inference framework for large
language models (LLMs) inference which enables LLMs to perform inferences with
incomplete prompts. By reallocating computational processes to prompt input
phase, we achieve a substantial reduction in latency, thereby significantly
enhancing t... | [
-0.2574795485,
-1.1709250212,
0.8665428162,
0.2893731296,
-0.1580431908,
0.1458546519,
-0.5022808909,
-0.3319764733,
0.4522789419,
0.3460126817,
-1.0617568493,
-0.3328184485,
-0.229469955,
-0.1254591346,
-0.0078348257,
0.3063944578,
0.1911522597,
0.1637526453,
-0.4284826815,
-0... |
http://arxiv.org/abs/2406.14318v1 | http://arxiv.org/pdf/2406.14318v1 | The Fire Thief Is Also the Keeper: Balancing Usability and Privacy in Prompts | 2024-06-20T13:52:25 | 2024-06-20T13:52:25 | Zhili Shen, Zihang Xi, Ying He, Wei Tong, Jingyu Hua, Sheng Zhong | cs.CR | cs.CR, cs.AI, cs.CL | The rapid adoption of online chatbots represents a significant advancement in
artificial intelligence. However, this convenience brings considerable privacy
concerns, as prompts can inadvertently contain sensitive information exposed to
large language models (LLMs). Limited by high computational costs, reduced task
usa... | [
-0.1244622618,
-0.8010131717,
0.4532234967,
0.2841273248,
-0.226145193,
0.4182177782,
0.1354787797,
-0.4673057497,
0.0837944224,
0.6630336642,
-0.7006735802,
-0.2696036696,
-0.2848324478,
0.3706990182,
-0.038741041,
0.9125946164,
0.3869673014,
-0.2542567551,
-0.1344407797,
0.25... |
http://arxiv.org/abs/2406.14314v2 | http://arxiv.org/pdf/2406.14314v2 | Identifying User Goals from UI Trajectories | 2024-06-20T13:46:10 | 2024-06-30T12:33:48 | Omri Berkovitch, Sapir Caduri, Noam Kahlon, Anatoly Efros, Avi Caciularu, Ido Dagan | cs.CL | cs.CL, cs.AI | Autonomous agents that interact with graphical user interfaces (GUIs) hold
significant potential for enhancing user experiences. To further improve these
experiences, agents need to be personalized and proactive. By effectively
comprehending user intentions through their actions and interactions with GUIs,
agents will ... | [
-0.1874296665,
-0.6508929133,
1.0377454758,
-0.1108596921,
0.0289494619,
0.0391730182,
0.094919391,
-0.0369419381,
0.2557512224,
0.600432992,
-0.8051232696,
-0.59471035,
-0.4018592536,
-0.0734961107,
-0.4466733336,
1.1653279066,
-0.2884034216,
-0.6893519759,
-0.570299983,
0.066... |
http://arxiv.org/abs/2406.14313v1 | http://arxiv.org/pdf/2406.14313v1 | Robust Few-shot Transfer Learning for Knowledge Base Question Answering with Unanswerable Questions | 2024-06-20T13:43:38 | 2024-06-20T13:43:38 | Riya Sawhney, Indrajit Bhattacharya, Mausam | cs.CL | cs.CL, cs.AI | Real-world KBQA applications require models that are (1) robust -- e.g., can
differentiate between answerable and unanswerable questions, and (2)
low-resource -- do not require large training data. Towards this goal, we
propose the novel task of few-shot transfer for KBQA with unanswerable
questions. We present FUn-FuS... | [
-0.1893886626,
-0.9827728271,
0.7905775309,
0.002007171,
-0.0287649166,
-0.0949618518,
-0.0434389934,
-0.3568395972,
0.101711534,
0.5431625247,
-1.0479917526,
-0.2154627293,
-0.467464745,
0.0249728113,
-0.4330100715,
1.0196698904,
-0.0591123216,
-0.2059222311,
-0.4024045765,
-0... |
http://arxiv.org/abs/2406.14312v1 | http://arxiv.org/pdf/2406.14312v1 | Infusing clinical knowledge into tokenisers for language models | 2024-06-20T13:43:03 | 2024-06-20T13:43:03 | Abul Hasan, Jinge Wu, Quang Ngoc Nguyen, Salomé Andres, Imane Guellil, Huayu Zhang, Arlene Casey, Beatrice Alex, Bruce Guthrie, Honghan Wu | cs.CL | cs.CL, cs.AI | This study introduces a novel knowledge enhanced tokenisation mechanism,
K-Tokeniser, for clinical text processing. Technically, at initialisation
stage, K-Tokeniser populates global representations of tokens based on semantic
types of domain concepts (such as drugs or diseases) from either a domain
ontology like Unifi... | [
0.1834060103,
-0.4211935103,
0.7358756065,
-0.0986540467,
-0.4060885012,
0.0658956021,
-0.1390879899,
-0.4549944997,
0.3870883882,
0.5683583021,
-0.0066512921,
-0.6475142241,
-0.8308807611,
0.0447471552,
-0.1724342406,
0.9724878073,
-0.0109503064,
0.6783195734,
-0.5134696364,
-... |
http://arxiv.org/abs/2406.14307v2 | http://arxiv.org/pdf/2406.14307v2 | QuST-LLM: Integrating Large Language Models for Comprehensive Spatial Transcriptomics Analysis | 2024-06-20T13:37:10 | 2024-07-01T19:38:03 | Chao Hui Huang | q-bio.GN | q-bio.GN, cs.CL, cs.CV | In this paper, we introduce QuST-LLM, an innovative extension of QuPath that
utilizes the capabilities of large language models (LLMs) to analyze and
interpret spatial transcriptomics (ST) data. In addition to simplifying the
intricate and high-dimensional nature of ST data by offering a comprehensive
workflow that inc... | [
-0.0479899794,
-1.0170456171,
1.1112226248,
0.0321353637,
-0.3230396211,
0.3574815989,
-0.1278108805,
-0.4507501423,
0.2694895566,
0.715847075,
-0.4465072751,
-1.0643565655,
-0.4715580344,
-0.031112574,
-0.6253515482,
0.681848824,
0.0669677854,
0.5807419419,
-0.2060740888,
-0.3... |
http://arxiv.org/abs/2406.14284v1 | http://arxiv.org/pdf/2406.14284v1 | VAIYAKARANA : A Benchmark for Automatic Grammar Correction in Bangla | 2024-06-20T13:09:29 | 2024-06-20T13:09:29 | Pramit Bhattacharyya, Arnab Bhattacharya | cs.CL | cs.CL, cs.AI | Bangla (Bengali) is the fifth most spoken language globally and, yet, the
problem of automatic grammar correction in Bangla is still in its nascent
stage. This is mostly due to the need for a large corpus of grammatically
incorrect sentences, with their corresponding correct counterparts. The present
state-of-the-art t... | [
0.3361803889,
-1.0748460293,
0.2002447098,
0.6466953158,
-0.1309252679,
-0.0721134096,
-0.1424242705,
-0.0743572265,
0.1687996089,
0.1566791087,
-0.1020908654,
-0.3428381085,
-0.1557015479,
0.4429542422,
-0.5996053815,
0.4721781313,
0.010557726,
0.1076452732,
-0.2676190436,
0.1... |
http://arxiv.org/abs/2406.14282v1 | http://arxiv.org/pdf/2406.14282v1 | Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs | 2024-06-20T13:07:38 | 2024-06-20T13:07:38 | Junjie Wang, Mingyang Chen, Binbin Hu, Dan Yang, Ziqi Liu, Yue Shen, Peng Wei, Zhiqiang Zhang, Jinjie Gu, Jun Zhou, Jeff Z. Pan, Wen Zhang, Huajun Chen | cs.CL | cs.CL, cs.AI | Improving the performance of large language models (LLMs) in complex
question-answering (QA) scenarios has always been a research focal point.
Recent studies have attempted to enhance LLMs' performance by combining
step-wise planning with external retrieval. While effective for advanced models
like GPT-3.5, smaller LLM... | [
-0.4101662636,
-0.5946242809,
0.7655358315,
-0.1510868371,
-0.0461574085,
0.2142393291,
-0.0246810336,
-0.2590997517,
0.1915559918,
0.6105570793,
-0.4755730331,
-0.4649886489,
-0.3438251019,
0.0173654761,
-0.2460372746,
0.472784102,
-0.1606326103,
0.0148317451,
-0.1269134283,
0... |
http://arxiv.org/abs/2406.14277v1 | http://arxiv.org/pdf/2406.14277v1 | Augmenting Query and Passage for Retrieval-Augmented Generation using LLMs for Open-Domain Question Answering | 2024-06-20T12:59:27 | 2024-06-20T12:59:27 | Minsang Kim, Cheoneum Park, Seungjun Baek | cs.CL | cs.CL, cs.AI | Retrieval-augmented generation (RAG) has received much attention for
Open-domain question-answering (ODQA) tasks as a means to compensate for the
parametric knowledge of large language models (LLMs). While previous approaches
focused on processing retrieved passages to remove irrelevant context, they
still rely heavily... | [
-0.7048559785,
-1.058036685,
0.543815434,
-0.1195786893,
-0.5019230843,
0.1530124247,
0.0134282997,
-0.4970546365,
0.3369935155,
0.8964565396,
-0.6336376667,
-0.1361422241,
-0.0989326909,
0.2641019523,
0.0662160516,
0.8406024575,
-0.1267004013,
-0.0534528159,
-0.4724907279,
-0.... |
http://arxiv.org/abs/2406.14275v1 | http://arxiv.org/pdf/2406.14275v1 | Step-Back Profiling: Distilling User History for Personalized Scientific Writing | 2024-06-20T12:58:26 | 2024-06-20T12:58:26 | Xiangru Tang, Xingyao Zhang, Yanjun Shao, Jie Wu, Yilun Zhao, Arman Cohan, Ming Gong, Dongmei Zhang, Mark Gerstein | cs.CL | cs.CL, cs.AI | Large language models (LLMs) excel at a variety of natural language
processing tasks, yet they struggle to generate personalized content for
individuals, particularly in real-world scenarios like scientific writing.
Addressing this challenge, we introduce Step-Back Profiling to personalize LLMs
by distilling user histo... | [
-0.2542947829,
-0.5098275542,
0.6712508798,
0.2802206278,
0.2943074703,
0.575699091,
-0.2556807995,
-0.3493479788,
0.1871092916,
0.8220821619,
-0.6964952946,
-0.8671668768,
-0.420170337,
0.4652850032,
-0.6236346364,
1.3953728676,
-0.0512633584,
-0.1377838552,
-0.4670694768,
-0.... |
http://arxiv.org/abs/2406.14267v1 | http://arxiv.org/pdf/2406.14267v1 | On the Evaluation Practices in Multilingual NLP: Can Machine Translation Offer an Alternative to Human Translations? | 2024-06-20T12:46:12 | 2024-06-20T12:46:12 | Rochelle Choenni, Sara Rajaee, Christof Monz, Ekaterina Shutova | cs.CL | cs.CL, cs.AI | While multilingual language models (MLMs) have been trained on 100+
languages, they are typically only evaluated across a handful of them due to a
lack of available test data in most languages. This is particularly problematic
when assessing MLM's potential for low-resource and unseen languages. In this
paper, we prese... | [
-0.3437885344,
-0.3722350895,
0.2597109973,
0.5314450264,
0.103754282,
0.0005929613,
-0.7237673998,
-0.4399840236,
-0.1500780284,
0.4935693741,
-0.6305904984,
-0.0706727058,
-0.618388474,
0.5803050995,
-0.1880073398,
1.1904617548,
-0.2160298377,
0.5788353682,
-0.229009375,
-0.5... |
http://arxiv.org/abs/2406.14230v1 | http://arxiv.org/pdf/2406.14230v1 | Raising the Bar: Investigating the Values of Large Language Models via Generative Evolving Testing | 2024-06-20T11:51:00 | 2024-06-20T11:51:00 | Han Jiang, Xiaoyuan Yi, Zhihua Wei, Shu Wang, Xing Xie | cs.CL | cs.CL, cs.AI, cs.CY | Warning: this paper contains model outputs exhibiting unethical information.
Large Language Models (LLMs) have achieved significant breakthroughs, but their
generated unethical content poses potential risks. Measuring value alignment of
LLMs becomes crucial for their regulation and responsible deployment. Numerous
data... | [
-0.0434831642,
-0.7913994193,
0.5378662944,
0.2471363097,
-0.2063671052,
0.2775978148,
-0.18577075,
-0.5305821896,
-0.1317105442,
0.360509038,
-0.5779120326,
-0.1849492639,
-0.6125989556,
-0.0479336381,
-0.2229484469,
1.0680546761,
0.067620337,
-0.0656450689,
-0.0373047031,
-0.... |
http://arxiv.org/abs/2406.14213v1 | http://arxiv.org/pdf/2406.14213v1 | Complexity of Symbolic Representation in Working Memory of Transformer Correlates with the Complexity of a Task | 2024-06-20T11:27:29 | 2024-06-20T11:27:29 | Alsu Sagirova, Mikhail Burtsev | cs.CL | cs.CL | Even though Transformers are extensively used for Natural Language Processing
tasks, especially for machine translation, they lack an explicit memory to
store key concepts of processed texts. This paper explores the properties of
the content of symbolic working memory added to the Transformer model decoder.
Such workin... | [
-0.1399815977,
-0.2756466568,
0.4022354782,
0.0733853653,
-0.2923889756,
0.0480885804,
-0.3326349258,
-0.4987370074,
-0.0209769774,
0.4431398213,
-0.2717044652,
-0.1737586558,
-1.095238924,
0.3607277572,
-0.5575418472,
0.8170450926,
-0.2646044791,
0.0092616184,
-0.1157987118,
-... |
http://arxiv.org/abs/2406.14197v1 | http://arxiv.org/pdf/2406.14197v1 | On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning | 2024-06-20T10:59:02 | 2024-06-20T10:59:02 | Franz Nowak, Anej Svete, Alexandra Butoi, Ryan Cotterell | cs.CL | cs.CL, cs.FL | The performance of modern language models (LMs) has been improved by
chain-of-thought (CoT) reasoning, i.e., the process of generating intermediate
results that guide the model towards a final answer. A possible explanation for
this improvement is that CoT reasoning extends an LM's computational power, as
RNNs and tran... | [
-0.42178756,
-0.4563975632,
0.9719401002,
-0.0247412957,
-0.2549151778,
0.3785798848,
-0.5055178404,
-0.4959956706,
0.1782195568,
0.703219831,
-0.25129655,
-0.1233542785,
-0.6423253417,
0.0563047454,
-0.5838424563,
0.7320077419,
-0.2528676689,
-0.0492957756,
-0.2760832012,
-0.1... |
http://arxiv.org/abs/2406.14192v1 | http://arxiv.org/pdf/2406.14192v1 | Timo: Towards Better Temporal Reasoning for Language Models | 2024-06-20T10:52:14 | 2024-06-20T10:52:14 | Zhaochen Su, Jun Zhang, Tong Zhu, Xiaoye Qu, Juntao Li, Min Zhang, Yu Cheng | cs.CL | cs.CL, cs.AI | Reasoning about time is essential for Large Language Models (LLMs) to
understand the world. Previous works focus on solving specific tasks, primarily
on time-sensitive question answering. While these methods have proven
effective, they cannot generalize to a wider spectrum of temporal reasoning
tasks. Therefore, we pro... | [
-0.1281760931,
-0.7189614773,
0.9278073907,
0.6111099124,
-0.2282586843,
0.1202748343,
-0.1768467873,
-0.752728045,
-0.2806167006,
0.3471491039,
-0.7205686569,
-0.3482840657,
-0.3231737912,
0.2118802965,
-0.1510401517,
0.7139396071,
0.1049493924,
-0.1047538817,
0.1332539171,
-0... |
http://arxiv.org/abs/2406.14191v1 | http://arxiv.org/pdf/2406.14191v1 | Temporal Knowledge Graph Question Answering: A Survey | 2024-06-20T10:51:06 | 2024-06-20T10:51:06 | Miao Su, ZiXuan Li, Zhuo Chen, Long Bai, Xiaolong Jin, Jiafeng Guo | cs.CL | cs.CL, cs.AI, cs.LG | Knowledge Base Question Answering (KBQA) has been a long-standing field to
answer questions based on knowledge bases. Recently, the evolving dynamics of
knowledge have attracted a growing interest in Temporal Knowledge Graph
Question Answering (TKGQA), an emerging task to answer temporal questions.
However, this field ... | [
-0.3033560812,
-1.1436519623,
0.8419474959,
0.1365022063,
-0.1946672946,
0.775834918,
0.3301144242,
-0.5528678298,
0.0421429314,
0.3366752863,
-0.6017523408,
-0.6084493399,
-0.5062968731,
-0.1879703552,
-0.041471079,
0.229118064,
-0.5326990485,
-0.0852297843,
-0.3481041789,
-0.... |
http://arxiv.org/abs/2406.14189v1 | http://arxiv.org/pdf/2406.14189v1 | In Tree Structure Should Sentence Be Generated | 2024-06-20T10:49:08 | 2024-06-20T10:49:08 | Yaguang Li, Xin Chen | cs.CL | cs.CL | Generative models reliant on sequential autoregression have been at the
forefront of language generation for an extensive period, particularly
following the introduction of widely acclaimed transformers. Despite its
excellent performance, there are always some issues that we face today. For
example, problems such as ha... | [
0.0089009479,
-1.0905526876,
0.3905089498,
0.2282384932,
-0.3076426685,
-0.02070277,
0.1658811271,
-0.4213396013,
-0.0305266418,
0.3010193408,
-0.4261421263,
-0.2624023855,
-0.570189476,
0.2200266868,
-0.9141149521,
0.9779285789,
0.225098595,
-0.409196794,
0.1998092681,
-0.0593... |
http://arxiv.org/abs/2406.14177v1 | http://arxiv.org/pdf/2406.14177v1 | SimulSeamless: FBK at IWSLT 2024 Simultaneous Speech Translation | 2024-06-20T10:34:46 | 2024-06-20T10:34:46 | Sara Papi, Marco Gaido, Matteo Negri, Luisa Bentivogli | cs.CL | cs.CL, cs.AI, cs.SD, eess.AS | This paper describes the FBK's participation in the Simultaneous Translation
Evaluation Campaign at IWSLT 2024. For this year's submission in the
speech-to-text translation (ST) sub-track, we propose SimulSeamless, which is
realized by combining AlignAtt and SeamlessM4T in its medium configuration. The
SeamlessM4T mode... | [
-0.2105159163,
-0.4719021618,
0.5643483996,
0.7904845476,
-0.6236496568,
0.3792219758,
-0.2521212995,
-0.5782958865,
0.141282931,
0.2449102849,
-0.9704465866,
-0.4088955224,
-0.5431078672,
0.1042003036,
-0.5811092854,
0.9632830024,
-0.7998139858,
0.5249676108,
-0.6189755797,
-0... |
http://arxiv.org/abs/2406.14171v1 | http://arxiv.org/pdf/2406.14171v1 | Ranking LLMs by compression | 2024-06-20T10:23:38 | 2024-06-20T10:23:38 | Peijia Guo, Ziguang Li, Haibo Hu, Chao Huang, Ming Li, Rui Zhang | cs.AI | cs.AI, cs.CL | We conceptualize the process of understanding as information compression, and
propose a method for ranking large language models (LLMs) based on lossless
data compression. We demonstrate the equivalence of compression length under
arithmetic coding with cumulative negative log probabilities when using a large
language ... | [
-0.428471148,
-0.5677004457,
0.727065444,
-0.0069279969,
-0.4176865816,
0.0028588569,
-0.6558853984,
-0.4581429362,
0.1346160173,
0.5077688694,
-0.3547556698,
-0.4535606205,
-0.8623383045,
0.2711846232,
-0.3976426721,
0.7088348866,
0.2418880165,
0.219515726,
-0.4357590377,
0.06... |
http://arxiv.org/abs/2406.14167v1 | http://arxiv.org/pdf/2406.14167v1 | Definition generation for lexical semantic change detection | 2024-06-20T10:13:08 | 2024-06-20T10:13:08 | Mariia Fedorova, Andrey Kutuzov, Yves Scherrer | cs.CL | cs.CL | We use contextualized word definitions generated by large language models as
semantic representations in the task of diachronic lexical semantic change
detection (LSCD). In short, generated definitions are used as `senses', and the
change score of a target word is retrieved by comparing their distributions in
two time ... | [
-0.5146170259,
-0.5462408662,
0.5611825585,
0.0620940514,
-0.4283154309,
-0.3850653172,
-0.3000785708,
-0.5008679628,
0.16177091,
0.3350934684,
-0.6237328649,
-0.5876218081,
-0.53753829,
0.3619852364,
-0.7944275141,
0.8336194754,
-0.3072615862,
-0.3755064905,
0.0391564146,
-0.1... |
http://arxiv.org/abs/2406.14164v1 | http://arxiv.org/pdf/2406.14164v1 | A Data-Driven Guided Decoding Mechanism for Diagnostic Captioning | 2024-06-20T10:08:17 | 2024-06-20T10:08:17 | Panagiotis Kaliosis, John Pavlopoulos, Foivos Charalampakos, Georgios Moschovis, Ion Androutsopoulos | cs.AI | cs.AI, cs.CL | Diagnostic Captioning (DC) automatically generates a diagnostic text from one
or more medical images (e.g., X-rays, MRIs) of a patient. Treated as a draft,
the generated text may assist clinicians, by providing an initial estimation of
the patient's condition, speeding up and helping safeguard the diagnostic
process. T... | [
0.0928488895,
-0.4648218453,
0.8544707298,
0.0308003351,
-0.9406549931,
0.1690438688,
0.2493557185,
0.0953538939,
0.0738512129,
0.4094042182,
-0.7823365927,
-1.0763715506,
-0.5958405137,
0.513782382,
-0.4485851228,
0.9132415652,
-0.1482248157,
0.4532574117,
-0.3793325424,
0.130... |
http://arxiv.org/abs/2406.14162v1 | http://arxiv.org/pdf/2406.14162v1 | DIRAS: Efficient LLM-Assisted Annotation of Document Relevance in Retrieval Augmented Generation | 2024-06-20T10:04:09 | 2024-06-20T10:04:09 | Jingwei Ni, Tobias Schimanski, Meihong Lin, Mrinmaya Sachan, Elliott Ash, Markus Leippold | cs.IR | cs.IR, cs.AI, cs.CL | Retrieval Augmented Generation (RAG) is widely employed to ground responses
to queries on domain-specific documents. But do RAG implementations leave out
important information or excessively include irrelevant information? To allay
these concerns, it is necessary to annotate domain-specific benchmarks to
evaluate infor... | [
-0.6018494964,
-0.4941906929,
0.5158225298,
0.0262197461,
-0.4724750817,
-0.0298633892,
0.0200019889,
-0.4116773903,
0.4405029416,
0.5417011976,
-0.3454548419,
-0.3340333402,
-0.5487405658,
0.0960375145,
-0.2962597311,
1.0488584042,
0.0910843536,
-0.1229780689,
-0.3038504422,
-... |
http://arxiv.org/abs/2406.14155v1 | http://arxiv.org/pdf/2406.14155v1 | Aligning Large Language Models with Diverse Political Viewpoints | 2024-06-20T09:53:23 | 2024-06-20T09:53:23 | Dominik Stammbach, Philine Widmer, Eunjung Cho, Caglar Gulcehre, Elliott Ash | cs.CL | cs.CL | Large language models such as ChatGPT often exhibit striking political
biases. If users query them about political information, they might take a
normative stance and reinforce such biases. To overcome this, we align LLMs
with diverse political viewpoints from 100,000 comments written by candidates
running for national... | [
-0.5047848821,
-0.4006489515,
0.6407297254,
0.027828414,
-0.4891405404,
0.1485122293,
-0.3660449982,
-0.1910338104,
0.2943909466,
0.85391289,
-0.6578546762,
-0.2953482866,
-0.6786344051,
-0.1571792811,
-0.2518491447,
1.0488357544,
0.3036475778,
0.2526390254,
-0.057439886,
-0.23... |
http://arxiv.org/abs/2406.14154v1 | http://arxiv.org/pdf/2406.14154v1 | Watching the Watchers: A Comparative Fairness Audit of Cloud-based Content Moderation Services | 2024-06-20T09:52:10 | 2024-06-20T09:52:10 | David Hartmann, Amin Oueslati, Dimitri Staufer | cs.CY | cs.CY, cs.CL, cs.HC, cs.LG | Online platforms face the challenge of moderating an ever-increasing volume
of content, including harmful hate speech. In the absence of clear legal
definitions and a lack of transparency regarding the role of algorithms in
shaping decisions on content moderation, there is a critical need for external
accountability. O... | [
-0.5711351037,
-0.1921745837,
0.3426740766,
-0.1153674573,
-0.2266301662,
0.1055263504,
0.2558816075,
-0.6505909562,
0.3157378435,
0.5115206242,
-0.8741257191,
0.0261365827,
-0.9842057824,
0.2339106947,
-0.6975638866,
1.3680127859,
0.1551898569,
0.3106337786,
0.0630490854,
-0.0... |
http://arxiv.org/abs/2406.14144v1 | http://arxiv.org/pdf/2406.14144v1 | Finding Safety Neurons in Large Language Models | 2024-06-20T09:35:22 | 2024-06-20T09:35:22 | Jianhui Chen, Xiaozhi Wang, Zijun Yao, Yushi Bai, Lei Hou, Juanzi Li | cs.CL | cs.CL, cs.AI, cs.LG | Large language models (LLMs) excel in various capabilities but also pose
safety risks such as generating harmful content and misinformation, even after
safety alignment. In this paper, we explore the inner mechanisms of safety
alignment from the perspective of mechanistic interpretability, focusing on
identifying and a... | [
0.1926083416,
-0.5938054919,
0.6078258157,
0.1973242015,
0.0288712867,
0.1361360848,
-0.1402873099,
-0.5136703253,
-0.0172689613,
0.116762273,
-0.4467543066,
-0.2741033733,
-0.7265675664,
-0.2313769311,
-0.3441494107,
0.9127780795,
0.0413340218,
-0.1509411782,
0.0612989627,
-0.... |
http://arxiv.org/abs/2406.14137v1 | http://arxiv.org/pdf/2406.14137v1 | MACAROON: Training Vision-Language Models To Be Your Engaged Partners | 2024-06-20T09:27:33 | 2024-06-20T09:27:33 | Shujin Wu, Yi R. Fung, Sha Li, Yixin Wan, Kai-Wei Chang, Heng Ji | cs.CL | cs.CL | Large vision-language models (LVLMs), while proficient in following
instructions and responding to diverse questions, invariably generate detailed
responses even when questions are ambiguous or unanswerable, leading to
hallucinations and bias issues. Thus, it is essential for LVLMs to proactively
engage with humans to ... | [
-0.6207350492,
-0.9699348807,
0.4797818959,
0.0342147574,
-0.1143392026,
0.0980261341,
-0.0954682752,
-0.6823154092,
0.4953331649,
0.698178947,
-0.6686654687,
-0.1437920928,
-0.2736239433,
0.1305689514,
-0.1130218506,
0.9547275901,
-0.2177556604,
-0.0482543297,
-0.1243323833,
-... |
http://arxiv.org/abs/2406.14129v1 | http://arxiv.org/pdf/2406.14129v1 | Towards Event-oriented Long Video Understanding | 2024-06-20T09:14:19 | 2024-06-20T09:14:19 | Yifan Du, Kun Zhou, Yuqi Huo, Yifan Li, Wayne Xin Zhao, Haoyu Lu, Zijia Zhao, Bingning Wang, Weipeng Chen, Ji-Rong Wen | cs.CV | cs.CV, cs.CL, cs.MM | With the rapid development of video Multimodal Large Language Models (MLLMs),
numerous benchmarks have been proposed to assess their video understanding
capability. However, due to the lack of rich events in the videos, these
datasets may suffer from the short-cut bias that the answers can be deduced
from a few frames,... | [
-0.4244332314,
-0.6575040221,
0.6249633431,
0.2273458987,
-0.546600461,
-0.2400372028,
-0.565820992,
-0.3805859089,
-0.3943163753,
0.3921978176,
-0.7761713266,
-0.5991834402,
-0.5888616443,
-0.3886221945,
-0.606552124,
0.879250288,
-0.3671390116,
0.0634709299,
-0.3168685436,
-0... |
http://arxiv.org/abs/2406.14117v1 | http://arxiv.org/pdf/2406.14117v1 | An Investigation of Prompt Variations for Zero-shot LLM-based Rankers | 2024-06-20T09:03:18 | 2024-06-20T09:03:18 | Shuoqi Sun, Shengyao Zhuang, Shuai Wang, Guido Zuccon | cs.IR | cs.IR, cs.CL | We provide a systematic understanding of the impact of specific components
and wordings used in prompts on the effectiveness of rankers based on zero-shot
Large Language Models (LLMs). Several zero-shot ranking methods based on LLMs
have recently been proposed. Among many aspects, methods differ across (1) the
ranking ... | [
-0.2110975981,
-0.7660834193,
0.8414003253,
0.3041597903,
-0.1051024497,
0.0868409947,
-0.0156743675,
-0.5130908489,
0.342307061,
0.3248519599,
-0.4909830987,
-0.5301952958,
-0.8176391721,
0.71081388,
0.1896652877,
0.923029542,
0.022245219,
0.1863119751,
-0.0449167937,
-0.19955... |
http://arxiv.org/abs/2406.14115v1 | http://arxiv.org/pdf/2406.14115v1 | Take the essence and discard the dross: A Rethinking on Data Selection for Fine-Tuning Large Language Models | 2024-06-20T08:58:58 | 2024-06-20T08:58:58 | Ziche Liu, Rui Ke, Feng Jiang, Haizhou Li | cs.CL | cs.CL | Data selection for fine-tuning Large Language Models (LLMs) aims to select a
high-quality subset from a given candidate dataset to train a Pending Fine-tune
Model (PFM) into a Selective-Enhanced Model (SEM). It can improve the model
performance and accelerate the training process. Although a few surveys have
investigat... | [
-0.5393379927,
-0.8415785432,
0.5457894802,
-0.1592679173,
-0.474970758,
0.0901292264,
-0.4552991986,
-0.316211015,
-0.0754727349,
0.5663437247,
-0.5080664754,
-0.4651761353,
-0.4052167833,
0.2597754002,
-0.431677103,
0.6953848004,
-0.1641869694,
0.0872280374,
-0.0460387953,
0.... |
http://arxiv.org/abs/2406.14106v1 | http://arxiv.org/pdf/2406.14106v1 | EasyECR: A Library for Easy Implementation and Evaluation of Event Coreference Resolution Models | 2024-06-20T08:40:21 | 2024-06-20T08:40:21 | Yuncong Li, Tianhua Xu, Sheng-hua Zhong, Haiqin Yang | cs.AI | cs.AI, cs.CL | Event Coreference Resolution (ECR) is the task of clustering event mentions
that refer to the same real-world event. Despite significant advancements, ECR
research faces two main challenges: limited generalizability across domains due
to narrow dataset evaluations, and difficulties in comparing models within
diverse EC... | [
-0.5873945951,
-0.3078670502,
0.5927869081,
-0.2341023237,
-0.2673538327,
-0.2306858152,
-0.6059330106,
-0.5899295211,
-0.272788316,
0.4248000979,
-0.299190402,
-0.690343082,
-0.7143219709,
0.2439529747,
-0.6229082346,
1.1800072193,
-0.0100047393,
-0.0083905915,
-0.0747767687,
... |
http://arxiv.org/abs/2406.14099v2 | http://arxiv.org/pdf/2406.14099v2 | Let Guidelines Guide You: A Prescriptive Guideline-Centered Data Annotation Methodology | 2024-06-20T08:24:57 | 2024-07-02T15:38:17 | Federico Ruggeri, Eleonora Misino, Arianna Muti, Katerina Korre, Paolo Torroni, Alberto Barrón-Cedeño | cs.CL | cs.CL | We introduce the Guideline-Centered annotation process, a novel data
annotation methodology focused on reporting the annotation guidelines
associated with each data sample. We identify three main limitations of the
standard prescriptive annotation process and describe how the
Guideline-Centered methodology overcomes th... | [
-0.4815301597,
-0.6403986812,
0.7688468099,
0.1262280643,
-0.7575543523,
-0.2455228865,
-0.182312876,
-0.3980497718,
0.3681714535,
1.2346383333,
-1.0501060486,
-0.9452541471,
-0.4354892075,
0.054264307,
-0.2847768962,
1.4938730001,
-0.31531775,
0.3529888988,
-0.2474658489,
-0.1... |
http://arxiv.org/abs/2406.14092v1 | http://arxiv.org/pdf/2406.14092v1 | Seamless Language Expansion: Enhancing Multilingual Mastery in Self-Supervised Models | 2024-06-20T08:13:30 | 2024-06-20T08:13:30 | Jing Xu, Minglin Wu, Xixin Wu, Helen Meng | cs.CL | cs.CL, eess.AS | Self-supervised (SSL) models have shown great performance in various
downstream tasks. However, they are typically developed for limited languages,
and may encounter new languages in real-world. Developing a SSL model for each
new language is costly. Thus, it is vital to figure out how to efficiently
adapt existed SSL ... | [
-0.1404843479,
-0.9354683757,
0.2375892997,
0.3843031228,
0.0031162293,
0.3478960097,
-0.3815728128,
-0.4220900834,
0.2494839281,
0.6514636278,
-0.2291445732,
-0.2902877033,
-0.3370863497,
-0.0936682448,
-0.3993240893,
0.9881687164,
0.1511232257,
-0.1298319846,
0.0516544431,
-0... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.