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/2406.12950v1 | http://arxiv.org/pdf/2406.12950v1 | MolecularGPT: Open Large Language Model (LLM) for Few-Shot Molecular Property Prediction | 2024-06-18T12:54:47 | 2024-06-18T12:54:47 | Yuyan Liu, Sirui Ding, Sheng Zhou, Wenqi Fan, Qiaoyu Tan | q-bio.QM | q-bio.QM, cs.AI, cs.CE, cs.CL, cs.LG | Molecular property prediction (MPP) is a fundamental and crucial task in drug
discovery. However, prior methods are limited by the requirement for a large
number of labeled molecules and their restricted ability to generalize for
unseen and new tasks, both of which are essential for real-world applications.
To address ... | [
-0.2640832961,
-0.775988996,
0.6435085535,
-0.3346303403,
-0.221713379,
-0.0186461005,
-0.4564536214,
-0.191707179,
-0.210258916,
0.5219493508,
-0.6277666092,
-0.6601384282,
-0.8284819126,
-0.1595397741,
0.1344342828,
1.1773990393,
0.0843370631,
0.2775609195,
-0.0007944872,
0.0... |
http://arxiv.org/abs/2406.12566v2 | http://arxiv.org/pdf/2406.12566v2 | RichRAG: Crafting Rich Responses for Multi-faceted Queries in Retrieval-Augmented Generation | 2024-06-18T12:52:51 | 2024-06-21T18:12:15 | Shuting Wang, Xin Yu, Mang Wang, Weipeng Chen, Yutao Zhu, Zhicheng Dou | cs.CL | cs.CL | Retrieval-augmented generation (RAG) effectively addresses issues of static
knowledge and hallucination in large language models. Existing studies mostly
focus on question scenarios with clear user intents and concise answers.
However, it is prevalent that users issue broad, open-ended queries with
diverse sub-intents,... | [
-0.8066781163,
-0.8380968571,
0.2615296245,
0.1854189336,
-0.0638768002,
0.1777306199,
-0.1664735973,
-0.4057587981,
0.5445910096,
0.6074695587,
-0.321317941,
-0.0841259658,
-0.5420148373,
0.0210759528,
-0.4332911074,
1.474075675,
0.1467788815,
-0.3185049593,
-0.1939959228,
-0.... |
http://arxiv.org/abs/2406.12564v1 | http://arxiv.org/pdf/2406.12564v1 | Low-Resource Machine Translation through the Lens of Personalized Federated Learning | 2024-06-18T12:50:00 | 2024-06-18T12:50:00 | Viktor Moskvoretskii, Nazarii Tupitsa, Chris Biemann, Samuel Horváth, Eduard Gorbunov, Irina Nikishina | cs.CL | cs.CL, cs.LG | We present a new approach based on the Personalized Federated Learning
algorithm MeritFed that can be applied to Natural Language Tasks with
heterogeneous data. We evaluate it on the Low-Resource Machine Translation
task, using the dataset from the Large-Scale Multilingual Machine Translation
Shared Task (Small Track #... | [
-0.2012296617,
-0.2798902988,
0.3952648044,
0.3168329298,
-0.1800519973,
0.1111869365,
-0.5320150256,
-0.3077380955,
0.3920668364,
0.5473710299,
-0.8659731746,
-0.5258844495,
-0.7207325101,
0.5835573673,
0.0102951294,
1.0431113243,
-0.2712421417,
0.3704702854,
-0.3903918266,
-0... |
http://arxiv.org/abs/2406.12549v1 | http://arxiv.org/pdf/2406.12549v1 | MultiSocial: Multilingual Benchmark of Machine-Generated Text Detection of Social-Media Texts | 2024-06-18T12:26:09 | 2024-06-18T12:26:09 | Dominik Macko, Jakub Kopal, Robert Moro, Ivan Srba | cs.CL | cs.CL, cs.AI | Recent LLMs are able to generate high-quality multilingual texts,
indistinguishable for humans from authentic human-written ones. Research in
machine-generated text detection is however mostly focused on the English
language and longer texts, such as news articles, scientific papers or student
essays. Social-media text... | [
-0.3470226228,
-0.8614925742,
0.4495704472,
0.4773660004,
-0.4917431772,
0.53231287,
-0.3683356047,
-0.7121799588,
0.6404844522,
0.1709690839,
-0.4756835699,
-0.8079921007,
-0.7482534647,
0.5691012144,
-0.3091177344,
0.8255877495,
-0.0907912105,
-0.0022789782,
-0.0081636608,
-0... |
http://arxiv.org/abs/2406.12548v1 | http://arxiv.org/pdf/2406.12548v1 | P-Tailor: Customizing Personality Traits for Language Models via Mixture of Specialized LoRA Experts | 2024-06-18T12:25:13 | 2024-06-18T12:25:13 | Yuhao Dan, Jie Zhou, Qin Chen, Junfeng Tian, Liang He | cs.CL | cs.CL | Personalized large language models (LLMs) have attracted great attention in
many applications, such as intelligent education and emotional support. Most
work focuses on controlling the character settings based on the profile (e.g.,
age, skill, experience, and so on). Conversely, the psychological theory-based
personali... | [
-0.7019733191,
-0.7683445215,
0.5743024349,
0.5380083919,
-0.072156176,
0.0878080502,
-0.4928387702,
-0.1765632033,
0.6425372958,
0.6417355537,
-0.6278699636,
-0.4582355022,
-0.4552003741,
0.123524107,
-0.0745494366,
0.6986299753,
0.1879927218,
0.0454739071,
0.1007108018,
0.021... |
http://arxiv.org/abs/2406.12546v1 | http://arxiv.org/pdf/2406.12546v1 | Liar, Liar, Logical Mire: A Benchmark for Suppositional Reasoning in Large Language Models | 2024-06-18T12:24:22 | 2024-06-18T12:24:22 | Philipp Mondorf, Barbara Plank | cs.CL | cs.CL | Knights and knaves problems represent a classic genre of logical puzzles
where characters either tell the truth or lie. The objective is to logically
deduce each character's identity based on their statements. The challenge
arises from the truth-telling or lying behavior, which influences the logical
implications of ea... | [
-0.0234153494,
-0.5540909171,
1.0764807463,
0.0341414921,
0.0259569064,
0.1277110279,
0.2058117986,
-0.333930254,
0.0458658449,
0.7588744164,
-0.5779988766,
-0.2579062581,
-0.5020827055,
0.2423524559,
-0.0869411007,
0.6092440486,
-0.0712846294,
0.2226329893,
-0.413213104,
-0.23... |
http://arxiv.org/abs/2406.12534v3 | http://arxiv.org/pdf/2406.12534v3 | Unified Active Retrieval for Retrieval Augmented Generation | 2024-06-18T12:09:02 | 2024-06-27T15:37:15 | Qinyuan Cheng, Xiaonan Li, Shimin Li, Qin Zhu, Zhangyue Yin, Yunfan Shao, Linyang Li, Tianxiang Sun, Hang Yan, Xipeng Qiu | cs.CL | cs.CL | In Retrieval-Augmented Generation (RAG), retrieval is not always helpful and
applying it to every instruction is sub-optimal. Therefore, determining whether
to retrieve is crucial for RAG, which is usually referred to as Active
Retrieval. However, existing active retrieval methods face two challenges: 1.
They usually r... | [
-0.2427019477,
-0.7010136247,
0.4523131549,
0.3628606498,
-0.30081141,
0.3471331298,
-0.0416019969,
-0.5798527598,
0.0234257281,
0.5453015566,
-0.0003515133,
0.1950746626,
-0.0859740451,
0.1207868978,
-0.3886191249,
0.894030869,
-0.1443982273,
-0.086232245,
-0.0945479199,
-0.28... |
http://arxiv.org/abs/2406.12527v1 | http://arxiv.org/pdf/2406.12527v1 | FuseGen: PLM Fusion for Data-generation based Zero-shot Learning | 2024-06-18T11:55:05 | 2024-06-18T11:55:05 | Tianyuan Zou, Yang Liu, Peng Li, Jianqing Zhang, Jingjing Liu, Ya-Qin Zhang | cs.CL | cs.CL | Data generation-based zero-shot learning, although effective in training
Small Task-specific Models (STMs) via synthetic datasets generated by
Pre-trained Language Models (PLMs), is often limited by the low quality of such
synthetic datasets. Previous solutions have primarily focused on single PLM
settings, where synth... | [
-0.158802554,
-0.5196380019,
0.3408258855,
0.0860830322,
-0.3724578917,
0.30733639,
-0.2019673884,
0.01215949,
0.0083389841,
0.5043466687,
-0.7479261756,
-0.6854243279,
-0.2803148627,
0.03853653,
-0.4760117233,
0.7399407625,
-0.057648845,
0.2198831588,
-0.6360750794,
-0.0795030... |
http://arxiv.org/abs/2406.12502v1 | http://arxiv.org/pdf/2406.12502v1 | Code-Optimise: Self-Generated Preference Data for Correctness and Efficiency | 2024-06-18T11:05:37 | 2024-06-18T11:05:37 | Leonidas Gee, Milan Gritta, Gerasimos Lampouras, Ignacio Iacobacci | cs.CL | cs.CL | Code Language Models have been trained to generate accurate solutions,
typically with no regard for runtime. On the other hand, previous works that
explored execution optimisation have observed corresponding drops in functional
correctness. To that end, we introduce Code-Optimise, a framework that
incorporates both cor... | [
-0.3334712088,
-0.2103167623,
0.4104640484,
0.2331523597,
-0.058456365,
-0.0362214744,
-0.2768038213,
-0.2177163661,
0.1215844229,
0.528986156,
-0.3995882273,
-0.4084540308,
-0.0248922538,
-0.0942977592,
-0.0159648061,
0.8938925266,
0.1253017932,
0.0416495688,
-0.393240273,
-0.... |
http://arxiv.org/abs/2406.12494v1 | http://arxiv.org/pdf/2406.12494v1 | LightPAL: Lightweight Passage Retrieval for Open Domain Multi-Document Summarization | 2024-06-18T10:57:27 | 2024-06-18T10:57:27 | Masafumi Enomoto, Kunihiro Takeoka, Kosuke Akimoto, Kiril Gashteovski, Masafumi Oyamada | cs.CL | cs.CL | Open-Domain Multi-Document Summarization (ODMDS) is crucial for addressing
diverse information needs, which aims to generate a summary as answer to user's
query, synthesizing relevant content from multiple documents in a large
collection. Existing approaches that first find relevant passages and then
generate a summary... | [
-0.1297788024,
-0.9788320661,
0.6821933389,
-0.0606998987,
-0.5633066297,
0.1871559471,
-0.0256536808,
-0.4592292309,
0.2683507204,
0.499763459,
-0.1763228476,
-0.4253520668,
-0.3377078176,
-0.0007386388,
-0.2420389354,
0.9343767166,
-0.1203543469,
0.0309518855,
0.0714675784,
-... |
http://arxiv.org/abs/2406.12480v1 | http://arxiv.org/pdf/2406.12480v1 | The Power of LLM-Generated Synthetic Data for Stance Detection in Online Political Discussions | 2024-06-18T10:36:21 | 2024-06-18T10:36:21 | Stefan Sylvius Wagner, Maike Behrendt, Marc Ziegele, Stefan Harmeling | cs.CL | cs.CL, cs.LG | Stance detection holds great potential for enhancing the quality of online
political discussions, as it has shown to be useful for summarizing
discussions, detecting misinformation, and evaluating opinion distributions.
Usually, transformer-based models are used directly for stance detection, which
require large amount... | [
-0.9177443385,
-0.7644960284,
0.6883178353,
-0.105203703,
-0.5909368992,
0.1382926852,
-0.2156136781,
-0.359820962,
0.5053722262,
0.4609923363,
-0.4214710891,
-0.5634939671,
-0.6488105059,
0.1578631401,
-0.4096120298,
1.0460710526,
0.3680691719,
0.2755844593,
-0.0632021576,
-0.... |
http://arxiv.org/abs/2406.12474v1 | http://arxiv.org/pdf/2406.12474v1 | Exploring Intra and Inter-language Consistency in Embeddings with ICA | 2024-06-18T10:24:50 | 2024-06-18T10:24:50 | Rongzhi Li, Takeru Matsuda, Hitomi Yanaka | cs.CL | cs.CL, stat.ME | Word embeddings represent words as multidimensional real vectors,
facilitating data analysis and processing, but are often challenging to
interpret. Independent Component Analysis (ICA) creates clearer semantic axes
by identifying independent key features. Previous research has shown ICA's
potential to reveal universal... | [
-0.3334112465,
-0.5559892654,
0.5477193594,
0.3973243535,
-0.2576045394,
-0.0332885832,
-0.2211781442,
-0.4925697148,
0.5636133552,
0.1810376793,
-0.3516809642,
-0.9197536111,
-0.6789650917,
0.1977123022,
-0.3611283302,
0.3904388547,
-0.4421483278,
0.5048021078,
-0.5059121847,
... |
http://arxiv.org/abs/2406.12471v1 | http://arxiv.org/pdf/2406.12471v1 | Fighting Randomness with Randomness: Mitigating Optimisation Instability of Fine-Tuning using Delayed Ensemble and Noisy Interpolation | 2024-06-18T10:20:36 | 2024-06-18T10:20:36 | Branislav Pecher, Jan Cegin, Robert Belanec, Jakub Simko, Ivan Srba, Maria Bielikova | cs.CL | cs.CL | While fine-tuning of pre-trained language models generally helps to overcome
the lack of labelled training samples, it also displays model performance
instability. This instability mainly originates from randomness in
initialisation or data shuffling. To address this, researchers either modify
the training process or a... | [
-0.6619558334,
-0.7612686753,
-0.0168106258,
0.2763210535,
-0.3378916979,
0.5643984079,
-0.4597539604,
-0.4781965613,
-0.0200000536,
0.3541551828,
-0.6781615019,
-0.0627156124,
-0.5825459361,
0.05794058,
-0.3438687921,
0.842682302,
0.1528750807,
0.2074205279,
-0.0551942699,
-0.... |
http://arxiv.org/abs/2406.12468v1 | http://arxiv.org/pdf/2406.12468v1 | Adaptive Token Biaser: Knowledge Editing via Biasing Key Entities | 2024-06-18T10:18:06 | 2024-06-18T10:18:06 | Baolong Bi, Shenghua Liu, Yiwei Wang, Lingrui Mei, Hongcheng Gao, Yilong Xu, Xueqi Cheng | cs.CL | cs.CL, cs.AI | The parametric knowledge memorized by large language models (LLMs) becomes
outdated quickly. In-context editing (ICE) is currently the most effective
method for updating the knowledge of LLMs. Recent advancements involve
enhancing ICE by modifying the decoding strategy, obviating the need for
altering internal model st... | [
-0.1986968368,
-0.6372849941,
0.3842023611,
0.0448171645,
-0.3511379063,
0.4185492694,
-0.2573580444,
-0.2575191259,
0.3939536214,
0.4594894648,
-0.1054604426,
-0.6663694978,
-0.6247490644,
0.0453176647,
-0.3100992143,
0.9601932168,
-0.2061974257,
0.2846558988,
-0.1069929749,
-... |
http://arxiv.org/abs/2406.12442v1 | http://arxiv.org/pdf/2406.12442v1 | Abstraction-of-Thought Makes Language Models Better Reasoners | 2024-06-18T09:46:44 | 2024-06-18T09:46:44 | Ruixin Hong, Hongming Zhang, Xiaoman Pan, Dong Yu, Changshui Zhang | cs.CL | cs.CL, cs.AI | Abstract reasoning, the ability to reason from the abstract essence of a
problem, serves as a key to generalization in human reasoning. However,
eliciting language models to perform reasoning with abstraction remains
unexplored. This paper seeks to bridge this gap by introducing a novel
structured reasoning format call... | [
-0.4881323278,
-0.8103280067,
0.8565919995,
-0.1375765055,
-0.1884218156,
-0.0456601642,
-0.0096997637,
-0.5481785536,
-0.4161962271,
0.3228769004,
-0.4249125421,
-0.4280540049,
-0.6379437447,
-0.0770245194,
-0.569227457,
0.8804633617,
-0.351824522,
-0.403304249,
0.0906855464,
... |
http://arxiv.org/abs/2406.12430v1 | http://arxiv.org/pdf/2406.12430v1 | PlanRAG: A Plan-then-Retrieval Augmented Generation for Generative Large Language Models as Decision Makers | 2024-06-18T09:25:35 | 2024-06-18T09:25:35 | Myeonghwa Lee, Seonho An, Min-Soo Kim | cs.CL | cs.CL, cs.AI, cs.LG, I.2.7 | In this paper, we conduct a study to utilize LLMs as a solution for decision
making that requires complex data analysis. We define Decision QA as the task
of answering the best decision, $d_{best}$, for a decision-making question $Q$,
business rules $R$ and a database $D$. Since there is no benchmark that can
examine D... | [
-0.4707974195,
-0.8861492872,
0.5531615615,
0.3828791678,
-0.1912716031,
0.2306476384,
-0.1399889439,
-0.1538547426,
0.3821661174,
0.5969666839,
-0.4542357326,
-0.5872611403,
-0.0764056891,
-0.1112697273,
-0.3306298256,
0.7188753486,
-0.412076354,
0.0828045234,
0.0192049518,
0.... |
http://arxiv.org/abs/2406.12428v1 | http://arxiv.org/pdf/2406.12428v1 | PSLM: Parallel Generation of Text and Speech with LLMs for Low-Latency Spoken Dialogue Systems | 2024-06-18T09:23:54 | 2024-06-18T09:23:54 | Kentaro Mitsui, Koh Mitsuda, Toshiaki Wakatsuki, Yukiya Hono, Kei Sawada | cs.CL | cs.CL, cs.AI, cs.LG, cs.SD, eess.AS | Multimodal language models that process both text and speech have a potential
for applications in spoken dialogue systems. However, current models face two
major challenges in response generation latency: (1) generating a spoken
response requires the prior generation of a written response, and (2) speech
sequences are ... | [
-0.250425458,
-0.7283644676,
0.8031990528,
0.1029337347,
-0.3136156797,
0.4199823141,
0.0289704446,
-0.2495542169,
0.122620821,
0.6814858913,
-0.8613513708,
-0.1646917015,
-0.3290037513,
0.2467589527,
-0.3227387965,
1.0055721998,
0.2369196415,
0.1982819289,
-0.5035990477,
0.054... |
http://arxiv.org/abs/2406.12422v1 | http://arxiv.org/pdf/2406.12422v1 | Open-Source Web Service with Morphological Dictionary-Supplemented Deep Learning for Morphosyntactic Analysis of Czech | 2024-06-18T09:14:58 | 2024-06-18T09:14:58 | Milan Straka, Jana Straková | cs.CL | cs.CL | We present an open-source web service for Czech morphosyntactic analysis. The
system combines a deep learning model with rescoring by a high-precision
morphological dictionary at inference time. We show that our hybrid method
surpasses two competitive baselines: While the deep learning model ensures
generalization for ... | [
-0.1916075647,
-0.5352945328,
0.3132734597,
0.3908809125,
-0.782513082,
0.2025541812,
-0.3948487043,
-0.5534645915,
0.3559698164,
0.6321150661,
-0.6246325374,
-1.0274438858,
-0.2406570911,
0.4210696816,
-0.1582048237,
0.6476980448,
0.1059288532,
0.8390794992,
-0.0201727822,
0.2... |
http://arxiv.org/abs/2406.12420v1 | http://arxiv.org/pdf/2406.12420v1 | MMUTF: Multimodal Multimedia Event Argument Extraction with Unified Template Filling | 2024-06-18T09:14:17 | 2024-06-18T09:14:17 | Philipp Seeberger, Dominik Wagner, Korbinian Riedhammer | cs.CL | cs.CL, cs.LG | With the advancement of multimedia technologies, news documents and
user-generated content are often represented as multiple modalities, making
Multimedia Event Extraction (MEE) an increasingly important challenge. However,
recent MEE methods employ weak alignment strategies and data augmentation with
simple classifica... | [
-0.6127410531,
-0.5576608181,
0.6093148589,
0.096181415,
-0.5607516766,
-0.2059185803,
-0.1214747578,
-0.2790728509,
0.0674073696,
0.300637275,
-0.8932769895,
-0.2699360251,
-0.5472385287,
0.2131905109,
-0.1259533018,
0.7165648341,
-0.0986414775,
-0.0172987115,
-0.1759360731,
-... |
http://arxiv.org/abs/2406.12419v1 | http://arxiv.org/pdf/2406.12419v1 | AI-Assisted Human Evaluation of Machine Translation | 2024-06-18T09:12:11 | 2024-06-18T09:12:11 | Vilém Zouhar, Tom Kocmi, Mrinmaya Sachan | cs.CL | cs.CL | Annually, research teams spend large amounts of money to evaluate the quality
of machine translation systems (WMT, inter alia). This is expensive because it
requires detailed human labor. The recently proposed annotation protocol, Error
Span Annotation (ESA), has annotators marking erroneous parts of the
translation. I... | [
-0.4688977897,
-0.4736120105,
0.3786315322,
0.4448523223,
-0.2418046445,
0.1274255961,
-0.0836920962,
-0.7607782483,
0.1941636205,
0.3116888702,
-0.3637654483,
-0.1379340589,
-0.885489285,
0.4765014648,
-0.4321958423,
0.851123333,
-0.3760139346,
0.1285639852,
0.1212335229,
-0.1... |
http://arxiv.org/abs/2406.12416v2 | http://arxiv.org/pdf/2406.12416v2 | Beyond Under-Alignment: Atomic Preference Enhanced Factuality Tuning for Large Language Models | 2024-06-18T09:07:30 | 2024-06-27T12:07:55 | Hongbang Yuan, Yubo Chen, Pengfei Cao, Zhuoran Jin, Kang Liu, Jun Zhao | cs.CL | cs.CL, cs.AI | Large language models (LLMs) have achieved remarkable success but still tend
to generate factually erroneous responses, a phenomenon known as hallucination.
A recent trend is to use preference learning to fine-tune models to align with
factuality. However, existing work primarily evaluates fine-tuned models on
in-domai... | [
-0.4674374759,
-0.7887582779,
0.7944875956,
-0.3731062114,
-0.3321945369,
-0.084616676,
-0.4125222564,
-0.4644942582,
0.0285952576,
0.5888593197,
-0.308436662,
-0.4315260649,
-0.4381439686,
0.1479897946,
-0.094599396,
1.06102705,
0.188601613,
-0.074728027,
-0.2101973742,
-0.195... |
http://arxiv.org/abs/2406.12403v1 | http://arxiv.org/pdf/2406.12403v1 | PDSS: A Privacy-Preserving Framework for Step-by-Step Distillation of Large Language Models | 2024-06-18T08:48:14 | 2024-06-18T08:48:14 | Tao Fan, Yan Kang, Weijing Chen, Hanlin Gu, Yuanfeng Song, Lixin Fan, Kai Chen, Qiang Yang | cs.CL | cs.CL, cs.AI | In the context of real-world applications, leveraging large language models
(LLMs) for domain-specific tasks often faces two major challenges:
domain-specific knowledge privacy and constrained resources. To address these
issues, we propose PDSS, a privacy-preserving framework for step-by-step
distillation of LLMs. PDSS... | [
-0.1422850192,
-0.7173639536,
0.6687986255,
0.3614540994,
-0.1961199641,
0.2021133453,
-0.2426294088,
0.2234181613,
-0.259483248,
0.6431539655,
-0.5841837525,
-0.5622187257,
-0.4574877024,
0.1230156347,
-0.5114825964,
1.2924971581,
-0.0236525703,
0.0397127308,
-0.1588095725,
0.... |
http://arxiv.org/abs/2406.12402v1 | http://arxiv.org/pdf/2406.12402v1 | Flee the Flaw: Annotating the Underlying Logic of Fallacious Arguments Through Templates and Slot-filling | 2024-06-18T08:44:45 | 2024-06-18T08:44:45 | Irfan Robbani, Paul Reisert, Naoya Inoue, Surawat Pothong, Camélia Guerraoui, Wenzhi Wang, Shoichi Naito, Jungmin Choi, Kentaro Inui | cs.CL | cs.CL | Prior research in computational argumentation has mainly focused on scoring
the quality of arguments, with less attention on explicating logical errors. In
this work, we introduce four sets of explainable templates for common informal
logical fallacies designed to explicate a fallacy's implicit logic. Using our
templat... | [
-0.4899458289,
-0.764962256,
0.7216919661,
0.0079785315,
-0.256663084,
-0.0712319836,
0.0958878472,
-0.0690081716,
0.0290335156,
0.2792763412,
-0.5871261954,
-0.1954825073,
-0.4937046766,
0.1943958551,
-0.5829516649,
1.0328444242,
-0.2215842754,
-0.1197083294,
-0.2441175133,
-0... |
http://arxiv.org/abs/2406.12399v1 | http://arxiv.org/pdf/2406.12399v1 | QueerBench: Quantifying Discrimination in Language Models Toward Queer Identities | 2024-06-18T08:40:29 | 2024-06-18T08:40:29 | Mae Sosto, Alberto Barrón-Cedeño | cs.CL | cs.CL, cs.AI, cs.CY | With the increasing role of Natural Language Processing (NLP) in various
applications, challenges concerning bias and stereotype perpetuation are
accentuated, which often leads to hate speech and harm. Despite existing
studies on sexism and misogyny, issues like homophobia and transphobia remain
underexplored and often... | [
-0.2598774731,
-0.7395499349,
0.2826535404,
0.3438502848,
0.1456544697,
-0.0555074811,
0.0610236265,
-0.8420264125,
0.1077699363,
0.4783982635,
-0.5022471547,
-0.4165267646,
-0.5403189063,
0.2810781598,
-0.8229976296,
0.7311317325,
0.1756608337,
0.2074251026,
-0.2776124775,
-0.... |
http://arxiv.org/abs/2406.12397v1 | http://arxiv.org/pdf/2406.12397v1 | Unveiling the Flaws: Exploring Imperfections in Synthetic Data and Mitigation Strategies for Large Language Models | 2024-06-18T08:38:59 | 2024-06-18T08:38:59 | Jie Chen, Yupeng Zhang, Bingning Wang, Wayne Xin Zhao, Ji-Rong Wen, Weipeng Chen | cs.CL | cs.CL | Synthetic data has been proposed as a solution to address the issue of
high-quality data scarcity in the training of large language models (LLMs).
Studies have shown that synthetic data can effectively improve the performance
of LLMs on downstream benchmarks. However, despite its potential benefits, our
analysis sugges... | [
-0.2719806433,
-1.1365989447,
0.67959553,
-0.1027072892,
-0.1897219568,
0.586253047,
-0.0956368744,
-0.3631881177,
-0.0790444762,
0.8105595708,
-0.6392840743,
-0.376002878,
-0.3240053654,
-0.0485120937,
-0.5370663404,
0.7588594556,
-0.1405702233,
0.2856270671,
-0.5022153258,
-0... |
http://arxiv.org/abs/2406.12946v1 | http://arxiv.org/pdf/2406.12946v1 | Instruction Data Generation and Unsupervised Adaptation for Speech Language Models | 2024-06-18T08:27:00 | 2024-06-18T08:27:00 | Vahid Noroozi, Zhehuai Chen, Somshubra Majumdar, Steve Huang, Jagadeesh Balam, Boris Ginsburg | eess.AS | eess.AS, cs.AI, cs.CL, cs.LG | In this paper, we propose three methods for generating synthetic samples to
train and evaluate multimodal large language models capable of processing both
text and speech inputs. Addressing the scarcity of samples containing both
modalities, synthetic data generation emerges as a crucial strategy to enhance
the perform... | [
-0.0782101005,
-0.8455294967,
0.5103577375,
0.0021543128,
-0.070131585,
0.1505527943,
-0.4618613124,
-0.2578707635,
-0.1276200712,
0.7118457556,
-0.6860015988,
-0.3255746365,
-0.1492179036,
-0.1414349526,
-0.5983524919,
0.9114655852,
0.2063127607,
0.5136265159,
-0.1983040869,
0... |
http://arxiv.org/abs/2406.12389v1 | http://arxiv.org/pdf/2406.12389v1 | EMO-KNOW: A Large Scale Dataset on Emotion and Emotion-cause | 2024-06-18T08:26:33 | 2024-06-18T08:26:33 | Mia Huong Nguyen, Yasith Samaradivakara, Prasanth Sasikumar, Chitralekha Gupta, Suranga Nanayakkara | cs.CL | cs.CL | Emotion-Cause analysis has attracted the attention of researchers in recent
years. However, most existing datasets are limited in size and number of
emotion categories. They often focus on extracting parts of the document that
contain the emotion cause and fail to provide more abstractive, generalizable
root cause. To ... | [
-0.4560075402,
-0.7662249804,
0.638450861,
0.0949011147,
-0.1505167782,
0.0293632206,
-0.1452722847,
-0.4314194918,
0.7039382458,
0.3059940338,
-0.7427173853,
-0.451511085,
-0.6349684596,
0.1650221348,
-0.5186184645,
0.9128513336,
0.0876290277,
-0.4167544842,
0.3031505048,
-0.0... |
http://arxiv.org/abs/2406.12387v1 | http://arxiv.org/pdf/2406.12387v1 | Performant ASR Models for Medical Entities in Accented Speech | 2024-06-18T08:19:48 | 2024-06-18T08:19:48 | Tejumade Afonja, Tobi Olatunji, Sewade Ogun, Naome A. Etori, Abraham Owodunni, Moshood Yekini | eess.AS | eess.AS, cs.CL, cs.SD | Recent strides in automatic speech recognition (ASR) have accelerated their
application in the medical domain where their performance on accented medical
named entities (NE) such as drug names, diagnoses, and lab results, is largely
unknown. We rigorously evaluate multiple ASR models on a clinical English
dataset of 93... | [
-0.0098154349,
-0.408208698,
0.5542776585,
-0.0280667488,
-0.09113276,
0.2252830863,
-0.3977795541,
-0.8793047071,
0.4412476122,
0.712539494,
-0.1308770478,
-0.4875865877,
-0.7632309198,
0.2875515223,
-0.1191971675,
0.6015145183,
0.107128948,
0.6528292894,
0.0469337441,
-0.5580... |
http://arxiv.org/abs/2406.12386v1 | http://arxiv.org/pdf/2406.12386v1 | IPEval: A Bilingual Intellectual Property Agency Consultation Evaluation Benchmark for Large Language Models | 2024-06-18T08:18:18 | 2024-06-18T08:18:18 | Qiyao Wang, Jianguo Huang, Shule Lu, Yuan Lin, Kan Xu, Liang Yang, Hongfei Lin | cs.CL | cs.CL | The rapid development of Large Language Models (LLMs) in vertical domains,
including intellectual property (IP), lacks a specific evaluation benchmark for
assessing their understanding, application, and reasoning abilities. To fill
this gap, we introduce IPEval, the first evaluation benchmark tailored for IP
agency and... | [
-0.4098105133,
-0.4576335549,
0.209235087,
0.9420602918,
-0.2085507512,
0.4299689531,
-0.1424319744,
-0.540155232,
-0.1010175943,
0.5642855763,
-0.1681223214,
-0.5941858888,
-0.3922323585,
0.0209173411,
-0.2156370133,
0.9744832516,
-0.153267473,
0.3080292344,
-0.0833462328,
-0.... |
http://arxiv.org/abs/2406.12382v1 | http://arxiv.org/pdf/2406.12382v1 | From Instance Training to Instruction Learning: Task Adapters Generation from Instructions | 2024-06-18T08:14:28 | 2024-06-18T08:14:28 | Huanxuan Liao, Yao Xu, Shizhu He, Yuanzhe Zhang, Yanchao Hao, Shengping Liu, Kang Liu, Jun Zhao | cs.CL | cs.CL | Large language models (LLMs) have acquired the ability to solve general tasks
by utilizing instruction finetuning (IFT). However, IFT still relies heavily on
instance training of extensive task data, which greatly limits the adaptability
of LLMs to real-world scenarios where labeled task instances are scarce and
broade... | [
-0.097422123,
-0.691033721,
0.501440227,
-0.1463258117,
-0.1689771116,
-0.288472116,
-0.2674061954,
-0.2272160351,
-0.430398047,
0.3524264991,
-0.6857850552,
-0.2932480276,
-0.5488443375,
-0.1566108465,
0.016429998,
1.0778331757,
-0.0130373659,
0.004595743,
-0.2283899635,
0.259... |
http://arxiv.org/abs/2406.12381v2 | http://arxiv.org/pdf/2406.12381v2 | QOG:Question and Options Generation based on Language Model | 2024-06-18T08:09:58 | 2024-07-02T09:21:03 | Jincheng Zhou | cs.CL | cs.CL, cs.LG | Question-Options Generation (QOG) is a task that involves generating a set of
question-options pairs given context. This task has various applications,
including fine-tuning large models, information retrieval, and automated
multiple-choice question generation for education. In this paper, we develop
QOG models using t... | [
-0.4008032084,
-1.2144975662,
0.6891438365,
-0.4319943786,
-0.170963794,
0.2068870366,
-0.0369330086,
-0.341507107,
-0.2865421176,
0.800141871,
-0.4739849865,
-0.2055522352,
-0.1024268344,
0.0781110227,
-0.3603649139,
0.9986391068,
0.0063258144,
-0.0077569759,
-0.4002315402,
-0... |
http://arxiv.org/abs/2406.12373v2 | http://arxiv.org/pdf/2406.12373v2 | WebCanvas: Benchmarking Web Agents in Online Environments | 2024-06-18T07:58:33 | 2024-06-27T16:56:13 | Yichen Pan, Dehan Kong, Sida Zhou, Cheng Cui, Yifei Leng, Bing Jiang, Hangyu Liu, Yanyi Shang, Shuyan Zhou, Tongshuang Wu, Zhengyang Wu | cs.CL | cs.CL, cs.AI, cs.LG, 68T50, I.2.7 | For web agents to be practically useful, they must adapt to the continuously
evolving web environment characterized by frequent updates to user interfaces
and content. However, most existing benchmarks only capture the static aspects
of the web. To bridge this gap, we introduce WebCanvas, an innovative online
evaluatio... | [
-0.7716379762,
-0.3323344886,
0.386875242,
0.324049443,
-0.2012484968,
0.2831048667,
-0.2099517733,
-0.8457427621,
0.1507867426,
0.0164288785,
-0.5946502686,
-0.3613560796,
-0.1984048039,
0.0634080246,
-0.2776207328,
0.7046630979,
-0.0034902943,
-0.0095364889,
-0.7219461203,
-0... |
http://arxiv.org/abs/2406.12369v1 | http://arxiv.org/pdf/2406.12369v1 | A Comparative Study of Continuous Sign Language Recognition Techniques | 2024-06-18T07:51:44 | 2024-06-18T07:51:44 | Sarah Alyami, Hamzah Luqman | cs.CL | cs.CL, cs.CV | Continuous Sign Language Recognition (CSLR) focuses on the interpretation of
a sequence of sign language gestures performed continually without pauses. In
this study, we conduct an empirical evaluation of recent deep learning CSLR
techniques and assess their performance across various datasets and sign
languages. The m... | [
0.1341762394,
-0.1052135825,
0.2478117496,
0.0744045153,
-0.3615996838,
0.6303107738,
-0.3461042047,
-0.8462737203,
-0.3046108484,
0.1830880642,
-0.3729503453,
-0.5836586952,
-0.729847014,
0.1593674719,
-0.4349900484,
0.5153599977,
-0.0606036671,
0.7300277948,
0.1842867434,
-0.... |
http://arxiv.org/abs/2406.12364v1 | http://arxiv.org/pdf/2406.12364v1 | Does Context Help Mitigate Gender Bias in Neural Machine Translation? | 2024-06-18T07:44:13 | 2024-06-18T07:44:13 | Harritxu Gete, Thierry Etchegoyhen | cs.CL | cs.CL | Neural Machine Translation models tend to perpetuate gender bias present in
their training data distribution. Context-aware models have been previously
suggested as a means to mitigate this type of bias. In this work, we examine
this claim by analysing in detail the translation of stereotypical professions
in English t... | [
-0.2228941917,
-0.6393787861,
0.4147194326,
0.2954609692,
0.025017174,
-0.5182133317,
-0.133082822,
-0.4107426703,
-0.2809426785,
0.5277418494,
-0.7711498141,
-0.3804627061,
-0.7628711462,
0.3417335749,
-0.5830899477,
0.8927582502,
0.0585558563,
0.7006005645,
-0.1815225631,
-0.... |
http://arxiv.org/abs/2406.12354v1 | http://arxiv.org/pdf/2406.12354v1 | Cross-Lingual Unlearning of Selective Knowledge in Multilingual Language Models | 2024-06-18T07:40:18 | 2024-06-18T07:40:18 | Minseok Choi, Kyunghyun Min, Jaegul Choo | cs.CL | cs.CL | Pretrained language models memorize vast amounts of information, including
private and copyrighted data, raising significant safety concerns. Retraining
these models after excluding sensitive data is prohibitively expensive, making
machine unlearning a viable, cost-effective alternative. Previous research has
focused o... | [
-0.0857747495,
-0.5423616171,
0.0062934798,
-0.0701961964,
-0.2843739092,
0.3429488838,
-0.4352398515,
-0.3740874231,
-0.199532479,
0.7615019679,
-0.4448982477,
-0.298186034,
-0.4989086986,
-0.2201840132,
-0.1900013983,
1.0142143965,
-0.008665395,
0.2158482373,
-0.0738853291,
-... |
http://arxiv.org/abs/2406.12347v1 | http://arxiv.org/pdf/2406.12347v1 | Interpreting Bias in Large Language Models: A Feature-Based Approach | 2024-06-18T07:28:15 | 2024-06-18T07:28:15 | Nirmalendu Prakash, Lee Ka Wei Roy | cs.CL | cs.CL | Large Language Models (LLMs) such as Mistral and LLaMA have showcased
remarkable performance across various natural language processing (NLP) tasks.
Despite their success, these models inherit social biases from the diverse
datasets on which they are trained. This paper investigates the propagation of
biases within LLM... | [
-0.3916598856,
-0.8970774412,
0.6086034775,
0.4078351259,
-0.0801586062,
-0.1540671736,
-0.1823457032,
-0.4819181561,
0.2487726808,
0.819331646,
-0.6953910589,
-0.3406884372,
-1.0188499689,
-0.0479093567,
-0.5547201633,
0.9994905591,
0.0680894107,
0.3545601368,
-0.0772925392,
-... |
http://arxiv.org/abs/2406.12336v1 | http://arxiv.org/pdf/2406.12336v1 | A Compass for Navigating the World of Sentence Embeddings for the Telecom Domain | 2024-06-18T07:03:34 | 2024-06-18T07:03:34 | Sujoy Roychowdhury, Sumit Soman, H. G. Ranjani, Vansh Chhabra, Neeraj Gunda, Subhadip Bandyopadhyay, Sai Krishna Bala | cs.CL | cs.CL, cs.LG, 68T50, I.2.7 | A plethora of sentence embedding models makes it challenging to choose one,
especially for domains such as telecom, rich with specialized vocabulary. We
evaluate multiple embeddings obtained from publicly available models and their
domain-adapted variants, on both point retrieval accuracies as well as their
(95\%) conf... | [
-0.2630442977,
-0.6661900878,
0.6527786255,
0.1258016229,
-0.3464951813,
-0.0752218515,
-0.2463781387,
-0.3595519662,
0.2937751412,
0.1844223291,
-0.2001929581,
-0.6220443845,
-0.653716445,
0.2011698633,
-0.146057412,
0.5138351321,
0.0792807713,
0.189706862,
-0.2163607627,
-0.3... |
http://arxiv.org/abs/2406.12335v1 | http://arxiv.org/pdf/2406.12335v1 | Attention Score is not All You Need for Token Importance Indicator in KV Cache Reduction: Value Also Matters | 2024-06-18T07:01:11 | 2024-06-18T07:01:11 | Zhiyu Guo, Hidetaka Kamigaito, Taro Watanabe | cs.CL | cs.CL, cs.LG | Scaling the context size of large language models (LLMs) enables them to
perform various new tasks, e.g., book summarization. However, the memory cost
of the Key and Value (KV) cache in attention significantly limits the practical
applications of LLMs. Recent works have explored token pruning for KV cache
reduction in ... | [
-0.3455727398,
-0.4262008369,
0.3149113953,
0.4674053192,
-0.5425118804,
-0.2577678859,
-0.3174467385,
-0.2341472507,
0.3899787366,
0.289506793,
-0.2504985034,
-0.4575097561,
-0.4889894426,
-0.2721546888,
-0.3720709682,
0.8755144477,
0.1500388384,
-0.1209479496,
-0.4491453767,
... |
http://arxiv.org/abs/2406.12331v1 | http://arxiv.org/pdf/2406.12331v1 | Retrieval Meets Reasoning: Dynamic In-Context Editing for Long-Text Understanding | 2024-06-18T06:54:28 | 2024-06-18T06:54:28 | Weizhi Fei, Xueyan Niu, Guoqing Xie, Yanhua Zhang, Bo Bai, Lei Deng, Wei Han | cs.CL | cs.CL, cs.AI | Current Large Language Models (LLMs) face inherent limitations due to their
pre-defined context lengths, which impede their capacity for multi-hop
reasoning within extensive textual contexts. While existing techniques like
Retrieval-Augmented Generation (RAG) have attempted to bridge this gap by
sourcing external infor... | [
-0.2453332841,
-1.0783647299,
1.3019160032,
0.0017572034,
-0.2829438448,
0.3596016467,
-0.5028567314,
-0.6582072973,
0.1430404633,
0.4639383256,
-0.49587816,
-0.1530811191,
-0.4095207155,
0.1114243567,
-0.2729533613,
0.6808590293,
-0.2969875634,
-0.4852571785,
-0.4397514462,
-0... |
http://arxiv.org/abs/2406.12329v1 | http://arxiv.org/pdf/2406.12329v1 | SNAP: Unlearning Selective Knowledge in Large Language Models with Negative Instructions | 2024-06-18T06:54:05 | 2024-06-18T06:54:05 | Minseok Choi, Daniel Rim, Dohyun Lee, Jaegul Choo | cs.CL | cs.CL | Instruction-following large language models (LLMs), such as ChatGPT, have
become increasingly popular with the general audience, many of whom are
incorporating them into their daily routines. However, these LLMs inadvertently
disclose personal or copyrighted information, which calls for a machine
unlearning method to r... | [
-0.2803471386,
-0.7838842869,
0.2963812351,
-0.1327487677,
-0.5333205462,
0.2623670697,
-0.5126861334,
-0.1286895871,
0.2489773482,
0.8066241741,
-0.6043202877,
-0.5135880113,
-0.5683564544,
-0.2881462574,
-0.6513841748,
1.1917709112,
0.23145248,
0.0488984063,
-0.1066381857,
-0... |
http://arxiv.org/abs/2406.12321v1 | http://arxiv.org/pdf/2406.12321v1 | Automatic benchmarking of large multimodal models via iterative experiment programming | 2024-06-18T06:43:46 | 2024-06-18T06:43:46 | Alessandro Conti, Enrico Fini, Paolo Rota, Yiming Wang, Massimiliano Mancini, Elisa Ricci | cs.AI | cs.AI, cs.CL, cs.CV | Assessing the capabilities of large multimodal models (LMMs) often requires
the creation of ad-hoc evaluations. Currently, building new benchmarks requires
tremendous amounts of manual work for each specific analysis. This makes the
evaluation process tedious and costly. In this paper, we present APEx,
Automatic Progra... | [
-0.3893199861,
-0.7148860097,
0.5595629215,
0.3391296268,
0.049056448,
-0.0558034778,
-0.4709773362,
-0.3317263424,
-0.1234175712,
0.1133558825,
-0.4259548485,
-0.5421605706,
-0.4656432271,
0.4354887903,
-0.5136653781,
0.438234359,
-0.2845824659,
0.166915834,
-0.5440196395,
-0.... |
http://arxiv.org/abs/2406.12319v1 | http://arxiv.org/pdf/2406.12319v1 | PRePair: Pointwise Reasoning Enhance Pairwise Evaluating for Robust Instruction-Following Assessments | 2024-06-18T06:43:04 | 2024-06-18T06:43:04 | Hawon Jeong, ChaeHun Park, Jimin Hong, Jaegul Choo | cs.CL | cs.CL | Pairwise evaluation using large language models (LLMs) is widely used for
evaluating natural language generation (NLG) tasks. However, the reliability of
LLMs is often compromised by biases, such as favoring verbosity and
authoritative tone. In the study, we focus on the comparison of two LLM-based
evaluation approache... | [
-0.152797088,
-0.823798418,
0.5714039207,
0.3739739358,
-0.0633706003,
0.3663488925,
-0.4169150889,
-0.0873316303,
-0.0014533979,
0.3306531906,
-0.1143622026,
-0.6798553467,
-0.6437894106,
0.0144962566,
-0.3202688694,
0.8337605,
-0.0073011927,
0.1375421137,
-0.5552355647,
-0.19... |
http://arxiv.org/abs/2406.12317v1 | http://arxiv.org/pdf/2406.12317v1 | Finding Task-specific Subnetworks in Multi-task Spoken Language Understanding Model | 2024-06-18T06:39:41 | 2024-06-18T06:39:41 | Hayato Futami, Siddhant Arora, Yosuke Kashiwagi, Emiru Tsunoo, Shinji Watanabe | cs.CL | cs.CL, eess.AS | Recently, multi-task spoken language understanding (SLU) models have emerged,
designed to address various speech processing tasks. However, these models
often rely on a large number of parameters. Also, they often encounter
difficulties in adapting to new data for a specific task without experiencing
catastrophic forge... | [
-0.2091821879,
-0.5673003197,
0.3179589212,
0.1240516007,
-0.3775823712,
-0.2915963829,
-0.2958550751,
-0.1102194861,
0.2501121163,
0.4413392246,
-0.820106864,
-0.413682282,
-0.3864795864,
-0.2331001014,
-0.3123579919,
0.7203917503,
0.0561113805,
-0.2810873091,
-0.3434647918,
-... |
http://arxiv.org/abs/2406.12307v1 | http://arxiv.org/pdf/2406.12307v1 | Can Tool-augmented Large Language Models be Aware of Incomplete Conditions? | 2024-06-18T06:28:06 | 2024-06-18T06:28:06 | Seungbin Yang, ChaeHun Park, Taehee Kim, Jaegul Choo | cs.CL | cs.CL | Recent advancements in integrating large language models (LLMs) with tools
have allowed the models to interact with real-world environments. However,
these tool-augmented LLMs often encounter incomplete scenarios when users
provide partial information or the necessary tools are unavailable. Recognizing
and managing suc... | [
-0.278432548,
-0.9204832911,
1.1121532917,
0.2552753687,
-0.1852341741,
0.3994095623,
-0.4457427561,
-0.6617426872,
0.0169200413,
0.4777802527,
-0.7003458738,
-0.2323648632,
-0.6052671075,
0.1137972623,
-0.6259568334,
0.6468371749,
-0.2826257944,
-0.1513524652,
-0.711879313,
-0... |
http://arxiv.org/abs/2406.12304v1 | http://arxiv.org/pdf/2406.12304v1 | COT: A Generative Approach for Hate Speech Counter-Narratives via Contrastive Optimal Transport | 2024-06-18T06:24:26 | 2024-06-18T06:24:26 | Linhao Zhang, Li Jin, Guangluan Xu, Xiaoyu Li, Xian Sun | cs.CL | cs.CL, 68U15, I.2.7 | Counter-narratives, which are direct responses consisting of non-aggressive
fact-based arguments, have emerged as a highly effective approach to combat the
proliferation of hate speech. Previous methodologies have primarily focused on
fine-tuning and post-editing techniques to ensure the fluency of generated
contents, ... | [
-0.659758389,
-0.6499403119,
0.332972765,
-0.1847680509,
-0.4337579012,
0.1905187964,
-0.0249145254,
-0.2557108104,
0.4414722621,
0.4048282802,
-0.5436825156,
-0.5140542984,
-0.5835854411,
-0.0772577152,
-0.7836617231,
1.114526391,
0.1964249462,
0.0149071617,
0.2239591628,
-0.5... |
http://arxiv.org/abs/2406.12295v1 | http://arxiv.org/pdf/2406.12295v1 | Fast and Slow Generating: An Empirical Study on Large and Small Language Models Collaborative Decoding | 2024-06-18T05:59:28 | 2024-06-18T05:59:28 | Kaiyan Zhang, Jianyu Wang, Ning Ding, Biqing Qi, Ermo Hua, Xingtai Lv, Bowen Zhou | cs.CL | cs.CL | Large Language Models (LLMs) demonstrate impressive performance in diverse
applications, yet they face significant drawbacks, including high inference
latency, expensive training cost, and generation of hallucination.
Collaborative decoding between large and small language models (SLMs) offers a
novel approach to addre... | [
-0.2962245345,
-0.7941974401,
0.7423627973,
0.0893698335,
-0.1421141624,
0.342184782,
-0.5337546468,
-0.5796703696,
0.1133515239,
0.3822900355,
-0.6294752359,
-0.0510952584,
-0.5758118629,
0.1477244943,
-0.2303194255,
0.6231539845,
0.074339129,
0.2956471741,
-0.2064801455,
0.13... |
http://arxiv.org/abs/2406.12277v1 | http://arxiv.org/pdf/2406.12277v1 | What Matters in Learning Facts in Language Models? Multifaceted Knowledge Probing with Diverse Multi-Prompt Datasets | 2024-06-18T05:11:35 | 2024-06-18T05:11:35 | Xin Zhao, Naoki Yoshinaga, Daisuke Oba | cs.CL | cs.CL | Large language models (LLMs) face issues in handling factual knowledge,
making it vital to evaluate their true ability to understand facts. In this
study, we introduce knowledge probing frameworks, BELIEF(-ICL), to evaluate the
knowledge understanding ability of not only encoder-based PLMs but also
decoder-based PLMs f... | [
-0.4286163151,
-0.6522180438,
1.0690288544,
0.2897406518,
-0.0991538465,
0.1163956821,
-0.1167376339,
-0.0720862076,
-0.4490491748,
0.2023724169,
-0.3598402739,
-0.6284592748,
-0.6580002904,
0.2105719894,
-0.0587128438,
0.9565030336,
0.1110615134,
0.1565731764,
-0.5427272916,
-... |
http://arxiv.org/abs/2406.12276v1 | http://arxiv.org/pdf/2406.12276v1 | CodeNav: Beyond tool-use to using real-world codebases with LLM agents | 2024-06-18T05:10:38 | 2024-06-18T05:10:38 | Tanmay Gupta, Luca Weihs, Aniruddha Kembhavi | cs.AI | cs.AI, cs.CL, cs.SE | We present CodeNav, an LLM agent that navigates and leverages previously
unseen code repositories to solve user queries. In contrast to tool-use LLM
agents that require ``registration'' of all relevant tools via manual
descriptions within the LLM context, CodeNav automatically indexes and searches
over code blocks in t... | [
-0.2199693322,
-0.190272063,
0.2748535275,
0.0970744342,
-0.1446655542,
0.6357680559,
-0.1191496626,
-0.421217531,
0.3793033361,
0.6766143441,
-0.3860582411,
-0.747730732,
0.0974947214,
-0.0669028759,
-0.5001681447,
0.4745383263,
0.0434978269,
-0.0736825094,
-0.4986565709,
-0.1... |
http://arxiv.org/abs/2406.12274v1 | http://arxiv.org/pdf/2406.12274v1 | SafeInfer: Context Adaptive Decoding Time Safety Alignment for Large Language Models | 2024-06-18T05:03:23 | 2024-06-18T05:03:23 | Somnath Banerjee, Soham Tripathy, Sayan Layek, Shanu Kumar, Animesh Mukherjee, Rima Hazra | cs.CL | cs.CL | Safety-aligned language models often exhibit fragile and imbalanced safety
mechanisms, increasing the likelihood of generating unsafe content. In
addition, incorporating new knowledge through editing techniques to language
models can further compromise safety. To address these issues, we propose
SafeInfer, a context-ad... | [
0.0363904499,
-0.7436228395,
0.0273046829,
0.2805584371,
-0.2319955379,
0.0748781562,
0.0469419062,
-0.6329650283,
-0.3641342521,
0.3753283024,
-0.3051525354,
-0.5236006379,
-0.4098547697,
-0.0318425223,
-0.5655896068,
0.9317361116,
0.1732466966,
-0.0867912024,
0.0235733911,
-0... |
http://arxiv.org/abs/2406.12269v1 | http://arxiv.org/pdf/2406.12269v1 | Unveiling Implicit Table Knowledge with Question-Then-Pinpoint Reasoner for Insightful Table Summarization | 2024-06-18T04:55:09 | 2024-06-18T04:55:09 | Kwangwook Seo, Jinyoung Yeo, Dongha Lee | cs.CL | cs.CL | Implicit knowledge hidden within the explicit table cells, such as data
insights, is the key to generating a high-quality table summary. However,
unveiling such implicit knowledge is a non-trivial task. Due to the complex
nature of structured tables, it is challenging even for large language models
(LLMs) to mine the i... | [
-0.3853511512,
-0.5434365869,
0.9387344718,
-0.0808783025,
-0.3746970892,
-0.0705656409,
0.1445165873,
-0.2531483471,
-0.1272023469,
0.4326037169,
-0.5349023342,
-0.5562207699,
-0.4378869534,
-0.2050773501,
-0.8262866735,
1.2353423834,
-0.1681352407,
0.0182973985,
-0.5663191676,
... |
http://arxiv.org/abs/2406.12266v2 | http://arxiv.org/pdf/2406.12266v2 | Towards a Client-Centered Assessment of LLM Therapists by Client Simulation | 2024-06-18T04:46:55 | 2024-06-20T22:06:47 | Jiashuo Wang, Yang Xiao, Yanran Li, Changhe Song, Chunpu Xu, Chenhao Tan, Wenjie Li | cs.CL | cs.CL | Although there is a growing belief that LLMs can be used as therapists,
exploring LLMs' capabilities and inefficacy, particularly from the client's
perspective, is limited. This work focuses on a client-centered assessment of
LLM therapists with the involvement of simulated clients, a standard approach
in clinical medi... | [
-0.4206168056,
-0.5004409552,
0.5322277546,
0.3819812238,
-0.1911614835,
0.3776297569,
-0.1786538661,
-0.5241723061,
0.337587893,
0.2491092533,
-0.4912164211,
-0.2856301665,
-0.5147558451,
0.0914698392,
0.066048488,
0.5703172684,
-0.1208877936,
0.8585609794,
-0.4047144353,
-0.2... |
http://arxiv.org/abs/2406.12263v1 | http://arxiv.org/pdf/2406.12263v1 | Defending Against Social Engineering Attacks in the Age of LLMs | 2024-06-18T04:39:40 | 2024-06-18T04:39:40 | Lin Ai, Tharindu Kumarage, Amrita Bhattacharjee, Zizhou Liu, Zheng Hui, Michael Davinroy, James Cook, Laura Cassani, Kirill Trapeznikov, Matthias Kirchner, Arslan Basharat, Anthony Hoogs, Joshua Garland, Huan Liu, Julia Hirschberg | cs.CL | cs.CL | The proliferation of Large Language Models (LLMs) poses challenges in
detecting and mitigating digital deception, as these models can emulate human
conversational patterns and facilitate chat-based social engineering (CSE)
attacks. This study investigates the dual capabilities of LLMs as both
facilitators and defenders... | [
-0.1922965944,
-0.8803128004,
0.3668246865,
-0.0600528046,
-0.095539026,
0.7788729072,
0.038879998,
-0.4951540232,
0.3925828934,
0.4179235697,
-0.6990145445,
-0.482056886,
-0.5806810856,
-0.0943096131,
-0.6411812901,
0.681923151,
0.4021059573,
0.0794601664,
-0.1301661134,
0.220... |
http://arxiv.org/abs/2406.12260v1 | http://arxiv.org/pdf/2406.12260v1 | Self-Supervised Time-Series Anomaly Detection Using Learnable Data Augmentation | 2024-06-18T04:25:56 | 2024-06-18T04:25:56 | Kukjin Choi, Jihun Yi, Jisoo Mok, Sungroh Yoon | cs.LG | cs.LG, cs.AI, cs.CL | Continuous efforts are being made to advance anomaly detection in various
manufacturing processes to increase the productivity and safety of industrial
sites. Deep learning replaced rule-based methods and recently emerged as a
promising method for anomaly detection in diverse industries. However, in the
real world, the... | [
-0.351574719,
-0.6506482363,
0.2779516578,
-0.0783616602,
-0.3933867812,
0.1222683266,
0.4002488852,
-0.6566318274,
0.2922464907,
0.4083887637,
-1.0225422382,
-0.4224818945,
-0.2880622745,
-0.0643626451,
-0.4525338113,
0.9573985338,
-0.2042753696,
-0.4379859567,
0.0942328274,
-... |
http://arxiv.org/abs/2406.12255v1 | http://arxiv.org/pdf/2406.12255v1 | A Hopfieldian View-based Interpretation for Chain-of-Thought Reasoning | 2024-06-18T04:07:13 | 2024-06-18T04:07:13 | Lijie Hu, Liang Liu, Shu Yang, Xin Chen, Hongru Xiao, Mengdi Li, Pan Zhou, Muhammad Asif Ali, Di Wang | cs.CL | cs.CL, cs.AI, cs.HC, cs.LG | Chain-of-Thought (CoT) holds a significant place in augmenting the reasoning
performance for large language models (LLMs). While some studies focus on
improving CoT accuracy through methods like retrieval enhancement, yet a
rigorous explanation for why CoT achieves such success remains unclear. In this
paper, we analyz... | [
-0.355235219,
-0.8748386502,
0.9053951502,
-0.118608281,
-0.2278492302,
0.3718183637,
-0.0925700739,
-0.5395563245,
0.1086085588,
0.6272456646,
-0.4568687379,
-0.3046337366,
-0.7098626494,
0.0893854722,
-0.2881298959,
0.9873555899,
-0.5272952318,
-0.4450672865,
0.2103680968,
0.... |
http://arxiv.org/abs/2406.12252v1 | http://arxiv.org/pdf/2406.12252v1 | Language and Multimodal Models in Sports: A Survey of Datasets and Applications | 2024-06-18T03:59:26 | 2024-06-18T03:59:26 | Haotian Xia, Zhengbang Yang, Yun Zhao, Yuqing Wang, Jingxi Li, Rhys Tracy, Zhuangdi Zhu, Yuan-fang Wang, Hanjie Chen, Weining Shen | cs.CL | cs.CL | Recent integration of Natural Language Processing (NLP) and multimodal models
has advanced the field of sports analytics. This survey presents a
comprehensive review of the datasets and applications driving these innovations
post-2020. We overviewed and categorized datasets into three primary types:
language-based, mul... | [
-0.3095016181,
-0.5919795036,
0.124259159,
0.3532368243,
0.1461047977,
0.3697567284,
-0.1530007273,
-0.7464552522,
0.2908690572,
0.148239553,
-1.0006738901,
-0.4948500693,
-0.5159561634,
-0.080743134,
-0.3481084406,
0.9763317704,
-0.2061000317,
0.2949729562,
-0.4980868399,
-0.5... |
http://arxiv.org/abs/2406.12251v1 | http://arxiv.org/pdf/2406.12251v1 | Mitigate Negative Transfer with Similarity Heuristic Lifelong Prompt Tuning | 2024-06-18T03:57:49 | 2024-06-18T03:57:49 | Chenyuan Wu, Gangwei Jiang, Defu Lian | cs.CL | cs.CL, cs.AI, cs.LG | Lifelong prompt tuning has significantly advanced parameter-efficient
lifelong learning with its efficiency and minimal storage demands on various
tasks. Our empirical studies, however, highlights certain transferability
constraints in the current methodologies: a universal algorithm that guarantees
consistent positive... | [
-0.3021096587,
-0.508587122,
0.4386049807,
0.2610283494,
-0.3303849697,
0.0867168978,
0.0143698491,
-0.3593972921,
0.1445457935,
0.2271805853,
-0.9934595227,
-0.1391461939,
-0.471457094,
-0.21892263,
-0.1520896256,
0.9523112178,
-0.0080247344,
-0.1580293924,
-0.2441104949,
-0.3... |
http://arxiv.org/abs/2406.12246v2 | http://arxiv.org/pdf/2406.12246v2 | TroL: Traversal of Layers for Large Language and Vision Models | 2024-06-18T03:42:00 | 2024-06-19T21:40:03 | Byung-Kwan Lee, Sangyun Chung, Chae Won Kim, Beomchan Park, Yong Man Ro | cs.LG | cs.LG, cs.CL, cs.CV | Large language and vision models (LLVMs) have been driven by the
generalization power of large language models (LLMs) and the advent of visual
instruction tuning. Along with scaling them up directly, these models enable
LLVMs to showcase powerful vision language (VL) performances by covering
diverse tasks via natural l... | [
-0.3066098392,
-0.4846278131,
0.5717635155,
0.2414258569,
-0.5600465536,
0.3584406078,
0.0044260141,
-0.7658684254,
-0.1602646261,
0.9609463811,
-0.4127394855,
-0.6134708524,
-0.4131507874,
-0.3424304724,
-0.3561069965,
0.5716854334,
0.0854106247,
-0.2013381869,
-0.0954649895,
... |
http://arxiv.org/abs/2406.12238v1 | http://arxiv.org/pdf/2406.12238v1 | PFID: Privacy First Inference Delegation Framework for LLMs | 2024-06-18T03:27:09 | 2024-06-18T03:27:09 | Haoyan Yang, Zhitao Li, Yong Zhang, Jianzong Wang, Ning Cheng, Ming Li, Jing Xiao | cs.CL | cs.CL | This paper introduces a novel privacy-preservation framework named PFID for
LLMs that addresses critical privacy concerns by localizing user data through
model sharding and singular value decomposition. When users are interacting
with LLM systems, their prompts could be subject to being exposed to
eavesdroppers within ... | [
-0.1616189331,
-0.8462510705,
0.5259479284,
0.3353956044,
-0.1310536563,
-0.1127277389,
0.160022065,
-0.1175681502,
0.1028720662,
0.6941256523,
-0.5036478043,
-0.6182989478,
-0.3436153829,
-0.2210089117,
-0.3748972416,
0.6741536856,
-0.0297192261,
-0.0175961498,
-0.5835548043,
... |
http://arxiv.org/abs/2406.12233v1 | http://arxiv.org/pdf/2406.12233v1 | SyncVSR: Data-Efficient Visual Speech Recognition with End-to-End Crossmodal Audio Token Synchronization | 2024-06-18T03:14:22 | 2024-06-18T03:14:22 | Young Jin Ahn, Jungwoo Park, Sangha Park, Jonghyun Choi, Kee-Eung Kim | cs.AI | cs.AI, cs.CL, cs.CV | Visual Speech Recognition (VSR) stands at the intersection of computer vision
and speech recognition, aiming to interpret spoken content from visual cues. A
prominent challenge in VSR is the presence of homophenes-visually similar lip
gestures that represent different phonemes. Prior approaches have sought to
distingui... | [
-0.0328131616,
-0.438557893,
0.0395923518,
-0.0479178876,
-0.4945423603,
0.0657688454,
-0.040866185,
-0.3210781217,
0.2350326329,
0.0507552885,
-0.4416778982,
-0.3710588217,
-0.5078270435,
-0.1138816178,
-0.687777102,
0.5164123774,
-0.0400154889,
0.3307713568,
-0.5667728186,
-0... |
http://arxiv.org/abs/2406.12232v1 | http://arxiv.org/pdf/2406.12232v1 | "You Gotta be a Doctor, Lin": An Investigation of Name-Based Bias of Large Language Models in Employment Recommendations | 2024-06-18T03:11:43 | 2024-06-18T03:11:43 | Huy Nghiem, John Prindle, Jieyu Zhao, Hal Daumé III | cs.AI | cs.AI, cs.CL | Social science research has shown that candidates with names indicative of
certain races or genders often face discrimination in employment practices.
Similarly, Large Language Models (LLMs) have demonstrated racial and gender
biases in various applications. In this study, we utilize GPT-3.5-Turbo and
Llama 3-70B-Instr... | [
-0.1077439934,
-0.3307189941,
0.9007693529,
-0.1386716962,
-0.0835531428,
-0.0209451951,
-0.1309215426,
-0.6317163706,
0.5666264296,
0.3701393902,
-0.4831427932,
-0.2290308475,
-0.7895553708,
-0.0672794655,
-0.4285692573,
1.0621620417,
0.328071028,
0.6066262722,
0.1781440079,
0... |
http://arxiv.org/abs/2406.12230v1 | http://arxiv.org/pdf/2406.12230v1 | MCSD: An Efficient Language Model with Diverse Fusion | 2024-06-18T03:08:01 | 2024-06-18T03:08:01 | Hua Yang, Duohai Li, Shiman Li | cs.CL | cs.CL, cs.AI | Transformers excel in Natural Language Processing (NLP) due to their prowess
in capturing long-term dependencies but suffer from exponential resource
consumption with increasing sequence lengths. To address these challenges, we
propose MCSD model, an efficient language model with linear scaling and fast
inference speed... | [
-0.6943216324,
-0.8420631886,
0.3602204025,
0.2349560857,
-0.0540595464,
0.3976571858,
-0.1615677774,
-0.3270709217,
0.2098591924,
0.5672050714,
-0.6444586515,
-0.4536091089,
-0.6010228992,
-0.1374083757,
-0.4149491191,
0.8370666504,
0.3902262151,
-0.1181195974,
-0.0211210009,
... |
http://arxiv.org/abs/2406.12223v1 | http://arxiv.org/pdf/2406.12223v1 | ToxiCloakCN: Evaluating Robustness of Offensive Language Detection in Chinese with Cloaking Perturbations | 2024-06-18T02:44:56 | 2024-06-18T02:44:56 | Yunze Xiao, Yujia Hu, Kenny Tsu Wei Choo, Roy Ka-wei Lee | cs.CL | cs.CL, cs.CY | Detecting hate speech and offensive language is essential for maintaining a
safe and respectful digital environment. This study examines the limitations of
state-of-the-art large language models (LLMs) in identifying offensive content
within systematically perturbed data, with a focus on Chinese, a language
particularl... | [
-0.0347197689,
-1.0496404171,
0.0063267383,
0.2144743502,
-0.4546781182,
0.1055227071,
-0.1308407485,
-0.6268938184,
-0.0659370497,
0.64903754,
-0.2665204406,
-0.6108970046,
-0.8400906324,
0.0922459289,
-0.5583461523,
0.8617831469,
0.1985972971,
0.0089963581,
0.5208500624,
-0.2... |
http://arxiv.org/abs/2406.12221v1 | http://arxiv.org/pdf/2406.12221v1 | On-Policy Fine-grained Knowledge Feedback for Hallucination Mitigation | 2024-06-18T02:43:49 | 2024-06-18T02:43:49 | Xueru Wen, Xinyu Lu, Xinyan Guan, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Le Sun | cs.CL | cs.CL | Hallucination occurs when large language models (LLMs) exhibit behavior that
deviates from the boundaries of their knowledge during the response generation
process. Previous learning-based methods focus on detecting knowledge
boundaries and finetuning models with instance-level feedback, but they suffer
from inaccurate... | [
-0.5087830424,
-1.2320599556,
0.5677331686,
0.1637101769,
0.1557707638,
-0.1310216039,
-0.4498018622,
-0.4006104767,
0.1981409043,
0.6199948788,
-0.717703104,
-0.2838586271,
-0.1246350259,
-0.1907391697,
0.0663056672,
1.0706710815,
0.1123277321,
0.0307087973,
-0.35273844,
0.127... |
http://arxiv.org/abs/2406.12216v1 | http://arxiv.org/pdf/2406.12216v1 | Is persona enough for personality? Using ChatGPT to reconstruct an agent's latent personality from simple descriptions | 2024-06-18T02:32:57 | 2024-06-18T02:32:57 | Yongyi Ji, Zhisheng Tang, Mayank Kejriwal | cs.CL | cs.CL, cs.AI | Personality, a fundamental aspect of human cognition, contains a range of
traits that influence behaviors, thoughts, and emotions. This paper explores
the capabilities of large language models (LLMs) in reconstructing these
complex cognitive attributes based only on simple descriptions containing
socio-demographic and ... | [
-0.3051172495,
-0.5060904026,
0.9430369735,
0.4508258104,
0.0249318574,
0.2617184222,
-0.440988034,
-0.3029722869,
0.3362399042,
0.7058078647,
-0.5761110783,
-0.0382314958,
-0.3821118176,
-0.1416287869,
-0.1277515739,
1.1513082981,
0.4504074156,
-0.0523678511,
-0.2708400786,
0.... |
http://arxiv.org/abs/2406.12213v3 | http://arxiv.org/pdf/2406.12213v3 | LLM-Oracle Machines | 2024-06-18T02:25:33 | 2024-07-03T12:59:21 | Jie Wang | cs.CL | cs.CL, cs.AI, cs.FL, F.1.1; F.4.1; I.2.0 | Contemporary AI applications leverage large language models (LLMs) to harness
their knowledge and reasoning abilities for natural language processing tasks.
This approach shares similarities with the concept of oracle Turing machines
(OTMs). To capture the broader potential of these computations, including those
not ye... | [
-0.3973117471,
-1.1126585007,
1.063121438,
0.0561295301,
0.0312631913,
0.0436141901,
-0.087523587,
-0.8788560629,
0.4407599568,
0.9242337346,
-0.6789339185,
-0.3894539475,
-0.4775512218,
0.4337551892,
-0.5695317388,
0.5779678822,
0.0839836001,
-0.1933869421,
-0.2395963967,
-0.0... |
http://arxiv.org/abs/2406.12209v1 | http://arxiv.org/pdf/2406.12209v1 | Interface Design for Self-Supervised Speech Models | 2024-06-18T02:13:48 | 2024-06-18T02:13:48 | Yi-Jen Shih, David Harwath | cs.SD | cs.SD, cs.CL, eess.AS | Self-supervised speech (SSL) models have recently become widely adopted for
many downstream speech processing tasks. The general usage pattern is to employ
SSL models as feature extractors, and then train a downstream prediction head
to solve a specific task. However, different layers of SSL models have been
shown to c... | [
-0.61267519,
-0.4021920562,
0.4020099044,
0.27595222,
-0.2317148596,
-0.0880719274,
0.2588335276,
-0.7334463596,
0.3485040665,
0.5588587523,
-0.5536281466,
-0.0884261131,
-0.4401593804,
-0.3346762061,
-0.6206191182,
0.9550172687,
0.25817734,
-0.0838074461,
-0.2771463692,
-0.053... |
http://arxiv.org/abs/2406.12208v1 | http://arxiv.org/pdf/2406.12208v1 | Knowledge Fusion By Evolving Weights of Language Models | 2024-06-18T02:12:34 | 2024-06-18T02:12:34 | Guodong Du, Jing Li, Hanting Liu, Runhua Jiang, Shuyang Yu, Yifei Guo, Sim Kuan Goh, Ho-Kin Tang | cs.CL | cs.CL, cs.AI, cs.CV, cs.NE | Fine-tuning pre-trained language models, particularly large language models,
demands extensive computing resources and can result in varying performance
outcomes across different domains and datasets. This paper examines the
approach of integrating multiple models from diverse training scenarios into a
unified model. T... | [
-0.4296179414,
-0.4865263104,
0.2433229536,
-0.1698131859,
-0.0298063979,
0.3167598844,
-0.3591941893,
-0.4653175473,
-0.0850731954,
0.5411432981,
-0.3997200727,
-0.1378030181,
-0.634552896,
-0.2223141342,
-0.4033521712,
0.795563221,
0.1517250687,
0.1713836193,
-0.531933248,
-0... |
http://arxiv.org/abs/2406.12197v1 | http://arxiv.org/pdf/2406.12197v1 | Debate as Optimization: Adaptive Conformal Prediction and Diverse Retrieval for Event Extraction | 2024-06-18T01:53:49 | 2024-06-18T01:53:49 | Sijia Wang, Lifu Huang | cs.CL | cs.CL, cs.AI | We propose a multi-agent debate as optimization (DAO) system for event
extraction, where the primary objective is to iteratively refine the large
language models (LLMs) outputs through debating without parameter tuning. In
DAO, we introduce two novel modules: the Diverse-RAG (DRAG) module and the
Adaptive Conformal Pre... | [
-0.666230619,
-0.8211539388,
0.4621621966,
0.0062624998,
-0.3494832218,
0.1722619534,
-0.4427641034,
-0.5518194437,
0.3661040068,
0.3882882893,
-0.7075029612,
-0.1349303424,
-0.6002271175,
0.1047785878,
-0.0246793088,
0.9748535156,
0.1234384775,
0.0763171017,
-0.0779651031,
-0.... |
http://arxiv.org/abs/2406.12182v1 | http://arxiv.org/pdf/2406.12182v1 | Aqulia-Med LLM: Pioneering Full-Process Open-Source Medical Language Models | 2024-06-18T01:30:07 | 2024-06-18T01:30:07 | Lulu Zhao, Weihao Zeng, Xiaofeng Shi, Hua Zhou, Donglin Hao, Yonghua Lin | cs.CL | cs.CL, cs.AI | Recently, both closed-source LLMs and open-source communities have made
significant strides, outperforming humans in various general domains. However,
their performance in specific professional fields such as medicine, especially
within the open-source community, remains suboptimal due to the complexity of
medical know... | [
0.0370672718,
-0.5929199457,
0.6496972442,
0.4105385542,
-0.3395064175,
0.0681945011,
-0.5329299569,
-0.3788669109,
0.3851181567,
0.5572871566,
-0.7262613773,
-0.5203318596,
-0.4465534985,
0.0787580311,
0.0649145246,
0.8372359872,
-0.3670288324,
0.8665833473,
0.0112942178,
-0.4... |
http://arxiv.org/abs/2406.12168v2 | http://arxiv.org/pdf/2406.12168v2 | BPO: Supercharging Online Preference Learning by Adhering to the Proximity of Behavior LLM | 2024-06-18T00:41:40 | 2024-06-19T05:25:27 | Wenda Xu, Jiachen Li, William Yang Wang, Lei Li | cs.LG | cs.LG, cs.AI, cs.CL | Direct alignment from preferences (DAP) has emerged as a promising paradigm
for aligning large language models (LLMs) to human desiderata from
pre-collected, offline preference datasets. While recent studies indicate that
existing offline DAP methods can directly benefit from online training samples,
we highlight the n... | [
-0.5664548278,
-0.4328991771,
0.6268063784,
-0.0928820893,
-0.0896485671,
0.0332807489,
-0.2140299976,
-0.6143444777,
0.3798556328,
0.2457521558,
-0.671777308,
-0.0871408731,
-0.4960431755,
-0.0388401076,
0.3333554268,
0.9559008479,
0.1442387551,
0.0913457051,
-0.3825278282,
0.... |
http://arxiv.org/abs/2406.12165v1 | http://arxiv.org/pdf/2406.12165v1 | Statistical Uncertainty in Word Embeddings: GloVe-V | 2024-06-18T00:35:02 | 2024-06-18T00:35:02 | Andrea Vallebueno, Cassandra Handan-Nader, Christopher D. Manning, Daniel E. Ho | cs.CL | cs.CL | Static word embeddings are ubiquitous in computational social science
applications and contribute to practical decision-making in a variety of fields
including law and healthcare. However, assessing the statistical uncertainty in
downstream conclusions drawn from word embedding statistics has remained
challenging. When... | [
-0.1589565128,
-0.9659944773,
0.5931869149,
0.1265976578,
-0.2014659196,
0.1359587908,
-0.0575299151,
-0.1251094043,
0.285390228,
0.18665497,
-0.3426093459,
-0.7591894865,
-0.9839764833,
-0.1567990184,
-0.3272871375,
0.3452005684,
-0.1640578508,
0.2757924199,
-0.353782177,
-0.2... |
http://arxiv.org/abs/2406.12159v1 | http://arxiv.org/pdf/2406.12159v1 | Exploring the Impact of a Transformer's Latent Space Geometry on Downstream Task Performance | 2024-06-18T00:17:30 | 2024-06-18T00:17:30 | Anna C. Marbut, John W. Chandler, Travis J. Wheeler | cs.CL | cs.CL, cs.LG | It is generally thought that transformer-based large language models benefit
from pre-training by learning generic linguistic knowledge that can be focused
on a specific task during fine-tuning. However, we propose that much of the
benefit from pre-training may be captured by geometric characteristics of the
latent spa... | [
-0.3754065931,
-0.5849965811,
0.5798467398,
0.0703609586,
-0.0535625815,
-0.1893212348,
-0.0812606066,
-0.1599726081,
0.1549926549,
0.2686948776,
-0.5588156581,
-0.2093279213,
-0.7785226703,
-0.4174250662,
-0.3245134056,
0.9719665051,
-0.1152122989,
-0.2344004363,
-0.3803033233,
... |
http://arxiv.org/abs/2406.12158v1 | http://arxiv.org/pdf/2406.12158v1 | LLMs Are Prone to Fallacies in Causal Inference | 2024-06-18T00:14:07 | 2024-06-18T00:14:07 | Nitish Joshi, Abulhair Saparov, Yixin Wang, He He | cs.CL | cs.CL, cs.AI | Recent work shows that causal facts can be effectively extracted from LLMs
through prompting, facilitating the creation of causal graphs for causal
inference tasks. However, it is unclear if this success is limited to
explicitly-mentioned causal facts in the pretraining data which the model can
memorize. Thus, this wor... | [
-0.046535939,
-1.0880427361,
1.1242153645,
0.1228943914,
-0.1933643818,
-0.2870341241,
0.2196269333,
-0.5431244373,
0.1127997264,
0.4931131899,
-0.7856307626,
-0.2903864682,
-0.5664741993,
0.1521194279,
-0.3694332242,
0.9076887369,
-0.06504336,
0.2503353953,
-0.5580375195,
-0.0... |
http://arxiv.org/abs/2406.15487v1 | http://arxiv.org/pdf/2406.15487v1 | Improving Text-To-Audio Models with Synthetic Captions | 2024-06-18T00:02:15 | 2024-06-18T00:02:15 | Zhifeng Kong, Sang-gil Lee, Deepanway Ghosal, Navonil Majumder, Ambuj Mehrish, Rafael Valle, Soujanya Poria, Bryan Catanzaro | cs.CL | cs.CL, cs.LG, cs.SD, eess.AS | It is an open challenge to obtain high quality training data, especially
captions, for text-to-audio models. Although prior methods have leveraged
\textit{text-only language models} to augment and improve captions, such
methods have limitations related to scale and coherence between audio and
captions. In this work, we... | [
-0.504663229,
-0.4415787756,
0.3566000462,
0.1475301981,
-0.332316041,
0.5246298909,
-0.3254705667,
-0.4463126361,
0.4886875749,
0.4838159382,
-0.9228663445,
-0.4123079479,
-0.4747800231,
0.2906636596,
-0.728987813,
1.0037420988,
-0.0013563191,
0.2558592856,
-0.1219660416,
-0.1... |
http://arxiv.org/abs/2406.12141v1 | http://arxiv.org/pdf/2406.12141v1 | A dual task learning approach to fine-tune a multilingual semantic speech encoder for Spoken Language Understanding | 2024-06-17T23:07:53 | 2024-06-17T23:07:53 | Gaëlle Laperrière, Sahar Ghannay, Bassam Jabaian, Yannick Estève | cs.CL | cs.CL, cs.SD, eess.AS | Self-Supervised Learning is vastly used to efficiently represent speech for
Spoken Language Understanding, gradually replacing conventional approaches.
Meanwhile, textual SSL models are proposed to encode language-agnostic
semantics. SAMU-XLSR framework employed this semantic information to enrich
multilingual speech r... | [
-0.1469500214,
-0.5047369599,
0.4112090468,
-0.1760983169,
-0.3960265815,
0.1744570732,
-0.4737823308,
-0.5914135575,
0.1137378663,
0.5007254481,
-0.7458643913,
-0.2491178513,
-0.6758844852,
0.1424509883,
0.198656857,
0.9058489203,
0.0498475581,
0.4277909994,
-0.1670318544,
-0.... |
http://arxiv.org/abs/2406.12138v1 | http://arxiv.org/pdf/2406.12138v1 | Bias in Text Embedding Models | 2024-06-17T22:58:36 | 2024-06-17T22:58:36 | Vasyl Rakivnenko, Nestor Maslej, Jessica Cervi, Volodymyr Zhukov | cs.AI | cs.AI, cs.CL | Text embedding is becoming an increasingly popular AI methodology, especially
among businesses, yet the potential of text embedding models to be biased is
not well understood. This paper examines the degree to which a selection of
popular text embedding models are biased, particularly along gendered
dimensions. More sp... | [
-0.1454959065,
-0.5979396701,
0.6078453064,
0.3416167498,
0.033358302,
-0.0237426907,
0.1819419116,
-0.5527012348,
-0.0717774555,
0.490883708,
-0.6217738986,
-0.3630359769,
-0.9180060029,
0.1976137012,
-0.2964016795,
1.0175340176,
0.1275061816,
0.6007446051,
-0.6877073646,
-0.3... |
http://arxiv.org/abs/2406.12131v1 | http://arxiv.org/pdf/2406.12131v1 | Gram2Vec: An Interpretable Document Vectorizer | 2024-06-17T22:42:14 | 2024-06-17T22:42:14 | Peter Zeng, Eric Sclafani, Owen Rambow | cs.CL | cs.CL | We present Gram2Vec, a grammatical style embedding algorithm that embeds
documents into a higher dimensional space by extracting the normalized relative
frequencies of grammatical features present in the text. Compared to neural
approaches, Gram2Vec offers inherent interpretability based on how the feature
vectors are ... | [
-0.1721354574,
-0.5921775699,
0.876942277,
0.0972564071,
-0.0194004271,
-0.0583011694,
-0.2360318452,
-0.2856282294,
0.3228605092,
0.2078951746,
-0.4608268738,
-0.70897609,
-0.9289343953,
-0.1405645311,
-0.5007347465,
0.7400990725,
0.0752684027,
0.0727805123,
-0.9056591392,
-0.... |
http://arxiv.org/abs/2406.12128v1 | http://arxiv.org/pdf/2406.12128v1 | AI "News" Content Farms Are Easy to Make and Hard to Detect: A Case Study in Italian | 2024-06-17T22:19:00 | 2024-06-17T22:19:00 | Giovanni Puccetti, Anna Rogers, Chiara Alzetta, Felice Dell'Orletta, Andrea Esuli | cs.CL | cs.CL | Large Language Models (LLMs) are increasingly used as "content farm" models
(CFMs), to generate synthetic text that could pass for real news articles. This
is already happening even for languages that do not have high-quality
monolingual LLMs. We show that fine-tuning Llama (v1), mostly trained on
English, on as little... | [
-0.4588137269,
-0.7392084002,
0.5659080744,
0.1219331101,
-0.4566921294,
0.2669976056,
-0.3299876153,
-0.7887620926,
0.3744677007,
0.4777883887,
-0.734521389,
-0.6304520965,
-0.7053153515,
0.4901334047,
-0.9188170433,
1.2257304192,
0.1237027422,
-0.0843228102,
0.0052075898,
-0.... |
http://arxiv.org/abs/2406.12125v1 | http://arxiv.org/pdf/2406.12125v1 | Efficient Sequential Decision Making with Large Language Models | 2024-06-17T22:13:22 | 2024-06-17T22:13:22 | Dingyang Chen, Qi Zhang, Yinglun Zhu | cs.LG | cs.LG, cs.CL | This paper focuses on extending the success of large language models (LLMs)
to sequential decision making. Existing efforts either (i) re-train or finetune
LLMs for decision making, or (ii) design prompts for pretrained LLMs. The
former approach suffers from the computational burden of gradient updates, and
the latter ... | [
-0.2758489549,
-0.709913671,
0.8731841445,
0.5752594471,
-0.1085379645,
0.3569169939,
-0.59464252,
-0.3394435346,
0.1021127775,
0.4866367579,
-0.6097055674,
-0.3559484482,
-0.3123579919,
0.2082492262,
-0.237508595,
0.5541698337,
0.0390942916,
0.2694151402,
0.0276556127,
0.04773... |
http://arxiv.org/abs/2406.12117v1 | http://arxiv.org/pdf/2406.12117v1 | Decoding the Narratives: Analyzing Personal Drug Experiences Shared on Reddit | 2024-06-17T21:56:57 | 2024-06-17T21:56:57 | Layla Bouzoubaa, Elham Aghakhani, Max Song, Minh Trinh, Rezvaneh Rezapour | cs.CL | cs.CL | Online communities such as drug-related subreddits serve as safe spaces for
people who use drugs (PWUD), fostering discussions on substance use
experiences, harm reduction, and addiction recovery. Users' shared narratives
on these forums provide insights into the likelihood of developing a substance
use disorder (SUD) ... | [
-0.3798598051,
-0.6675958037,
0.7596011162,
0.1876327693,
0.0026065211,
0.2383116335,
0.0579865798,
-0.7462482452,
0.3774664998,
0.5821440816,
-0.4741318226,
-0.9108029008,
-0.9288399816,
0.3584493697,
-0.2493254989,
1.5500888824,
-0.2191214561,
-0.1955801696,
-0.0699079409,
-0... |
http://arxiv.org/abs/2406.12114v1 | http://arxiv.org/pdf/2406.12114v1 | Enhancing Text Classification through LLM-Driven Active Learning and Human Annotation | 2024-06-17T21:45:48 | 2024-06-17T21:45:48 | Hamidreza Rouzegar, Masoud Makrehchi | cs.CL | cs.CL, cs.AI, cs.LG, 68T50, I.2.7 | In the context of text classification, the financial burden of annotation
exercises for creating training data is a critical issue. Active learning
techniques, particularly those rooted in uncertainty sampling, offer a
cost-effective solution by pinpointing the most instructive samples for manual
annotation. Similarly,... | [
0.0448400714,
-0.648278296,
0.463391304,
-0.0610694289,
-0.2714776397,
0.1936378032,
-0.3683559299,
-0.7695431113,
-0.2555817664,
0.5128957629,
-0.0914732367,
-0.4870477319,
-0.3478068709,
0.0724654272,
-0.3109591901,
1.0086599588,
-0.4895354509,
0.0034477788,
0.0610237382,
-0.... |
http://arxiv.org/abs/2406.12109v1 | http://arxiv.org/pdf/2406.12109v1 | Can LLMs Learn Macroeconomic Narratives from Social Media? | 2024-06-17T21:37:09 | 2024-06-17T21:37:09 | Almog Gueta, Amir Feder, Zorik Gekhman, Ariel Goldstein, Roi Reichart | cs.CL | cs.CL, cs.CE | This study empirically tests the $\textit{Narrative Economics}$ hypothesis,
which posits that narratives (ideas that are spread virally and affect public
beliefs) can influence economic fluctuations. We introduce two curated datasets
containing posts from X (formerly Twitter) which capture economy-related
narratives (D... | [
-0.2177867591,
-1.0544832945,
0.7201854587,
0.2513239384,
-0.4112802446,
0.4651578367,
-0.1411765963,
-0.7447482944,
0.6436052918,
0.5733317137,
-0.7189412713,
-0.4599235058,
-0.6956709623,
0.1326245517,
-0.3532796204,
1.0188772678,
0.1584384292,
-0.2468692213,
-0.0049398821,
0... |
http://arxiv.org/abs/2406.12104v1 | http://arxiv.org/pdf/2406.12104v1 | End-to-end Text-to-SQL Generation within an Analytics Insight Engine | 2024-06-17T21:33:01 | 2024-06-17T21:33:01 | Karime Maamari, Amine Mhedhbi | cs.CL | cs.CL, cs.AI, cs.DB, cs.LG | Recent advancements in Text-to-SQL have pushed database management systems
towards greater democratization of data access. Today's language models are at
the core of these advancements. They enable impressive Text-to-SQL generation
as experienced in the development of Distyl AI's Analytics Insight Engine. Its
early dep... | [
-0.2396319211,
-0.8218050003,
0.6624807119,
0.0465992019,
-0.199430719,
0.1649887562,
-0.0975307524,
-0.4947988689,
0.1195011586,
0.5180294514,
-1.0053297281,
-0.5291337371,
-0.3943827748,
0.1707584858,
-0.4754730165,
1.1169711351,
0.2918060422,
-0.1706410348,
-0.3949711323,
0.... |
http://arxiv.org/abs/2406.12094v1 | http://arxiv.org/pdf/2406.12094v1 | Who's asking? User personas and the mechanics of latent misalignment | 2024-06-17T21:15:12 | 2024-06-17T21:15:12 | Asma Ghandeharioun, Ann Yuan, Marius Guerard, Emily Reif, Michael A. Lepori, Lucas Dixon | cs.CL | cs.CL, cs.AI | Despite investments in improving model safety, studies show that misaligned
capabilities remain latent in safety-tuned models. In this work, we shed light
on the mechanics of this phenomenon. First, we show that even when model
generations are safe, harmful content can persist in hidden representations and
can be extra... | [
-0.2264152169,
-0.7444691658,
0.7108077407,
0.136926204,
0.105101712,
0.0801217183,
0.4320004284,
-0.5927485824,
-0.125774622,
0.663197875,
-0.7537274361,
0.180740118,
-0.5387117267,
-0.1839096546,
-0.2554041147,
0.9314032197,
0.4437652528,
0.0050043617,
-0.2758048475,
0.089291... |
http://arxiv.org/abs/2406.12091v2 | http://arxiv.org/pdf/2406.12091v2 | Is poisoning a real threat to LLM alignment? Maybe more so than you think | 2024-06-17T21:06:00 | 2024-06-19T17:56:17 | Pankayaraj Pathmanathan, Souradip Chakraborty, Xiangyu Liu, Yongyuan Liang, Furong Huang | cs.LG | cs.LG, cs.CL, cs.CR | Recent advancements in Reinforcement Learning with Human Feedback (RLHF) have
significantly impacted the alignment of Large Language Models (LLMs). The
sensitivity of reinforcement learning algorithms such as Proximal Policy
Optimization (PPO) has led to new line work on Direct Policy Optimization
(DPO), which treats R... | [
0.0282128323,
-0.6221215129,
0.579046011,
0.0091822445,
0.1600678116,
-0.037821319,
0.0347306281,
-0.3806845546,
-0.3015056551,
0.8324315548,
-0.6452638507,
-0.3935510218,
-0.3422668874,
-0.0341972299,
-0.075057894,
1.0322757959,
0.1150185615,
-0.1026337072,
-0.024129048,
0.164... |
http://arxiv.org/abs/2406.12084v1 | http://arxiv.org/pdf/2406.12084v1 | When Reasoning Meets Information Aggregation: A Case Study with Sports Narratives | 2024-06-17T20:49:35 | 2024-06-17T20:49:35 | Yebowen Hu, Kaiqiang Song, Sangwoo Cho, Xiaoyang Wang, Wenlin Yao, Hassan Foroosh, Dong Yu, Fei Liu | cs.CL | cs.CL, cs.AI | Reasoning is most powerful when an LLM accurately aggregates relevant
information. We examine the critical role of information aggregation in
reasoning by requiring the LLM to analyze sports narratives. To succeed at this
task, an LLM must infer points from actions, identify related entities,
attribute points accuratel... | [
-0.2471360117,
-0.8978893161,
1.0049412251,
0.1360600889,
-0.1984641999,
0.3110970855,
0.1547011733,
-0.6736109853,
0.0143453442,
0.1309775412,
-0.3522591591,
-0.4879389405,
-0.7149367332,
-0.0624680966,
-0.4422495961,
0.7538025379,
-0.2467080206,
-0.121790722,
-0.5844677091,
-... |
http://arxiv.org/abs/2406.12074v1 | http://arxiv.org/pdf/2406.12074v1 | COMMUNITY-CROSS-INSTRUCT: Unsupervised Instruction Generation for Aligning Large Language Models to Online Communities | 2024-06-17T20:20:47 | 2024-06-17T20:20:47 | Zihao He, Rebecca Dorn, Siyi Guo, Minh Duc Chu, Kristina Lerman | cs.CL | cs.CL | Social scientists use surveys to probe the opinions and beliefs of
populations, but these methods are slow, costly, and prone to biases. Recent
advances in large language models (LLMs) enable creating computational
representations or "digital twins" of populations that generate human-like
responses mimicking the popula... | [
-0.4876997471,
-0.5639490485,
0.5875973701,
-0.1007754058,
-0.1313222498,
0.5974447727,
-0.5340625048,
-0.1948463321,
0.6203243732,
0.4480758309,
-0.5042006373,
-0.5802484155,
-0.5834769607,
-0.3133607805,
-0.4649567008,
0.8742007017,
0.1749842912,
0.3065263331,
-0.6172260642,
... |
http://arxiv.org/abs/2406.12069v1 | http://arxiv.org/pdf/2406.12069v1 | Satyrn: A Platform for Analytics Augmented Generation | 2024-06-17T20:14:16 | 2024-06-17T20:14:16 | Marko Sterbentz, Cameron Barrie, Shubham Shahi, Abhratanu Dutta, Donna Hooshmand, Harper Pack, Kristian J. Hammond | cs.CL | cs.CL | Large language models (LLMs) are capable of producing documents, and
retrieval augmented generation (RAG) has shown itself to be a powerful method
for improving accuracy without sacrificing fluency. However, not all
information can be retrieved from text. We propose an approach that uses the
analysis of structured data... | [
-0.0947126821,
-0.8625720143,
0.5703908801,
0.21302028,
-0.1060542911,
0.2698630989,
-0.5145446062,
-0.3559931219,
0.8047480583,
0.6743362546,
-0.3897040486,
-0.5988800526,
-0.4323556423,
0.0507945828,
-0.442935288,
1.2795990705,
0.1845998913,
0.0892187357,
-0.3127659261,
-0.00... |
http://arxiv.org/abs/2406.12066v2 | http://arxiv.org/pdf/2406.12066v2 | Language Models are Surprisingly Fragile to Drug Names in Biomedical Benchmarks | 2024-06-17T20:09:24 | 2024-06-19T03:59:41 | Jack Gallifant, Shan Chen, Pedro Moreira, Nikolaj Munch, Mingye Gao, Jackson Pond, Leo Anthony Celi, Hugo Aerts, Thomas Hartvigsen, Danielle Bitterman | cs.CL | cs.CL | Medical knowledge is context-dependent and requires consistent reasoning
across various natural language expressions of semantically equivalent phrases.
This is particularly crucial for drug names, where patients often use brand
names like Advil or Tylenol instead of their generic equivalents. To study
this, we create ... | [
0.1455884278,
-0.6559506059,
0.5263931751,
-0.1363982409,
0.0183289312,
0.2365332246,
-0.2579194307,
-0.7983614802,
-0.1487935781,
0.4336665273,
-0.2991248071,
-0.4524292648,
-1.031580925,
0.3457994163,
-0.2454990149,
1.037592411,
-0.2338998616,
0.3238812983,
-0.2079607844,
-0.... |
http://arxiv.org/abs/2406.12060v1 | http://arxiv.org/pdf/2406.12060v1 | Not Eliminate but Aggregate: Post-Hoc Control over Mixture-of-Experts to Address Shortcut Shifts in Natural Language Understanding | 2024-06-17T20:00:04 | 2024-06-17T20:00:04 | Ukyo Honda, Tatsushi Oka, Peinan Zhang, Masato Mita | cs.CL | cs.CL, cs.LG | Recent models for natural language understanding are inclined to exploit
simple patterns in datasets, commonly known as shortcuts. These shortcuts hinge
on spurious correlations between labels and latent features existing in the
training data. At inference time, shortcut-dependent models are likely to
generate erroneou... | [
-0.3291302025,
-1.0223915577,
0.570366025,
0.0648836419,
-0.4789965451,
0.3077371716,
-0.1422483623,
-0.3059121668,
0.345328927,
0.6432911158,
-0.4239291847,
-0.4607926011,
-1.1229654551,
0.0851617455,
-0.4181503952,
1.2079681158,
0.0345066488,
-0.4933867753,
-0.1605225652,
-0.... |
http://arxiv.org/abs/2406.12058v3 | http://arxiv.org/pdf/2406.12058v3 | WellDunn: On the Robustness and Explainability of Language Models and Large Language Models in Identifying Wellness Dimensions | 2024-06-17T19:50:40 | 2024-06-28T04:08:12 | Seyedali Mohammadi, Edward Raff, Jinendra Malekar, Vedant Palit, Francis Ferraro, Manas Gaur | cs.AI | cs.AI, cs.CL | Language Models (LMs) are being proposed for mental health applications where
the heightened risk of adverse outcomes means predictive performance may not be
a sufficient litmus test of a model's utility in clinical practice. A model
that can be trusted for practice should have a correspondence between
explanation and ... | [
-0.2954874635,
-0.6931148171,
0.8857215643,
0.0847238302,
0.0017403855,
-0.1320241988,
-0.3233802617,
-0.5648424625,
0.0257500764,
0.1591050327,
-0.7234556675,
-0.5241872072,
-0.7370151281,
0.0510606617,
-0.0222333204,
1.0384539366,
0.2617094517,
0.0101345982,
-0.2743825316,
-0... |
http://arxiv.org/abs/2406.12053v1 | http://arxiv.org/pdf/2406.12053v1 | InternalInspector $I^2$: Robust Confidence Estimation in LLMs through Internal States | 2024-06-17T19:46:05 | 2024-06-17T19:46:05 | Mohammad Beigi, Ying Shen, Runing Yang, Zihao Lin, Qifan Wang, Ankith Mohan, Jianfeng He, Ming Jin, Chang-Tien Lu, Lifu Huang | cs.CL | cs.CL | Despite their vast capabilities, Large Language Models (LLMs) often struggle
with generating reliable outputs, frequently producing high-confidence
inaccuracies known as hallucinations. Addressing this challenge, our research
introduces InternalInspector, a novel framework designed to enhance confidence
estimation in L... | [
-0.3174098432,
-0.3955266774,
0.7156773806,
0.1118928716,
-0.0534878857,
0.296831578,
-0.315099448,
-0.5992876887,
0.1534167975,
0.2087233961,
-0.4312941432,
-0.4318730831,
-0.5865760446,
0.1380246729,
-0.3474722207,
0.3958410621,
0.0899679139,
0.3339124024,
-0.4955879748,
0.33... |
http://arxiv.org/abs/2406.12052v1 | http://arxiv.org/pdf/2406.12052v1 | UniGLM: Training One Unified Language Model for Text-Attributed Graphs | 2024-06-17T19:45:21 | 2024-06-17T19:45:21 | Yi Fang, Dongzhe Fan, Sirui Ding, Ninghao Liu, Qiaoyu Tan | cs.CL | cs.CL, cs.AI, cs.IR, cs.LG | Representation learning on text-attributed graphs (TAGs), where nodes are
represented by textual descriptions, is crucial for textual and relational
knowledge systems and recommendation systems. Currently, state-of-the-art
embedding methods for TAGs primarily focus on fine-tuning language models
(e.g., BERT) using stru... | [
-0.0468358062,
-0.4877000153,
0.255679965,
-0.0827898681,
-0.5836957097,
0.5691556334,
-0.0051535745,
-0.2077588588,
0.1047521606,
0.5046112537,
-0.2199161649,
-0.896912694,
-0.3404580951,
-0.2854735255,
-0.1484627128,
1.1551806927,
-0.1235801056,
0.1325164437,
-0.3020012379,
-... |
http://arxiv.org/abs/2406.12050v1 | http://arxiv.org/pdf/2406.12050v1 | Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning | 2024-06-17T19:42:22 | 2024-06-17T19:42:22 | Zhihan Zhang, Zhenwen Liang, Wenhao Yu, Dian Yu, Mengzhao Jia, Dong Yu, Meng Jiang | cs.CL | cs.CL | Supervised fine-tuning enhances the problem-solving abilities of language
models across various mathematical reasoning tasks. To maximize such benefits,
existing research focuses on broadening the training set with various data
augmentation techniques, which is effective for standard single-round
question-answering set... | [
-0.6986575127,
-0.8415898085,
0.6401121616,
-0.0212642588,
-0.0652125701,
0.2870259881,
-0.3910203874,
-0.364417702,
-0.0890021846,
0.8029794693,
-0.9156633615,
-0.2064932734,
-0.4672335684,
-0.0991806686,
-0.1471178532,
1.0523426533,
-0.0134750307,
0.3407360315,
-0.5410789847,
... |
http://arxiv.org/abs/2406.12045v1 | http://arxiv.org/pdf/2406.12045v1 | $τ$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains | 2024-06-17T19:33:08 | 2024-06-17T19:33:08 | Shunyu Yao, Noah Shinn, Pedram Razavi, Karthik Narasimhan | cs.AI | cs.AI, cs.CL | Existing benchmarks do not test language agents on their interaction with
human users or ability to follow domain-specific rules, both of which are vital
for deploying them in real world applications. We propose $\tau$-bench, a
benchmark emulating dynamic conversations between a user (simulated by language
models) and ... | [
-0.410605073,
-0.9718640447,
0.4667912722,
0.1087231115,
-0.0800730661,
0.4065762758,
-0.17815575,
-0.6946101785,
0.087266095,
0.1092365161,
-0.5409728885,
-0.3723369539,
-0.1661889553,
-0.0562622659,
-0.2664324343,
0.5770798922,
-0.1185775101,
-0.0833771974,
-0.1099137515,
-0.... |
http://arxiv.org/abs/2406.12038v1 | http://arxiv.org/pdf/2406.12038v1 | Soft Prompting for Unlearning in Large Language Models | 2024-06-17T19:11:40 | 2024-06-17T19:11:40 | Karuna Bhaila, Minh-Hao Van, Xintao Wu | cs.CL | cs.CL, cs.AI, cs.LG | The widespread popularity of Large Language Models (LLMs), partly due to
their unique ability to perform in-context learning, has also brought to light
the importance of ethical and safety considerations when deploying these
pre-trained models. In this work, we focus on investigating machine unlearning
for LLMs motivat... | [
0.0363436826,
-0.7198810577,
0.4219562709,
-0.1116086319,
-0.3927876055,
-0.1894582063,
-0.7466349006,
0.2103367448,
-0.3701810539,
0.7007257342,
-0.6687093377,
-0.4541378319,
-0.3695099354,
-0.06652987,
-0.2117521167,
1.3385545015,
0.1166399866,
-0.0100125168,
-0.3919450641,
-... |
http://arxiv.org/abs/2406.12036v4 | http://arxiv.org/pdf/2406.12036v4 | MedCalc-Bench: Evaluating Large Language Models for Medical Calculations | 2024-06-17T19:07:21 | 2024-06-30T15:12:10 | Nikhil Khandekar, Qiao Jin, Guangzhi Xiong, Soren Dunn, Serina S Applebaum, Zain Anwar, Maame Sarfo-Gyamfi, Conrad W Safranek, Abid A Anwar, Andrew Zhang, Aidan Gilson, Maxwell B Singer, Amisha Dave, Andrew Taylor, Aidong Zhang, Qingyu Chen, Zhiyong Lu | cs.CL | cs.CL, cs.AI | As opposed to evaluating computation and logic-based reasoning, current
benchmarks for evaluating large language models (LLMs) in medicine are
primarily focused on question-answering involving domain knowledge and
descriptive reasoning. While such qualitative capabilities are vital to medical
diagnosis, in real-world s... | [
-0.1254877597,
-0.6931762695,
1.1926929951,
0.296620369,
-0.1051794365,
-0.1249360293,
-0.3694680929,
-0.2398169339,
-0.0492852964,
0.3914113045,
-0.4625399411,
-0.7681150436,
-0.6575661898,
0.1218819246,
-0.2184176445,
0.5665224195,
-0.0802182481,
0.8002517819,
-0.6128386259,
... |
http://arxiv.org/abs/2406.12034v1 | http://arxiv.org/pdf/2406.12034v1 | Self-MoE: Towards Compositional Large Language Models with Self-Specialized Experts | 2024-06-17T19:06:54 | 2024-06-17T19:06:54 | Junmo Kang, Leonid Karlinsky, Hongyin Luo, Zhen Wang, Jacob Hansen, James Glass, David Cox, Rameswar Panda, Rogerio Feris, Alan Ritter | cs.CL | cs.CL, cs.LG | We present Self-MoE, an approach that transforms a monolithic LLM into a
compositional, modular system of self-specialized experts, named MiXSE (MiXture
of Self-specialized Experts). Our approach leverages self-specialization, which
constructs expert modules using self-generated synthetic data, each equipped
with a sha... | [
-0.5239217877,
-0.6892952919,
0.4995174706,
0.3572759032,
0.1798511893,
0.2298208028,
-0.4034211636,
-0.391536057,
0.6806899905,
0.6381005645,
-0.5344188809,
-0.3608502448,
-0.4891055524,
0.0556835085,
-0.3599374592,
0.6477968693,
0.2512806356,
-0.0806096643,
-0.2266620398,
0.0... |
http://arxiv.org/abs/2406.12033v2 | http://arxiv.org/pdf/2406.12033v2 | Unveiling and Mitigating Bias in Mental Health Analysis with Large Language Models | 2024-06-17T19:05:32 | 2024-06-19T18:28:22 | Yuqing Wang, Yun Zhao, Sara Alessandra Keller, Anne de Hond, Marieke M. van Buchem, Malvika Pillai, Tina Hernandez-Boussard | cs.CL | cs.CL | The advancement of large language models (LLMs) has demonstrated strong
capabilities across various applications, including mental health analysis.
However, existing studies have focused on predictive performance, leaving the
critical issue of fairness underexplored, posing significant risks to
vulnerable populations. ... | [
-0.358191222,
-0.7111181021,
0.9544341564,
0.2233427167,
0.153532505,
-0.1080885902,
-0.277592808,
-0.4207704961,
-0.1210225746,
0.3103136122,
-0.7924649715,
-0.3779721856,
-0.9342877865,
0.1040705219,
0.0619428866,
1.3294248581,
0.3063579202,
0.5553945899,
-0.0578323007,
-0.01... |
http://arxiv.org/abs/2406.12031v1 | http://arxiv.org/pdf/2406.12031v1 | Large Scale Transfer Learning for Tabular Data via Language Modeling | 2024-06-17T18:58:20 | 2024-06-17T18:58:20 | Josh Gardner, Juan C. Perdomo, Ludwig Schmidt | cs.LG | cs.LG, cs.AI, cs.CL | Tabular data -- structured, heterogeneous, spreadsheet-style data with rows
and columns -- is widely used in practice across many domains. However, while
recent foundation models have reduced the need for developing task-specific
datasets and predictors in domains such as language modeling and computer
vision, this tra... | [
-0.1835456789,
-0.879596889,
0.2938061357,
0.109384045,
0.0821363851,
-0.0162805263,
-0.3523477614,
-0.5592234731,
-0.1722940058,
0.8143846989,
-0.1335324794,
-0.6487967968,
-0.5107588768,
-0.2741244435,
-0.5868760943,
1.1144483089,
-0.1478399783,
0.0021229188,
0.1070239246,
0.... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.