paper_id stringlengths 9 58 | openreview_id stringlengths 9 58 | title stringlengths 5 214 ⌀ | conference stringclasses 4
values | year int64 2.01k 2.03k | pdf_sha256 stringlengths 64 64 | text stringlengths 1 1.3M | text_sha256 stringlengths 64 64 | source_text_sha256 stringlengths 64 64 | cleanup_version stringclasses 1
value | flag_count int64 0 2.73k | stats_json stringlengths 73 506 |
|---|---|---|---|---|---|---|---|---|---|---|---|
0UK8c2kg7c | 0UK8c2kg7c | InstructAV: Instruction Fine-tuning Large Language Models for Authorship Verification | colm | 2,024 | f71db4adab0775a79b42bb60f90d3ec2ef26f36d7a9806806b4db63b7b12623f | InstructAV: Instruction Fine-tuning Large Language Models
for Authorship Verification
Yujia Hu , ∗ Zhiqiang Hu , ∗ Chun-Wei Seah, Roy Ka-Wei Lee Information Systems Technology and Design Singapore University of Technology and Design
{yujia hu, chunwei seah, roy lee}@sutd.edu.sg
{zhiqiang hu}@mymail.sutd.edu.sg
Abstract... | 79526e561763915022cf7f06ecbe9ab9c70264f2841a322580e89f4a372711f9 | f44b2e21cbdf0b83ace86fb37ed7bf87eb65c49a5eda54fa15c6a912b3fcf014 | positioned-clean-v2 | 10 | {"removed_words": 144, "retained_words": 8614, "rules": {"dehyphenated": 15, "joined_hyphenated_wrap": 1, "page_number": 16, "repeated_review_header": 128}, "source_words": 8758} |
0VLBwQGWpA | 0VLBwQGWpA | ReAct Meets ActRe: Autonomous Annotation of Agent Trajectories for Contrastive Self-Training | colm | 2,024 | 3547184a04aed247ab08fa59e5b52a7f24962dfb5be57c4d0f2f2b9b56214df4 | ReAct Meets ActRe: Autonomous Annotation of Agent Trajectories for Contrastive Self-Training
Zonghan Yang 1 , Peng Li 2 ∗ , Ming Yan 3 , Ji Zhang 3 , Fei Huang 3 , Yang Liu 1,2,4 ∗
1 Department of Computer Science and Technology, Tsinghua University, Beijing, China 2 Institute for AI Industry Research (AIR), Tsinghua U... | a6bc8b85e14ce0ac2aad52b40091b6d4b7d8ac83d3e3fea4c07bd4544207e9e4 | da767f9135abb7ff128c4f8a6ecf3a705b4aac65a52d0e83750b3f68cb0f977f | positioned-clean-v2 | 15 | {"removed_words": 198, "retained_words": 10967, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 5, "page_number": 22, "repeated_review_header": 176}, "source_words": 11165} |
0o95CVdNuz | 0o95CVdNuz | Effective Prompt Extraction from Language Models | colm | 2,024 | eb4b7c46e113145b14975f26c2ae77a01fa9940e922c74baead2f98b5c28f7dc | Effective Prompt Extraction from Language Models
Yiming Zhang 1 ∗ Nicholas Carlini 2 Daphne Ippolito 1,2
2 Google DeepMind
1 Carnegie Mellon University
Abstract
The text generated by large language models is commonly controlled by prompting, where a prompt prepended to a user’s query guides the model’s output. The prom... | 4b18589e1cb366bd86a9f453708d7f254899f383252879199bfa8584acac02ae | 7a1b2193cf7d8bf0f055f55e86a7a2d6b77280641a264890d44b45a5ae104d13 | positioned-clean-v2 | 9 | {"removed_words": 248, "retained_words": 12718, "rules": {"combined_negation": 6, "composed_negation": 6, "dehyphenated": 11, "joined_hyphenated_wrap": 4, "joined_small_caps": 28, "page_number": 26, "repeated_review_header": 208, "separated_prose_lane": 1, "unmapped_glyph": 8}, "source_words": 12966} |
0oiG1KigYN | 0oiG1KigYN | SPEER: Sentence-Level Planning of Long Clinical Summaries via Embedded Entity Retrieval | colm | 2,024 | f62737322b783a4106e5140ab77243486d9a5165e727cec2fc248a9bc03ada52 | SPEER: Sentence-Level Planning of Long Clinical Summaries via Embedded Entity Retrieval
Griffin Adams
Department of Computer Science
Columbia University
New York, NY, USA
griffin.adams@columbia.edu
Jason Zucker & Noémie Elhadad
Department of Biomedical Informatics
Columbia University
New York, NY
{jz2700,ne60}@cumc.col... | b5b8f609b28befdd429929816d23ab338f354bfe500ad79f5db2d51b79bd9673 | 8ffd5e22bae8574119adbc6358f394ac4ad06249af8945b2eae95c85be2fd177 | positioned-clean-v2 | 13 | {"removed_words": 180, "retained_words": 9693, "rules": {"dehyphenated": 11, "joined_hyphenated_wrap": 2, "page_number": 20, "repeated_review_header": 160}, "source_words": 9873} |
18iNTRPx8c | 18iNTRPx8c | See What LLMs Cannot Answer: A Self-Challenge Framework for Uncovering LLM Weaknesses | colm | 2,024 | f2b402243eb15627eb9d32ea3643c3fbb2790c5066c897918cb7078e22e3826f | See What LLMs Cannot Answer: A Self-Challenge Framework for Uncovering LLM Weaknesses
Yulong Chen 1,2 ∗ Yang Liu 3 Jianhao Yan 1 Xuefeng Bai 1 Ming Zhong 4
Yinghao Yang 1 Ziyi Yang 3 Chenguang Zhu Yue Zhang 1,5†
1 Westlake University
2 University of Cambridge 3 Microsoft GenAI
4 UIUC 5 Westlake Institute for Advanced S... | 7634c7cbac38aeece27711f5a6cd38c75b58747f3fa491c83e0175aeee27229d | e81276b1cd7ae8a090e0a424a5dcaa83baf776154b4ef5b2da58370590d4bb14 | positioned-clean-v2 | 28 | {"removed_words": 279, "retained_words": 15078, "rules": {"dehyphenated": 31, "joined_hyphenated_wrap": 7, "page_number": 31, "repeated_review_header": 248}, "source_words": 15357} |
1Tny4KgGO2 | 1Tny4KgGO2 | From Strategic Narratives to Code-Like Cognitive Models: An LLM-Based Approach in A Sorting Task | colm | 2,024 | 0abb9f6ba3c6f0c326bcfdb18cebe625f31fd66875d4f929d601fa2ebf6d70a1 | From Strategic Narratives to Code-Like Cognitive Models: An LLM-Based Approach in A Sorting Task
Hanbo Xie 1,2 , Huadong Xiong 1,2 & Robert C. Wilson 1,2 ∗
1 School of Psychology, Georgia Institute of Technology, Atlanta, GA, USA 2 Department of Psychology, University of Arizona, Tucson, AZ, USA
{hanboxie1997,hdx}@gate... | 33d2a0c41c10a089dc5b9895936c3d3946ea324dbcac3aa8c6e8467bee01d139 | f7ecde0b32d8988e3a26998c89b5de5e2a84d6a99ba064a0e0660a89b2f07781 | positioned-clean-v2 | 8 | {"removed_words": 127, "retained_words": 5271, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 13, "page_number": 14, "repeated_review_header": 112}, "source_words": 5398} |
1ba209BACA | 1ba209BACA | Agent-DocEdit: Language-Instructed LLM Agent for Content-Rich Document Editing | colm | 2,024 | fb22ce7021bf6a6abcd3f88f51d8b96ffc0de7798a435807d030da7149d1d701 | Agent-DocEdit: Language-Instructed LLM Agent for Content-
Rich Document Editing
Te-Lin Wu 1 , ∗ Rajiv Jain 2 , Yufan Zhou 2 , Puneet Mathur 2 , Vlad I Morariu 2
1 Character.AI, Menlo Park, CA, USA
telinwu@character.ai
2 Adobe Research, San Jose, CA, USA
{rajijain,yufzhou,puneetm,morariu}@adobe.com
Abstract
Editing cont... | 12ac987b7a9fe3feebc00260841971c16cef64adebde6a4664cc523ed7f1a6ae | 142819be9992f13b8e551c3678677913f4c5c6fde2315168228c8fd58c36398b | positioned-clean-v2 | 9 | {"removed_words": 180, "retained_words": 9596, "rules": {"dehyphenated": 12, "joined_hyphenated_wrap": 2, "page_number": 20, "repeated_review_header": 160}, "source_words": 9776} |
1eg6UnpYu7 | 1eg6UnpYu7 | Prompt Public Large Language Models to Synthesize Data for Private On-device Applications | colm | 2,024 | 5a08a455614fd6906bc9b6eb8d1a6ac24a633c8fbdb6b19a735fbced1c462d61 | Prompt Public Large Language Models to Synthesize Data for Private On-device Applications
Shanshan Wu , ∗ Zheng Xu ∗ , Yanxiang Zhang ∗ , Yuanbo Zhang, Daniel Ramage
Google
{shanshanw, xuzheng, zhangyx, zyb, dramage}@google.com
Abstract
Pre-training on public data is an effective method to improve the performance for f... | df72d6fb4fadb876595b0ceeac5f9496801aa36c37517a9cd9222e1d1ca51cef | 3bfc810926aeaec1bef9321ea3dd6416e4d4084d6388cd2d9c7e9a1daa4961a9 | positioned-clean-v2 | 7 | {"removed_words": 173, "retained_words": 10118, "rules": {"dehyphenated": 6, "expanded_ligature": 2, "joined_hyphenated_wrap": 6, "page_number": 19, "repeated_review_header": 152, "separated_prose_lane": 1, "unmapped_glyph": 2}, "source_words": 10291} |
1pgfvZj0Rx | 1pgfvZj0Rx | Risks from Language Models for Automated Mental Healthcare: Ethics and Structure for Implementation | colm | 2,024 | f9ba5a1701f0031a8ce80fd2715aaadbbf527c2f12971701adc968c04f3f8ff6 | Risks from Language Models for Automated Mental Healthcare: Ethics and Structure for Implementation
Declan Grabb 1,2 ∗ Max Lamparth 2, ∗ Nina Vasan 2
2 Stanford University
1 Northwestern University
Abstract
Amidst the growing interest in developing task-autonomous AI for automated mental health care, this paper address... | d1be49a891c4f228e919b1bface13aa134320e0ac132f6520e7beff739517341 | d88da0c4c4b2719649cc32887bba984f8a99069aa6eacac0447f4a0ef72bd216 | positioned-clean-v2 | 22 | {"removed_words": 351, "retained_words": 19183, "rules": {"dehyphenated": 21, "joined_hyphenated_wrap": 6, "page_number": 39, "repeated_review_header": 312}, "source_words": 19534} |
2cop2jmQVL | 2cop2jmQVL | Stream of Search (SoS): Learning to Search in Language | colm | 2,024 | 2d8931191a7ce2bd15e1637fa64900d642898283324c07b383a034d35d1563de | Stream of Search (SoS): Learning to Search in Language
Kanishk Gandhi ∗
Stanford University
Denise Lee
Stanford University
Winson Cheng
Stanford University
Archit Sharma
Stanford University
Gabriel Grand
MIT
Muxin Liu
Harvey Mudd
Noah D. Goodman
Stanford University
Abstract
Language models are rarely shown fruitful mis... | d50a0a52db3fa7088ccc01fe651cfcf0250c4f569ee774ae5deca01e08885042 | 81251e890b19ecf6cf007e2574586899012986061b3774da75fe9fd62da0979a | positioned-clean-v2 | 5 | {"removed_words": 154, "retained_words": 8865, "rules": {"dehyphenated": 6, "joined_hyphenated_wrap": 1, "page_number": 17, "repeated_review_header": 136, "unmapped_glyph": 1}, "source_words": 9019} |
2nTzomzjjb | 2nTzomzjjb | ProLLM: Protein Chain-of-Thoughts Enhanced LLM for Protein-Protein Interaction Prediction | colm | 2,024 | 5b4733c225808bd0ce942a9f0c715ba0827ad548d9313f2230e1b2a38542fb8a | ProLLM: Protein Chain-of-Thoughts Enhanced LLM for
Protein-Protein Interaction Prediction
Mingyu Jin ∗
Rutgers University
Boming Kang
Peking University
Haochen Xue ∗
University of Liverpool
Ruosong Ye
Rutgers University
Zhenting Wang
Rutgers University
Kaixiong Zhou
Massachusetts Institute of Technology
Yongfeng Zhang ... | 2a515bab33685194285e3d0904206c86761e2c5714759a7c0894eda1da7ba588 | 5eb73fc5d4968c37e4c996d15e773c6eb646123aca404e0c9ddd0a2af8a4b2c6 | positioned-clean-v2 | 15 | {"removed_words": 153, "retained_words": 8162, "rules": {"dehyphenated": 14, "joined_hyphenated_wrap": 5, "page_number": 17, "repeated_review_header": 136}, "source_words": 8315} |
2oHnsM9M9D | 2oHnsM9M9D | ACORN: Aspect-wise Commonsense Reasoning Explanation Evaluation | colm | 2,024 | 5bb55b4440d24159b1ffad3bafb39a3324516d9a271fdd4cdf19db2412f24727 | ACORN : Aspect-wise Commonsense Reasoning
Explanation Evaluation
Ana Brassard, αβ Benjamin Heinzerling, αβ Keito Kudo, βα
Keisuke Sakaguchi βα & Kentaro Inui γβα
α RIKEN, β Tohoku University, γ MBZUAI
{ana.brassard, benjamin.heinzerling}@riken.jp
{keisuke.sakaguchi, kentaro.inui}@tohoku.ac.jp
keito.kudo.q4@dc.tohoku.ac... | 31a4bb3e059599adfadda4218c38b23c23e8ace768cfa72ed13133520ad4a8da | bc583c7f2f9f56610323aed2b691fc30b4bd31d12b03ca81abe010effef4730d | positioned-clean-v2 | 8 | {"removed_words": 162, "retained_words": 8603, "rules": {"dehyphenated": 14, "dropped_invisible_character": 10, "joined_hyphenated_wrap": 4, "joined_small_caps": 8, "page_number": 18, "repeated_review_header": 144}, "source_words": 8765} |
2wtj0up8rv | 2wtj0up8rv | Enhancing Language Models with Idiomatic Reasoning | colm | 2,024 | 774832aadabd57f65a6c7121298faaf7333ac6895d4098ef6033412952de889f | Enhancing Language Models with Idiomatic Reasoning
Jianing Zhou
University of Illinois Urbana-Champaign Urbana, IL 61801, USA
zjn1746@illinois.edu
Ziheng Zeng
University of Illinois Urbana-Champaign Urbana, IL 61801, USA
zzeng13@illinois.edu
Hongyu Gong
Facebook AI aaaaaaaaaaaaaaaaaaaaaaaaaa
Seattle, WA 98109, USA
hygo... | b5b26ecfdf4d20ad1139cbd3d36ce7ea3dcc56121314970e2b5064d3db90402b | 78c6a2983630160d292ef0651355240d29f341dcc1370e627c44c5f57085bb73 | positioned-clean-v2 | 10 | {"removed_words": 174, "retained_words": 9250, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 11, "page_number": 19, "repeated_review_header": 152, "unmapped_glyph": 3}, "source_words": 9424} |
3GhOWfSLrD | 3GhOWfSLrD | Will the Real Linda Please Stand up...to Large Language Models? Examining the Representativeness Heuristic in LLMs | colm | 2,024 | efe3c7211a8aa8ea594b0116b194aa9c52638cf299301be6ce2a0516c98c3008 | Will the Real Linda Please Stand Up...to Large Language Mod-
els? Examining the Representativeness Heuristic in LLMs
Pengda Wang ∗ , Zilin Xiao ∗ , Hanjie Chen, Frederick L. Oswald
Rice University
{pw32,zilin,hc86,fo3}@rice.edu
Abstract
Although large language models (LLMs) have demonstrated remarkable proficiency in m... | cda47f49cac25892fe000f7b518ae6dbbf269925a61cc00dd0b5887da16c90f6 | d1b3e41b28d83237a64f76826c3beaec4a91e2fa99a267b585f01173b8baf519 | positioned-clean-v2 | 15 | {"removed_words": 279, "retained_words": 13066, "rules": {"dehyphenated": 33, "joined_hyphenated_wrap": 2, "joined_small_caps": 66, "page_number": 31, "repeated_review_header": 248}, "source_words": 13345} |
3HTVP34WWE | 3HTVP34WWE | Bot or Human? Detecting ChatGPT Imposters with A Single Question | colm | 2,024 | a78ce0901c26a2bc2ac3d2400d01cf1497df8a0a76ff25a5022e8dedd92e1c61 | Bot or Human? Detecting ChatGPT Imposters with A Single
Question
Hong Wang 1 , Xuan Luo 1 , Weizhi Wang 1 , Melody Yu 2 , Xifeng Yan 1
1 University of California, Santa Barbara, CA, US 2 Sage Hill School, Newport Coast, CA, US
{hongwang600,xuan_luo,weizhiwang}@ucsb.edu
ocmelodyyu@gmail.com, xyan@cs.ucsb.edu
Abstract
La... | 246ddec966005bdfe5e038fde77fffedd8caf9e828a11df4732f7cac9c05a0d5 | fec19a9ecc47acd9d38d188cc997a8d58af01b66c6b81ba921e22aab43706d88 | positioned-clean-v2 | 7 | {"removed_words": 117, "retained_words": 6822, "rules": {"dehyphenated": 8, "joined_hyphenated_wrap": 1, "page_number": 13, "repeated_review_header": 104}, "source_words": 6939} |
3TzGD95Jw1 | 3TzGD95Jw1 | Timo: Towards Better Temporal Reasoning for Language Models | colm | 2,024 | bb66250ba41e68351a09e5d5a4887d7b8b70e56bc49806f53b8bcd88545f7bab | Timo: Towards Better Temporal Reasoning for Language
Models
Zhaochen Su 1∗ , Jun Zhang 1 , Tong Zhu 1 ,
Xiaoye Qu 2 , Juntao Li 1† , Min Zhang 1 , Yu Cheng 3
1
Institute of Computer Science and Technology, Soochow University, China;
2
Shanghai AI Laboratory; 3 The Chinese University of Hong Kong
{suzhaochen0110,junzhan... | ed78bf38f9c00658a624748c93ba1854f0f9f433cf41db6ce4d711601608cb87 | 2834a98cfc7fd8d23e567daae26f027766483161d6a094c3e505343ef96e0201 | positioned-clean-v2 | 18 | {"removed_words": 173, "retained_words": 9909, "rules": {"dehyphenated": 27, "joined_hyphenated_wrap": 4, "page_number": 21, "repeated_review_header": 152, "separated_prose_lane": 1}, "source_words": 10082} |
3X2L2TFr0f | 3X2L2TFr0f | MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies | colm | 2,024 | 051f26973f45ccfeb0049897ba3fc3c89467548bb83634e3f5f79ed1418e1a31 | MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies
Shengding Hu 1 , Yuge Tu 2 , Xu Han 1,3 ∗ , Chaoqun He 1 , Ganqu Cui 1 , Xiang Long 2 ,
Zhi Zheng 2 , Yewei Fang 2 , Yuxiang Huang 1 , Weilin Zhao 1 , Xinrong Zhang 1 ,
Zheng Leng Thai 1 ,Kaihuo Zhang 2 , Chongyi Wang 2 , Yuan Y... | 31cd16ec1dc02cb1d5eb181af7c7d8dbef2f8d3de950c7ead349ce5871fa7942 | 605a14604af00d5c9ebd93b1e3afba4568db8924ba47bdedac8ac385f074f654 | positioned-clean-v2 | 17 | {"removed_words": 300, "retained_words": 16390, "rules": {"dehyphenated": 16, "joined_hyphenated_wrap": 3, "mapped_symbol_piece": 6, "page_number": 33, "repeated_review_header": 264, "unmapped_glyph": 3}, "source_words": 16690} |
3nTbuygoop | 3nTbuygoop | StateFlow: Enhancing LLM Task-Solving through State-Driven Workflows | colm | 2,024 | 6d16f4fc03bc9f646dc33f41564e2cc7e6727b238a6d99b42fcd1854225e8b87 | StateFlow: Enhancing LLM Task-Solving through State-
Driven Workflows
Yiran Wu
Pennsylvania State University
yiran.wu@psu.deu
Tianwei Yue
MathGPTPro
tianwei.yue@mathgptpro.com
Shaokun Zhang
Pennsylvania State University
shaokun.zhang@psu.edu
Chi Wang
Microsoft Research Redmond
wang.chi@microsoft.com
Qingyun Wu
Pennsylv... | 04df220109ed8f40bb2552ae82b0fbecc00dc7ed4df5a96aee789ee44107f9d1 | aeb810fdd2cd1b12578d2352abfc7fc6de8aab1133ff1e288536d33f0daad2aa | positioned-clean-v2 | 12 | {"removed_words": 279, "retained_words": 15954, "rules": {"composed_negation": 1, "dehyphenated": 8, "dropped_invisible_character": 22, "joined_hyphenated_wrap": 2, "joined_split_letters": 24, "page_number": 31, "repeated_review_header": 248, "separated_prose_lane": 2}, "source_words": 16233} |
3ypWPhMGhV | 3ypWPhMGhV | Cohesive Conversations: Enhancing Authenticity in Multi-Agent Simulated Dialogues | colm | 2,024 | 09b8b300530104ef4e0e30df57e7db968b3a4e8c1c4cfb7e9eb8e01a40982079 | Cohesive Conversations: Enhancing Authenticity in Multi-
Agent Simulated Dialogues
KuanChao Chu, Yi-Pei Chen & Hideki Nakayama ∗
The University of Tokyo
{kcchu,ypc}@nlab.ci.i.u-tokyo.ac.jp
Abstract
This paper investigates the quality of multi-agent dialogues in simulations powered by Large Language Models (LLMs). Analy... | 3a5f43d8372cf9f120917393ccd1b7cd6af632298596d99937898924fe873785 | 646270db3014f403086166a19779a7c4ee6a31f4cf98a2948e91f8b0bbf8f3d6 | positioned-clean-v2 | 20 | {"removed_words": 209, "retained_words": 10252, "rules": {"combined_negation": 2, "composed_negation": 2, "dehyphenated": 26, "joined_hyphenated_wrap": 5, "joined_small_caps": 48, "mapped_symbol_piece": 3, "page_number": 23, "repeated_review_header": 184, "separated_prose_lane": 1}, "source_words": 10461} |
46Zgqo4QIU | 46Zgqo4QIU | Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation | colm | 2,024 | 9b8133d18c5d6a6089b79507459ec441fbc88ca32fc78c0c7a11cef6df2ed292 | Self-Taught Optimizer (STOP): Recursively Self-Improving
Code Generation
Eric Zelikman
Stanford University*
eric@zelikman.me
Eliana Lorch
elianalorch@gmail.com
Lester Mackey
Microsoft Research
Adam Kalai
OpenAI*
lmackey@microsoft.com adam@kal.ai
Abstract
Several recent advances in AI systems solve problems by providing... | f3efc34492914d43a8d737fffd9dfa7350e92e5e0f88429db343f0e863bf3128 | 5f8d64ffe0ad97cf29bcc0f2cee84bd0f900d249486077e05941e5a35a70f857 | positioned-clean-v2 | 17 | {"removed_words": 423, "retained_words": 23024, "rules": {"dehyphenated": 13, "joined_hyphenated_wrap": 2, "mapped_symbol_piece": 13, "page_number": 46, "repeated_review_header": 368, "separated_prose_lane": 1, "unmapped_glyph": 9}, "source_words": 23447} |
4HNAwZFDcH | 4HNAwZFDcH | WorkBench: a Benchmark Dataset for Agents in a Realistic Workplace Setting | colm | 2,024 | 458ac67cd6934026830ded1d079cb3ece2199aaefaff5c252422d0389282bdc6 | WorkBench: a Benchmark Dataset for Agents in a Realistic
Workplace Setting
Olly Styles , ∗ Sam Miller , ∗ Patricio Cerda-Mardini
Mindsdb
ollystyles@gmail.com, sam@artanis.ai, patricio@mindsdb.com
Tanaya Guha
University of Glasgow
tanaya.guha@glasgow.ac.uk
Victor Sanchez
University of Warwick
V.F.Sanchez-Silva@warwick.a... | 3f6b5c20fd3068dc1f12ab84e1a0c1de3a111879fe63f9175b012d72d7f19682 | aaaf89565ced8215c28f31522efd080f02590dae4583985fc0fc3ded16b1996d | positioned-clean-v2 | 11 | {"removed_words": 351, "retained_words": 10764, "rules": {"dehyphenated": 8, "page_number": 39, "repeated_review_header": 312}, "source_words": 11115} |
4aqq9xTtih | 4aqq9xTtih | Prompt-prompted Adaptive Structured Pruning for Efficient LLM Generation | colm | 2,024 | 1de3c4694b9bd4dc05792db295560fbc4cb88504507b6a10ebaa86a9da9f985b | Prompt-prompted Adaptive Structured Pruning for
Efficient LLM Generation
Harry Dong, Beidi Chen, Yuejie Chi
Department of Electrical and Computer Engineering
Carnegie Mellon University
Pittsburgh, PA 15213, USA
{harryd,beidic,yuejiec}@andrew.cmu.edu
Abstract
With the development of transformer-based large language mode... | ac4860008c0833df44fe71c7c9235c8105dab0e8ccd06357538d0e795bce1c5a | fc0bdbd34a3793aa62c891f0daa64d2da0487ff5d14cb68f53f669281ab5173e | positioned-clean-v2 | 2 | {"removed_words": 173, "retained_words": 7172, "rules": {"joined_small_caps": 81, "page_number": 19, "repeated_review_header": 152, "unmapped_glyph": 2}, "source_words": 7345} |
5B2K4LRgmz | 5B2K4LRgmz | Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data | colm | 2,024 | 62bbe68c69d14da7b24f4327eaba04720d9facc686512d9d63eaa5b9541d0af4 | Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data
Matthias Gerstgrasser ∗
Harvard & Stanford University
matthias@seas.harvard.edu
Rafael Rafailov ∗
Stanford University
rafailov@stanford.edu
Rajashree Agrawal †
Constellation
Rylan Schaeffer ∗
Stanford University
rschae... | 8fe8719adef07b5e17b19fa4f4367a0531a586112344fa6b1df9c7978e3c400a | ce6fa64a520ac3cd516374737b86922dae2ed3fabcc201f7d26f946c0b02772d | positioned-clean-v2 | 4 | {"removed_words": 264, "retained_words": 11441, "rules": {"dehyphenated": 8, "joined_hyphenated_wrap": 1, "mapped_symbol_piece": 4, "page_number": 29, "repeated_review_header": 232, "unmapped_glyph": 3}, "source_words": 11705} |
5Evv4tIjUI | 5Evv4tIjUI | Exploiting the Potential of Seq2Seq Models as Robust Few-Shot Learners | colm | 2,024 | 0295062fa14535a2cdae01573ebc0294fb027f52599e98fb26edbde5163106a4 | Exploiting the Potential of Seq2Seq Models as Robust Few-
Shot Learners
Jihyeon Lee *†2 , Dain Kim *†3 , Doohae Jung *1 , Boseop Kim 1 , Kyoung-Woon On 1
Kakao Corp., NVIDIA, Kyunghee University
jihyeonl@nvidia.com, dannykm@khu.ac.kr, {wavy.jung,mat.mul,kloud.off}@kakaocorp.com
Abstract
In-context learning, which offer... | d77b9455f092114cadeb2da2ce1ac238fdbeafd3fb9a5535ff1086f50442955d | bf494032a32b7f730d0dd335bcb953a3178133bc6c43c7ecc1c40a8b5e7fe09a | positioned-clean-v2 | 18 | {"removed_words": 180, "retained_words": 11167, "rules": {"dehyphenated": 13, "joined_hyphenated_wrap": 7, "page_number": 20, "repeated_review_header": 160}, "source_words": 11347} |
5Nsl0nlStc | 5Nsl0nlStc | Pack of LLMs: Model Fusion at Test-Time via Perplexity Optimization | colm | 2,024 | 6bfb7394e109a6cdc7f3beb389d0d4416c26e9da4e91a3736e1c059283406996 | Pack of LLMs: Model Fusion at Test-Time via Perplexity
Optimization
Costas Mavromatis
University of Minnesota
mavro016@umn.edu
Petros Karypis
University of California San Diego
pkarypis@ucsd.edu
George Karypis
University of Minnesota
karypis@umn.edu
Abstract
Fusing knowledge from multiple Large Language Models (LLMs) c... | 0d0d1abb13794ed03c62b1755df4f8c7e0fa2e74d0509769a799a3b301c9fa2d | ccc94fcf993da8c0c28c548d4cb534d2427b19a9b2a53e9ac4eb5c9484a35251 | positioned-clean-v2 | 6 | {"removed_words": 203, "retained_words": 9516, "rules": {"dehyphenated": 13, "joined_hyphenated_wrap": 1, "page_number": 21, "repeated_review_header": 168, "separated_prose_lane": 2, "unmapped_glyph": 14}, "source_words": 9719} |
5RdIMlGLXL | 5RdIMlGLXL | LLM-Datasets: An Open Framework for Pretraining Datasets of Large Language Models | colm | 2,024 | b8ffcd52ccc273894a82abf39355ba54f0ba79c97f086cf5630c32f45efaf77d | LLM-Datasets: An Open Framework for Pretraining Datasets of Large Language Models
Malte Ostendorff 1,2 , Pedro Ortiz Suarez 3 , Lucas Fonseca Lage 4 , Georg Rehm 4,1
1 Occiglot, 2 Deutsche Telekom, 3 Common Crawl Foundation,
4 Deutsches Forschungszentrum für Künstliche Intelligenz GmbH (DFKI)
{firstname}@occiglot.org
A... | ee4453a0611daebbe8d11a12ff66c51ef951f2df3f07a61faf4822b70a05dcb7 | ec9b34292f5da3d0a5d70908e8a94d172e128d13ff4184e94c77cdec1ba120fe | positioned-clean-v2 | 33 | {"removed_words": 225, "retained_words": 11524, "rules": {"dehyphenated": 26, "joined_hyphenated_wrap": 5, "joined_split_letters": 34, "page_number": 25, "repeated_review_header": 200}, "source_words": 11749} |
5fg0VtRxgi | 5fg0VtRxgi | SteP: Stacked LLM Policies for Web Actions | colm | 2,024 | f088b5b09e13681d7d94972a2d48b4994853fbac070707f020cfb63f2d71f7ea | SteP: Stacked LLM Policies for Web Actions
Paloma Sodhi 1 ∗ , S.R.K. Branavan 1 , Yoav Artzi 1,2 , Ryan McDonald 1
1 ASAPP Research, NY, USA
2 Cornell University, NY, USA
{psodhi, branavan, rmcdonald}@asapp.com, yoav@cs.cornell.edu
Abstract
Performing tasks on the web presents fundamental challenges to large language m... | ce3dae2b52041d0c3e1d94715b5e055ddab7fbf0f39fd499676d891103a6c3cf | 09638f3103a828425358d5b9b5b59946ea731e496e79527e685fec720f3cdca4 | positioned-clean-v2 | 17 | {"removed_words": 296, "retained_words": 15532, "rules": {"dehyphenated": 10, "page_number": 32, "repeated_review_header": 264}, "source_words": 15828} |
5u1GpUkKtG | 5u1GpUkKtG | Helping or Herding? Reward Model Ensembles Mitigate but do not Eliminate Reward Hacking | colm | 2,024 | e9d55fcc47cfa2f80abbbd0dff48f4ec08b051ab29bf92a0f83d8367345c8a2c | Helping or Herding?
Reward Model Ensembles Mitigate but do not Eliminate Re-
ward Hacking
Jacob Eisenstein 1, ∗
Ahmad Beirami 1
Katherine Heller 2
Chirag Nagpal 2, ∗
Alex D’Amour 1
Stephen Pfohl 2
Alekh Agarwal 2, ∗
DJ Dvijotham 1
Adam Fisch 1
Deepak Ramachandran 1
Peter Shaw 1
Jonathan Berant 1, ∗
1. Google DeepMind
2... | f14e0dff26553bd34553b48e2936e81933cbc944b0d47cfac55f6b3321690bdc | 22f0d5d12270aaad48123cdc9a91f748b6ea1c525b3de6a13d3fea584ea6dcb7 | positioned-clean-v2 | 5 | {"removed_words": 220, "retained_words": 10112, "rules": {"dehyphenated": 7, "expanded_ligature": 3, "joined_small_caps": 6, "page_number": 24, "repeated_review_header": 192, "unmapped_glyph": 4}, "source_words": 10332} |
60a1SAtH4e | 60a1SAtH4e | Measuring and Controlling Instruction (In)Stability in Language Model Dialogs | colm | 2,024 | 235aa4a9b8801b654805393d7e1e82ad8a2d4f6fc7dfe831587bfa3cde54cddc | Measuring and Controlling Instruction (In)Stability in Language Model Dialogs
Kenneth Li 1 , ∗ Tianle Liu 1 , Naomi Bashkansky 1 ,
David Bau 2 , Fernanda Viégas 1 , Hanspeter Pfister 1 , Martin Wattenberg 1
1 Harvard University, 2 Northeastern University
Abstract
System-prompting is a standard tool for customizing lang... | 49dc6f4c11d5e4e23bdc0235c92632c11491c494c74026e1e7ff718fb2c99c11 | ac32f84a7d10bdc2aa1ffb113555ae11821dcf4713bf5a44c1722aa020aec3d7 | positioned-clean-v2 | 16 | {"removed_words": 192, "retained_words": 10572, "rules": {"dehyphenated": 11, "dropped_invisible_character": 3, "joined_hyphenated_wrap": 2, "joined_small_caps": 1, "mapped_symbol_piece": 3, "page_number": 19, "repeated_review_header": 152, "separated_prose_lane": 1, "unmapped_glyph": 21}, "source_words": 10764} |
6U1FEKP7Ar | 6U1FEKP7Ar | ExoViP: Step-by-step Verification and Exploration with Exoskeleton Modules for Compositional Visual Reasoning | colm | 2,024 | 6ff45a22f8ac8e38fbcbf5037173801156468f2896a4f946babd08d96296c0e7 | ExoViP: Step-by-step Verification and Exploration with Exoskeleton Modules for Compositional Visual Reasoning
Yuxuan Wang 1
Alan Yuille 2
Zhuowan Li 2
Zilong Zheng 1
1 State Key Laboratory of General Artificial Intelligence, BIGAI, Beijing, China 2 Johns Hopkins University, Baltimore, Maryland, USA
wangyuxuan1@bigai.ai... | af7c1dfac5e2e7be811f4b1544c2f44addc3aa2b5bd6aa9f994a5f7bd7add325 | c21c2ee47e6d4653c995f1034c633a4eddb54c1719db75b9f2818a81b59b5dd3 | positioned-clean-v2 | 97 | {"removed_words": 302, "retained_words": 10700, "rules": {"dehyphenated": 59, "dropped_invisible_character": 8, "dropped_unmapped_glyph": 39, "joined_hyphenated_wrap": 5, "joined_small_caps": 130, "page_number": 25, "repeated_review_header": 200, "separated_prose_lane": 2, "unmapped_glyph": 77}, "source_words": 11002} |
6vEfyp0o68 | 6vEfyp0o68 | MANGO: A Benchmark for Evaluating Mapping and Navigation Abilities of Large Language Models | colm | 2,024 | c5866b3c8fc70c46e18310288f3eafdaa9df01e6192ae105507f6505d3824dfb | MANGO: A Benchmark for Evaluating Mapping and Navigation Abilities of Large Language Models
Peng Ding 1 Jiading Fang 2
Peng Li 2
Kangrui Wang 2
Xiaochen Zhou 3,5
4
5
2
2
Mo Yu
Jing Li
Matthew R. Walter
Hongyuan Mei
1 University of Chicago 2 Toyota Technological Institute at Chicago 3 Syracuse University 4 WeChat AI 5 N... | c7401e5e3ac67979abe94d53ec8e6eef66951fb8ca78e5dbaaf001419c2faaa4 | 2629c89dd91a8df4e330053e52f3ac585a52b5e98b8af1049777f1d9716da3c4 | positioned-clean-v2 | 11 | {"removed_words": 270, "retained_words": 17710, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 4, "joined_small_caps": 192, "page_number": 30, "repeated_review_header": 240}, "source_words": 17980} |
7BCmIWVT0V | 7BCmIWVT0V | Corex: Pushing the Boundaries of Complex Reasoning through Multi-Model Collaboration | colm | 2,024 | 0a45f6dbc9d1eb02123f18a93ddc0f9d0cf8977c800d2b720bd32692038eeb27 | Corex: Pushing the Boundaries of Complex Reasoning
through Multi-Model Collaboration
♢♡∗
♣
♠
♢†
♣
Qiushi Sun
Zhangyue Yin Xiang Li Zhiyong Wu Xipeng Qiu Lingpeng Kong
♡
Shanghai AI Laboratory The University of Hong Kong
♣
♠
Fudan University East China Normal University
qiushisun@connect.hku.hk, yinzy21@m.fudan.edu.cn, ... | 2ce818600f241913d6268b8d6acb83633b227fe6cae35bf414dcf36d79407ddd | b1213199758ffb9b1a394d0cfc3a7e98d4481908a77d982ea6568a910d0aed9a | positioned-clean-v2 | 37 | {"removed_words": 296, "retained_words": 15924, "rules": {"dehyphenated": 27, "joined_hyphenated_wrap": 3, "joined_small_caps": 93, "page_number": 32, "repeated_review_header": 264, "separated_prose_lane": 3}, "source_words": 16220} |
7QaEO9WYMa | 7QaEO9WYMa | Poly-Visual-Expert Vision-Language Models | colm | 2,024 | 713818f8abcf9e7f3cb4a504582fd1275ae9783bd93956680ac6ffa3dbfff2f2 | Poly-Visual-Expert Vision-Language Models
Xiaoran Fan ∗ , Tao Ji ∗ , Changhao Jiang ∗ , Shuo Li ∗ , Senjie Jin ∗ , Sirui Song
Junke Wang, Boyang Hong, Lu Chen, Guodong Zheng, Ming Zhang, Caishuang Huang Rui Zheng, Zhiheng Xi, Yuhao Zhou, Shihan Dou, Junjie Ye, Hang Yan Tao Gui † , Qi Zhang , † Xipeng Qiu, Xuanjing Huan... | 35e2de54acc9f97afe1528bb7116a69655ce8e322814c3cbb49bd4c5109abcd0 | 9decfa2d65d3dc46a2462901cd616d663d0217daee8559cfd750007aead0c1a1 | positioned-clean-v2 | 19 | {"removed_words": 169, "retained_words": 8280, "rules": {"dehyphenated": 6, "joined_hyphenated_wrap": 4, "page_number": 18, "repeated_review_header": 144, "unmapped_glyph": 7}, "source_words": 8449} |
7VPKtz8CHN | 7VPKtz8CHN | Beyond Relevance: Evaluate and Improve Retrievers on Perspective Awareness | colm | 2,024 | 6338e03c0f674c459144bd0057648ca0e0884b9ac6bf38a85962cd032e9477bb | Beyond Relevance: Evaluate and Improve Retrievers on
Perspective Awareness
Xinran Zhao 1 , Tong Chen 2 , Sihao Chen 3 , Hongming Zhang 4 , Tongshuang Wu 1
1 Carnegie Mellon University, 2 University of Washington, 3 University of Pennsylvania, 4 Tencent AI Lab, Bellevue
{xinranz3, sherryw}@andrew.cmu.edu
Abstract
The ta... | ef20643065d89b15e5b9c08570dcca0121404473a09c96d4443ce93a7c91a106 | 0f8d772e1e2ea4e2784f25181dcabf169b62e88971abbb5df3a75f3ff00bc98b | positioned-clean-v2 | 11 | {"removed_words": 189, "retained_words": 11855, "rules": {"dehyphenated": 32, "joined_hyphenated_wrap": 5, "page_number": 21, "repeated_review_header": 168}, "source_words": 12044} |
7iaAlIlV2H | 7iaAlIlV2H | Pairwise Proximal Policy Optimization: Language Model Alignment with Comparative RL | colm | 2,024 | d32f4137db2ae35e649165ed0dda105c03d4287eb43711caa70c1e5056e5852c | Pairwise Proximal Policy Optimization: Language Model Alignment with Comparative RL
Tianhao Wu 1 ∗ Banghua Zhu 1 Ruoyu Zhang 1
Kannan Ramchandran 1 Jiantao Jiao 1
1 University of California, Berkeley
Zhaojin Wen 1
Abstract
LLMs may exhibit harmful behavior without aligning with human values. The dominant approach for s... | 7d3a61c8299647a2c1e2dede9fc28f008310f40a8f2c5b57dbdb0e00317a3927 | 963964d37050a355de1d27fd9730e329827920eb0201bda3f528068d282abba2 | positioned-clean-v2 | 10 | {"removed_words": 248, "retained_words": 11396, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 4, "page_number": 21, "repeated_review_header": 168, "unmapped_glyph": 59}, "source_words": 11644} |
7jSMMvXLri | 7jSMMvXLri | Measuring Taiwanese Mandarin Language Understanding | colm | 2,024 | ca52a8b05a9dafa220b1b1ed11e322b152d9ab7bf126882bca4314c0dd82021a | Measuring Taiwanese Mandarin Language Understanding
Po-Heng Chen, Sijia Cheng, Wei-Lin Chen, Yen-Ting Lin, Yun-Nung Chen National Taiwan University, Taipei, Taiwan
{ytl, yvchen}@ieee.org
Abstract
The evaluation of large language models (LLMs) has drawn substantial attention in the field recently. This work focuses on e... | 1471661823a2a0627ac64b011029ebb1c6de8baf91c7dc82c61860260582eb71 | d8239f40c9b437d2db86c5a55d2136759c4b0bc1ed4644b3c370efa281d94b48 | positioned-clean-v2 | 18 | {"removed_words": 162, "retained_words": 8277, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 6, "joined_small_caps": 22, "page_number": 18, "repeated_review_header": 144}, "source_words": 8439} |
7xUtka9ck9 | 7xUtka9ck9 | Yes, no, maybe? Revisiting language models' response stability under paraphrasing for the assessment of political leaning | colm | 2,024 | 31fa6b2a52fe9b5d73d4ecdf8c0672fa40d105d6af1932a6fb3c03c95244e250 | Yes, no, maybe? Revisiting language models’ response stability under paraphrasing for the assessment of political leaning
Patrick Haller, Jannis Vamvas, Lena A. Jäger Department of Computational Linguistics
University of Zurich
{haller, vamvas, jaeger}@cl.uzh.ch
Abstract
An increasing number of studies are aimed at unc... | 8a274effb483d3406f8571c066e373d102ab8cec1380e8872d615969b1218cb8 | 26a59c0621f2abf27b2f19b0f52de65332f4ab949969932c92c5b1a7fe50ccf9 | positioned-clean-v2 | 17 | {"removed_words": 185, "retained_words": 10861, "rules": {"combined_negation": 14, "composed_negation": 14, "dehyphenated": 8, "joined_hyphenated_wrap": 3, "joined_small_caps": 4, "page_number": 19, "repeated_review_header": 152, "separated_prose_lane": 1}, "source_words": 11046} |
7ysaJGs7zY | 7ysaJGs7zY | IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models | colm | 2,024 | 916eff655e940ace9d7df44f902a425a48b6609d8b9041ef578a377c2ec0d127 | IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models
Haz Sameen Shahgir 1,3 , ∗ Khondker Salman Sayeed 1 , ∗ Abhik Bhattacharjee 1 ,
Wasi Uddin Ahmad 2 , Yue Dong 3 , Rifat Shahriyar 1
Bangladesh University of Engineering and Technology 1 , AWS AI Labs 2† , University of California Riverside 3... | 34a30f2bd9a4639590363742aeb00daae20b766eac802a8a20ca7f3c7a243e38 | d4114c743f7595c128011f58103e13628d176ebe7ea55fb6d05ebfa857162c3c | positioned-clean-v2 | 9 | {"removed_words": 189, "retained_words": 8472, "rules": {"dehyphenated": 9, "joined_hyphenated_wrap": 4, "page_number": 21, "repeated_review_header": 168}, "source_words": 8661} |
8TdcXwfNRB | 8TdcXwfNRB | PAPERCLIP: Associating Astronomical Observations and Natural Language with Multi-Modal Models | colm | 2,024 | d55d271b530c17a3ea7919b69bd72eebca858781652df0348a1a091baa36370a | PAPERCLIP: Associating Astronomical Observations and Natural Language with Multi-Modal Models
Siddharth Mishra-Sharma, Yiding Song, & Jesse Thaler
MIT & IAIFI
{smsharma,ydsong,jthaler}@mit.edu
Abstract
We present PAPERCLIP (Proposal Abstracts Provide an Effective Repre-
sentation for Contrastive Language-Image Pre-trai... | ae712f49f517efa6846392de002c33d3da03400d655a7b681bebce640f9339de | 77ae96100496adf45fee2e4cce459d68428700011fadb63f88d02cbbb3b2319b | positioned-clean-v2 | 14 | {"removed_words": 176, "retained_words": 9349, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 11, "joined_hyphenated_wrap": 7, "joined_small_caps": 6, "page_number": 19, "repeated_review_header": 152, "unmapped_glyph": 4}, "source_words": 9525} |
8tKjqqMM5z | 8tKjqqMM5z | Keep the Cost Down: A Review on Methods to Optimize LLM’s KV-Cache Consumption | colm | 2,024 | c87f9a494a864b17b55f75d1afad45b3cf37bfc01e9bd35dafd27dfdc393cedb | Keep the Cost Down: A Review on Methods to Optimize LLM’s KV Cache Consumption
Shi Luohe & Zhang Hongyi
National Engineering Research Center
for Multimedia Software,
School of Computer Science,
Wuhan University,
Wuhan, 430072, P. R. China
{shiluohe, harryzhang}@whu.edu.cn
Yao Yao
Department of Computer Science
and Engi... | c6fdc71508e4fad271d81e10f8a504cc6858fdefb31af4e7b7092a516dd644a3 | e93693fca83c8b055c862da3a963ebc4642ecab61b34dad44f87626430001a52 | positioned-clean-v2 | 11 | {"removed_words": 171, "retained_words": 9996, "rules": {"dehyphenated": 9, "joined_hyphenated_wrap": 1, "page_number": 19, "repeated_review_header": 152}, "source_words": 10167} |
8w0RApM5yG | 8w0RApM5yG | BumbleBee: Dynamic KV-Cache Streaming Submodular Summarization for Infinite-Context Transformers | colm | 2,024 | a88da619c2aeafb9a2994b1c3e6a1b3daf46aac239407be2af897d4cce8a8b3c | BumbleBee: Dynamic KV-Cache Streaming Submodular Summarization for Infinite-Context Transformers
Lilly Kumari ⋄ ‡
Shengjie Wang ⋄ †
Tianyi Zhou ⋄ ⋆
Nikhil Sarda ‡
§
⋄
Anthony Rowe
Jeff Bilmes
⋄ University of Washington, Seattle, † NYU Shanghai, ⋆ University of Maryland, ‡ Google, § Carnegie Mellon University
{lkumari, ... | b73b2e817d1d9f442ee0847aa95848ce7db397be59e5e7e3d1a1000f055b3677 | 00407ecc0631fffd17df10349dd11cb0f3ef16dcfa27a77f508b77390e6d2884 | positioned-clean-v2 | 21 | {"removed_words": 199, "retained_words": 11675, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 13, "joined_hyphenated_wrap": 1, "joined_split_letters": 8, "page_number": 22, "repeated_review_header": 176}, "source_words": 11874} |
95TayIeqJ4 | 95TayIeqJ4 | TMMLU+: An Improved Traditional Chinese Evaluation Suite for Foundation Models | colm | 2,024 | 56b236d0dc8f195cdbbd8dd3b2fcae5c6e48ce7df34bcfb05563005fc4261936 | TMMLU+: An Improved Traditional Chinese Evaluation Suite
for Foundation Models
Zhi Rui Tam ∗ † , Ya-Ting Pai ∗ † , Yen-Wei Lee, Jun-Da Chen, Wei-Min Chu, Sega Cheng
iKala AI Lab
{avery.pai, patrick.lee}@ikala.ai
Hong-Han Shuai
National Yang Ming Chiao Tung University
Abstract
We present TMMLU+, a new benchmark designed... | 629af936a8381312677742d14181ab6279ac0b6ccee13c10f08485a32c4b9517 | 6aa8f66f98faa96dd71c360ab1faa7113c11e38a02da3bb7a79c48f36ee543b9 | positioned-clean-v2 | 13 | {"removed_words": 531, "retained_words": 17166, "rules": {"dehyphenated": 25, "joined_hyphenated_wrap": 5, "page_number": 59, "repeated_review_header": 472}, "source_words": 17697} |
98ekcwQqb7 | 98ekcwQqb7 | Latent Causal Probing: A Formal Perspective on Probing with Causal Models of Data | colm | 2,024 | 2ad892b091ca4ba6a94ccdc9e446be4774286465e44f286efefeb9458c078554 | Latent Causal Probing:
A Formal Perspective on Probing with Causal Models of Data
Charles Jin ∗
MIT CSAIL
ccj@csail.mit.edu
Abstract
As language models (LMs) deliver increasing performance on a range of NLP tasks, probing classifiers have become an indispensable technique in the effort to better understand their inner ... | 47fda1d8c303e558d000ba321b06879d5e0b0312400a9b7a563fb3548f650276 | 7ca1aa0b219e1009979c553e35993e21ac6576b45cbeb37c817d165d4f4ad182 | positioned-clean-v2 | 13 | {"removed_words": 176, "retained_words": 8988, "rules": {"combined_negation": 2, "composed_negation": 2, "dehyphenated": 6, "page_number": 19, "repeated_review_header": 152, "unmapped_glyph": 3}, "source_words": 9164} |
9Ik05cycLq | 9Ik05cycLq | Certifying LLM Safety against Adversarial Prompting | colm | 2,024 | 5a2b222e82b62ded82969b7096bcf8e94d354cf5fa9d12f990d9aa2d263d7c43 | Certifying LLM Safety against Adversarial Prompting
Aounon Kumar
Harvard University
Cambridge, MA
aokumar@hbs.edu
Chirag Agarwal
Harvard University
Cambridge, MA
Suraj Srinivas
Harvard University
Cambridge, MA
Aaron Jiaxun Li
Harvard University
Cambridge, MA
Soheil Feizi
University of Maryland
College Park, MD
Himabind... | 630a62f2e4270294e2328c60588f9cc4e3bc4b113cb792186c4964dcdd5e60c3 | b1f6c20cf780678612ed0b1c944cd83bc6eef1cfcd49a2ce00f22e4fe67c06b8 | positioned-clean-v2 | 14 | {"removed_words": 255, "retained_words": 14132, "rules": {"dehyphenated": 42, "joined_hyphenated_wrap": 2, "page_number": 27, "repeated_review_header": 216, "separated_prose_lane": 2, "unmapped_glyph": 12}, "source_words": 14387} |
9JY1QLVFPZ | 9JY1QLVFPZ | Forcing Diffuse Distributions out of Language Models | colm | 2,024 | e7c706ef504fd72f0636fc4aee53197e5c060ed86557a20280bbe3ea00a28931 | Forcing Diffuse Distributions out of Language Models
Yiming Zhang 1 ∗ Avi Schwarzschild 1
Nicholas Carlini 2 Zico Kolter 1,3 Daphne Ippolito 1,2 1 Carnegie Mellon University
2 Google DeepMind
3 Bosch Center for AI
Abstract
Despite being trained specifically to follow user instructions, today’s instruction-tuned languag... | 8396c6ade35406f471ef96674a9e95fc544358b4c87aba8ac7fa574d66c82b5e | eed6cc6645df2db5ce9a4aac2c218868b080daca1029c9564e0e7ac70e2f64d3 | positioned-clean-v2 | 30 | {"removed_words": 186, "retained_words": 8190, "rules": {"dehyphenated": 33, "joined_hyphenated_wrap": 2, "joined_small_caps": 62, "page_number": 20, "repeated_review_header": 160, "unmapped_glyph": 6}, "source_words": 8376} |
9Wmdk94oKF | 9Wmdk94oKF | CHOPS: CHat with custOmer Profile Systems for Customer Service with LLMs | colm | 2,024 | 7a6c2a37477fe8d9728b9e537f3440285655ea1d36681fcd2e309542ae683100 | CHOPS: CHat with custOmer Profile Systems for Customer
Service with LLMs
Jingzhe Shi 1,4 , Jialuo Li 1,4 , Qinwei Ma 1,4 , Zaiwen Yang 1,4 , Huan Ma 1,4 , Lei Li
2,3,4
1 Tsinghua University, 2 University of Copenhagen, 3 University of Washington, 4 CPHOS ∗
1 {shi-jz21, lijialuo21, mqw21, yangzw23, mah21}@mails.tsinghua... | 2865184be7ce77f2d6594254e1ea6e3f5cef50aa201bbecd32e2ad68b1a8a174 | 59b40d24396ecbd4def7da401b72be49e9694947374828cf6436cb2599bbe728 | positioned-clean-v2 | 12 | {"removed_words": 144, "retained_words": 7269, "rules": {"dehyphenated": 6, "page_number": 16, "repeated_review_header": 128}, "source_words": 7413} |
9gdZI7c6yr | 9gdZI7c6yr | Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators | colm | 2,024 | e7e6bba9fc6219f844413f02841a3fc2c8a56d552440b53d001c2af8f9002728 | Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators
Yinhong Liu 1 ∗ Han Zhou 1 ∗ Zhijiang Guo 1 Ehsan Shareghi 2
Anna Korhonen 1 Nigel Collier 1
1 University of Cambridge 2 Monash University
{yl535,hz416,zg283,iv250,alk23,nhc30}@cam.ac.uk
Ivan Vulić 1
ehsan.shareghi@monash.... | d1d219c7edae532b17e9717180a0d816a5f64ab867399841d54b893121c8e002 | 8e607a86bb5b2a25106f1cb4bdb6c3e53b8e975f93278bd62c0159dd1c4ed082 | positioned-clean-v2 | 20 | {"removed_words": 182, "retained_words": 11069, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 21, "joined_hyphenated_wrap": 2, "joined_small_caps": 86, "page_number": 20, "repeated_review_header": 160, "separated_prose_lane": 1, "unmapped_glyph": 1}, "source_words": 11251} |
ADtL6fgNRv | ADtL6fgNRv | Inspecting and Editing Knowledge Representations in Language Models | colm | 2,024 | 7b7ce5a01895adfeb4d3eb7b074739f58599206c389ef32ddfa7d6b6a7174c56 | Inspecting and Editing Knowledge Representations in
Language Models
Evan Hernandez, Belinda Z. Li & Jacob Andreas Computer Science & Artificial Intelligence Laboratory Massachusetts Institute of Technology
{dez,bzl,jda}@mit.edu
Abstract
Neural language models (LMs) represent facts about the world described by text. Som... | 2523545ab87aedbb95b36f3a42741e6eea27209759540c622482ee13a5224540 | bfb71aef36dd7719aceaeedccd4f575d587a99c3e857e3abb4f647aa0dd20c9f | positioned-clean-v2 | 15 | {"removed_words": 201, "retained_words": 13386, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 17, "joined_hyphenated_wrap": 1, "joined_small_caps": 44, "page_number": 22, "repeated_review_header": 176, "separated_prose_lane": 1, "unmapped_glyph": 2}, "source_words": 13587} |
Aaz6R4Tlwv | Aaz6R4Tlwv | SynerGPT: In-Context Learning for Personalized Drug Synergy Prediction and Drug Design | colm | 2,024 | e53cd6a7dbe2a7c99846902303726a41abf5fea81df01affb1c72f003d1b9fcb | SynerGPT: In-Context Learning for Personalized Drug Synergy Prediction and Drug Design
Carl Edwards 1 , Aakanksha Naik 2 , Tushar Khot 2 , Martin D. Burke 1 , Heng Ji 1 , Tom Hope 2,3
1 University of Illinois Urbana-Champaign, 2 Allen Institute for Artificial Intelligence, 3 The Hebrew University of Jerusalem
{cne2, md... | a442c2149dc642959391893152e0f7ecafd0baaad4207dbb8b476373504d668c | 589ebc87bd38db44e0afcf2821ad092abaf1e1de703773b0a875d1a1a788662d | positioned-clean-v2 | 14 | {"removed_words": 290, "retained_words": 15273, "rules": {"dehyphenated": 24, "joined_hyphenated_wrap": 2, "page_number": 32, "repeated_review_header": 256, "separated_prose_lane": 2, "unmapped_glyph": 2}, "source_words": 15563} |
B41hNBoWLo | B41hNBoWLo | TOFU: A Task of Fictitious Unlearning for LLMs | colm | 2,024 | b30b14709526cb622ea81c5c2b3e2e24ffc0e7879cc89a897b2362be4ec05fd6 | TOFU: A Task of Fictitious Unlearning for LLMs
Pratyush Maini ∗
pratyushmaini@cmu.edu
Carnegie Mellon University
Zhili Feng ∗
zhilif@andrew.cmu.edu
Carnegie Mellon University
Avi Schwarzschild ∗
schwarzschild@cmu.edu
Carnegie Mellon University
Zachary C. Lipton
Carnegie Mellon University
J. Zico Kolter
Carnegie Mellon ... | 8a9f8ff8f082e1f8b49643c0c5043f9cdbeffd6164ee64e4a0a06ca0c11b63fc | e5e511c439675f0aa7ec74f95caf1c557d3c223e73132c309337af04e58bb73b | positioned-clean-v2 | 6 | {"removed_words": 239, "retained_words": 13303, "rules": {"dehyphenated": 11, "page_number": 26, "repeated_review_header": 208, "separated_prose_lane": 1, "unmapped_glyph": 5}, "source_words": 13542} |
BAakY1hNKS | BAakY1hNKS | AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversations | colm | 2,024 | b1050948cf2c3aceaed607f650b02c04d1e3225bf364bcb0b3690bfe5d90bad8 | AutoGen: Enabling Next-Gen LLM Applications
via Multi-Agent Conversations
Qingyun Wu ∗
Penn State University
qingyun@autogen.team
Gagan Bansal
Microsoft Research
gaganbansal@microsoft.com
Jieyu Zhang
University of Washington
jieyuz2@cs.washington.edu
Yiran Wu
Penn State University
yiran.wu@psu.edu
Beibin Li
Microsoft R... | 5145c555344b583715ba3a98d8c00e00a79b98295f1e19d19b5abf4243fbb412 | 70ce66e2ad0a8af89dc727c51382d170ced4a898757d62244e8fb1f1571bb417 | positioned-clean-v2 | 23 | {"removed_words": 414, "retained_words": 26548, "rules": {"dehyphenated": 23, "joined_hyphenated_wrap": 17, "joined_split_letters": 565, "page_number": 46, "repeated_review_header": 368}, "source_words": 26962} |
BDBdblmyzY | BDBdblmyzY | Automata-based constraints for language model decoding | colm | 2,024 | 3d4e6c3b20bce5081081936d07d0132b65ae2610b6ff44cf7ee6ccf572231475 | Automata-based constraints for language model decoding
Terry Koo ∗ , Frederick Liu ∗ , and Luheng He
Google DeepMind
{terrykoo,frederickliu,luheng}@google.com
Abstract
Language models (LMs) are often expected to generate strings in some formal language; for example, structured data, API calls, or code snip-
pets. Altho... | 0de4de0f8b10d27fc174c5832c20e2d100d980db3199b06f422281df667e149c | 8fd6058cf293acb82fd38cc5ac5f45b92940a68d41c97df8132fdeba81ee2d35 | positioned-clean-v2 | 13 | {"removed_words": 185, "retained_words": 11165, "rules": {"dehyphenated": 8, "joined_hyphenated_wrap": 3, "joined_small_caps": 23, "page_number": 20, "repeated_review_header": 160, "unmapped_glyph": 5}, "source_words": 11350} |
BaOAvPUyBO | BaOAvPUyBO | Do Language Models Plan Ahead for Future Tokens? | colm | 2,024 | f77ae321b8e376b4b258f8aacbdd2754dc8c2cc7949f5f30ae5f5aadf5051f7f | Do Language Models Plan Ahead for Future Tokens?
John X. Morris
Department of Computer Science
Cornell University
jxm3@cornell.edu
Wilson Wu
Department of Mathematics
University of Colorado Boulder
wiwu2390@colorado.edu
Lionel Levine
Department of Mathematics
Cornell University
levine@math.cornell.edu
Abstract
Do trans... | 13c066868fed0144c78f09ef43e513df39b7b4517d0260e25ebf645b5ed5abca | f12ba62de91b0c6f17776f5cd9f863660a9dc2c8592f930bb8c94d4843c9c826 | positioned-clean-v2 | 7 | {"removed_words": 231, "retained_words": 11590, "rules": {"combined_negation": 2, "composed_negation": 2, "dehyphenated": 7, "mapped_symbol_piece": 16, "page_number": 24, "repeated_review_header": 192, "separated_prose_lane": 2, "unmapped_glyph": 13}, "source_words": 11821} |
BgvgMxY8s5 | BgvgMxY8s5 | Handling Open-Vocabulary Constructs in Formalizing Specifications: Retrieval Augmented Parsing with Expert Knowledge | colm | 2,024 | 52a346d6abda4feb3f7c21fa0d94d1704ffc53aac8781163a75fa7bf68c7dff7 | Handling Open-Vocabulary Constructs in Formalizing Speci-
fications: Retrieval-Augmented Parsing with Expert Knowl-
edge
Mohammad Saqib Hasan ∗ , Sayontan Ghosh ∗ , Dhruv Verma ∗ , Geoff Kuenning † , Erez Zadok ∗ ,
Scott A. Smolka ∗ & Niranjan Balasubramanian ∗
∗ Department of Computer Science, Stony Brook University †... | bbb1c311752929f76ff3081f8d35d6a3c0c00b50c5c3427b916e941263b648fa | 8d555afefac70fb7357c43583163a45ac6699e76ddee70e03c0ef921a164dc38 | positioned-clean-v2 | 52 | {"removed_words": 182, "retained_words": 11023, "rules": {"dehyphenated": 5, "dropped_unmapped_glyph": 38, "joined_hyphenated_wrap": 7, "joined_small_caps": 18, "page_number": 20, "repeated_review_header": 160, "unmapped_glyph": 2}, "source_words": 11205} |
C0j44uRPcl | C0j44uRPcl | On Robustness-Accuracy Characterization of Language Models using Synthetic Datasets | colm | 2,024 | 121a940e251a3fa250e6835c93f268107a01ab645c8a050316fa6a69194125fd | On Robustness-Accuracy Characterization of Language Models using Synthetic Datasets
Pin-Yu Chen & Payel Das
IBM
pin-yu.chen@ibm.com, daspa@us.ibm.com
Ching-Yun Ko
MIT
cyko@mit.edu
Yung-Sung Chuang & Luca Daniel
MIT
yungsung@mit.edu, luca@mit.edu
Abstract
In recent years, language models (LMs) that were pretrained at sc... | f0248dae792ec8fdc357851372615c14c1ec2e4db9e2f0540c6b4e03522abda1 | f2a85d571c22d805da3e3f1775cc4c66d9e6426d5933999386431592b4c48e5c | positioned-clean-v2 | 20 | {"removed_words": 293, "retained_words": 16492, "rules": {"combined_negation": 6, "composed_negation": 6, "dehyphenated": 12, "dropped_unmapped_glyph": 1, "joined_hyphenated_wrap": 3, "page_number": 30, "repeated_review_header": 240, "separated_prose_lane": 1, "unmapped_glyph": 17}, "source_words": 16785} |
CI7D2kiih1 | CI7D2kiih1 | Should We Attend More or Less? Modulating Attention for Fairness | colm | 2,024 | 3150497f47848976a969f85ff5745c71fbef694d89246d3d4b917cb46be773f8 | Should We Attend More or Less? Modulating Attention for
Fairness
Abdelrahman Zayed 1,2,3 *, Gonçalo Mordido 1,2,3 , Samira Shabanian 4 , ∗ Sarath Chandar 1,2,3,5
1 Chandar Research Lab
2 Mila - Quebec AI Institute
3 Polytechnique Montreal
4 Independent Researcher
5 Canada CIFAR AI Chair
{zayedabd,goncalo-filipe.torcato... | 00435bc1d0cbd55f0457d80b6ab05ec2a933f49f8377e73408f0058b3ffc0bce | a9ca66f862f75ebe0c29d40ba6caa30094ab6ee01f5cfeb68554b82d8b348f1b | positioned-clean-v2 | 7 | {"removed_words": 198, "retained_words": 10108, "rules": {"dehyphenated": 13, "joined_hyphenated_wrap": 1, "page_number": 22, "repeated_review_header": 176}, "source_words": 10306} |
CrzAj0kZjR | CrzAj0kZjR | STaR-GATE: Teaching Language Models to Ask Clarifying Questions | colm | 2,024 | d8294027875af979b5152abcbc2bdb61170f44197c0a0a71051847be2d218cbf | STaR-GATE:
Teaching Language Models to Ask Clarifying Questions
Chinmaya Andukuri ∗
Stanford University
andukuri@stanford.edu
Jan-Philipp Fränken ∗
Stanford University
jphilipp@stanford.edu
Tobias Gerstenberg
Stanford University
gerstenberg@stanford.edu
Noah D. Goodman
Stanford University
ngoodman@stanford.edu
Abstract... | 1858a13e48afc0c9eb333c1c8ea8c469392bb576d2636f844f86bccacf5c9d78 | 5207161fd2643a5f6fa4fadc97346453922b10867d8796b3fb2bf66be52fd07e | positioned-clean-v2 | 16 | {"removed_words": 227, "retained_words": 10038, "rules": {"combined_negation": 2, "composed_negation": 2, "dehyphenated": 16, "joined_hyphenated_wrap": 3, "page_number": 25, "repeated_review_header": 200, "separated_prose_lane": 1}, "source_words": 10265} |
CybBmzWBX0 | CybBmzWBX0 | Length-Controlled AlpacaEval: A Simple Debiasing of Automatic Evaluators | colm | 2,024 | 20cbf04d52414f03895e678ff29a4bdebdc82ec44b30c2ff2d9bc67380f9c0a1 | Length-Controlled AlpacaEval:
A Simple Way to Debias Automatic Evaluators
Yann Dubois & Percy Liang & Tatsunori B. Hashimoto
Stanford University
{yanndubs,pliang,thashim}@stanford.edu
Abstract
LLM-based auto-annotators have become a key component of the LLM development process due to their cost-effectiveness and scalab... | 2ffded5938cc597e90c7c5bcf15a0c28c6149ba7a9b89d5e2d6fd0724b66010f | e2ef8c379638955e9d7ad54568a3acbad34dc79fb7d044ee126be666a8be0a04 | positioned-clean-v2 | 12 | {"removed_words": 107, "retained_words": 5644, "rules": {"dehyphenated": 7, "joined_hyphenated_wrap": 1, "page_number": 11, "repeated_review_header": 88, "unmapped_glyph": 8}, "source_words": 5751} |
D06yk3DBas | D06yk3DBas | Can It Edit? Evaluating the Ability of Large Language Models to Follow Code Editing Instructions | colm | 2,024 | 435112e6929eaa0c356a8f6960407ef25f187eee99ef848c512d229a206a3052 | Can It Edit? Evaluating the Ability of Large Language Models to Follow Code Editing Instructions
Federico Cassano
Northeastern University
cassano.f@northeastern.edu
Luisa Li
Northeastern University
Akul Sethi
Northeastern University
Noah Shinn
Northeastern University
Abby Brennan-Jones
Wellesley College
Jacob Ginesin
N... | 666a14b4006feed5c71637e66529d0996d4c95597e9c86af2aa583d25e52f721 | e18306edf175203bc25e004c13158632ccb19bdcfc94cdc5daed4555f94fa2d5 | positioned-clean-v2 | 28 | {"removed_words": 414, "retained_words": 20808, "rules": {"dehyphenated": 20, "joined_hyphenated_wrap": 4, "joined_small_caps": 250, "page_number": 46, "repeated_review_header": 368, "separated_prose_lane": 1}, "source_words": 21222} |
DMUGTMWrKZ | DMUGTMWrKZ | Enhancing Adversarial Robustness of LLMs with Analytic Hierarchy Process | colm | 2,024 | 64eec6c279bc4eee9f6ce253d9615dc99f91be85c8090d1b984030b30fb9b6f8 | Enhancing Adversarial Robustness of LLMs with Analytic
Hierarchy Process
Jiahao Zhao & Minzheng Wang
MAIS, Institute of Automation, Chinese Academy of Sciences School of Artificial Intelligence, University of Chinese Academy of Sciences
Beijing, China
{zhaojiahao2019, wangminzheng2023}@ia.ac.cn
Nan Xu ∗
MAIS, Institute... | e65c92b4f6d807e5adbfb9dd670c94c890f2efb83f72ad51a06ba9572b3a9b4d | 2ad9a79614fd4ad154be1b014c4944c8102b9b8a4db80c2ab220427d29e3c103 | positioned-clean-v2 | 9 | {"removed_words": 162, "retained_words": 9311, "rules": {"dehyphenated": 18, "joined_hyphenated_wrap": 3, "page_number": 18, "repeated_review_header": 144}, "source_words": 9473} |
DOMP5AgwQz | DOMP5AgwQz | CTIKG: LLM-Powered Knowledge Graph Construction from Cyber Threat Intelligence | colm | 2,024 | 5a6f523fc31fa22b9fadc701eb714d25a65b12a9901d1400060826087fa026bb | CTIKG: LLM-Powered Knowledge Graph Construction from
Cyber Threat Intelligence
Liangyi Huang, Xusheng Xiao
School of Computing and Augmented Intelligence Arizona State University, Tempe, AZ, USA
{lhuan139, xusheng.xiao}@asu.edu
Abstract
To gain visibility into evolving threat landscape, knowledge of cyber threats has b... | 0e303c129beb2d52f97adf82dcd91e8cebe0614c59ee34acc076695f342936a2 | d955a155651613b4b22f095d9b369322e0fbb4ed30d0b73c571051698e490e18 | positioned-clean-v2 | 6 | {"removed_words": 216, "retained_words": 10623, "rules": {"dehyphenated": 11, "joined_hyphenated_wrap": 8, "joined_small_caps": 117, "page_number": 24, "repeated_review_header": 192, "separated_prose_lane": 2}, "source_words": 10839} |
DRffhKBVlE | DRffhKBVlE | LITE: Modeling Environmental Ecosystems with Multimodal Large Language Models | colm | 2,024 | f80c104fc441a07b758b3fa538bcd60e2aa6f6f540f84ea99bdd72740f5719ef | LITE: Modeling Environmental Ecosystems with Multimodal
Large Language Models
Haoran Li ‡ , Junqi Liu ¶ , Zexian Wang § , Shiyuan Luo † , Xiaowei Jia † , Huaxiu Yao ∗
∗ UNC-Chapel Hill, ‡ PolyU, § UMich, ¶ Independent Researcher, † University of Pittsburgh
haoran.li.cs@gmail.com, huaxiu@cs.unc.edu
Abstract
The modeling... | 50a630848a3fc092af1299e6641058c2679f4b882ea87790c151dcb57aa9d06c | 5e41814254b8a05345638e76c0d9cbaf69e8f6637791abfc6c47e2c02fac1b45 | positioned-clean-v2 | 20 | {"removed_words": 136, "retained_words": 8316, "rules": {"dehyphenated": 26, "dropped_invisible_character": 1, "joined_hyphenated_wrap": 6, "page_number": 15, "repeated_review_header": 120, "separated_prose_lane": 1, "unmapped_glyph": 1}, "source_words": 8452} |
DbsLm2KAqP | DbsLm2KAqP | CULTURE-GEN: Revealing Global Cultural Perception in Language Models through Natural Language Prompting | colm | 2,024 | b461f84f3991f716c03813dd5882e0108cad44a5654b5bb3068ef18426c7ed89 | 🗺 ⽂ A CULTURE -GEN : Revealing Global Cultural Perception in Language Models through Natural Language Prompting
Huihan Li 1 , Liwei Jiang 2 , Nouha Dziri 3 , Xiang Ren 1 & Yejin Choi 2,3
1 University of Southern California
2 University of Washington
3 Allen Institute of Artificial Intelligence
huihanl@usc.edu, lwjiang... | 0187d442dc3b0fa000604fb98e75b15296c4b91c42cc0fd830dba131753ba34b | a5e46e8e480e371721a656183c7089eb63a28c8345659ea7ce6246b438887872 | positioned-clean-v2 | 9 | {"removed_words": 180, "retained_words": 7943, "rules": {"dehyphenated": 22, "joined_hyphenated_wrap": 6, "joined_small_caps": 26, "page_number": 20, "repeated_review_header": 160, "separated_prose_lane": 1}, "source_words": 8123} |
DomBynQsqt | DomBynQsqt | 3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation | colm | 2,024 | dbbceaeaf1b2ac9800315f6cbf8f849d35e587f6041a501cfeb9869187cebb3c | 3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation
Huaisheng Zhu* 1,2 , Teng Xiao* 1,2 & Vasant G Honavar 1,2,3,4
1 Artificial Intelligence Research Laboratory 2 College of Information Sciences and Technology 3 Center for Artificial Intelligence Foundations and Scientific Appl... | 977600a244c828033f8913609199e89f6b9d51af80b8f6f922750d2bf291b44c | e1451e51f7b0c806094672f48d1f16349ce0f0c400c7869fd6341dcc1f87931d | positioned-clean-v2 | 21 | {"removed_words": 236, "retained_words": 12246, "rules": {"dehyphenated": 20, "expanded_ligature": 8, "joined_hyphenated_wrap": 4, "page_number": 23, "repeated_review_header": 184, "separated_prose_lane": 2, "unmapped_glyph": 29}, "source_words": 12482} |
Dt6qXZsgaU | Dt6qXZsgaU | Self-Guide: Better Task-Specific Instruction Following via Self-Synthetic Finetuning | colm | 2,024 | 1e8046ba82667ec3498e5ab3fc2c5b9376dc7bd96752e2993cee323f9d76fa8e | SELF -GUIDE : Better Task-Specific Instruction Following via
Self-Synthetic Finetuning
Chenyang Zhao 1* , Xueying Jia 2* ,
Vijay Viswanathan 2 , Graham Neubig 2 , Tongshuang Wu 2
1 Department of Computer Science and Technology, Tsinghua University, Beijing, China 2 Language Technologies Institute, Carnegie Mellon Unive... | bb15af316e0e0e9b3b3167671627951ccedc9a0d3b65e2b6d8d8461ba5043d47 | 3f90def379133d949d05328d9abd2c0de9fc1a15533e0cdaed930133f40f0761 | positioned-clean-v2 | 9 | {"removed_words": 153, "retained_words": 8357, "rules": {"dehyphenated": 8, "joined_hyphenated_wrap": 3, "joined_small_caps": 142, "joined_split_letters": 190, "page_number": 17, "repeated_review_header": 136}, "source_words": 8510} |
EEPBOB2Xww | EEPBOB2Xww | Ferret-v2: An Improved Baseline for Referring and Grounding with Large Language Models | colm | 2,024 | a2b03d0b674f56eca51f8355532f15e2290c10ca9332659f414d5c55c469eace | Ferret-v2: An Improved Baseline for Referring and Ground-
ing with Large Language Models
Haotian Zhang 1∗ , Haoxuan You 2∗ , Philipp Dufter 1 , Bowen Zhang 1 , Chen Chen 1 ,
Hong-You Chen 1 , Tsu-Jui Fu 3 , William Yang Wang 3 , Shih-Fu Chang 2 , Zhe Gan 1 , Yinfei Yang 1
1
Apple AI/ML, 2 Columbia University, 3 UC Sant... | ac409662be033c3f278c45a4895c4f92151e3512680644ed8d555970d1abfa5a | 4e5b5db1c5fb44285f325d564d779bd6c40802f290042ba85bc8e9c88674c3df | positioned-clean-v2 | 10 | {"removed_words": 135, "retained_words": 8247, "rules": {"dehyphenated": 17, "joined_hyphenated_wrap": 4, "page_number": 15, "repeated_review_header": 120}, "source_words": 8382} |
EHPns3hVkj | EHPns3hVkj | Tower: An Open Multilingual Large Language Model for Translation-Related Tasks | colm | 2,024 | 6db253e160ff28b72ffaf2ad28fc1968e17d2ed352e8b3257e562b255c8181df | TOWER : An Open Multilingual Large Language
Model for Translation-Related Tasks
Duarte M. Alves ∗ 2,3 , José Pombal ∗ 1,2,3 , Nuno M. Guerreiro ∗ 1,2,3,4
Pedro H. Martins 1 , João Alves 1 , Amin Farajian 1 , Ben Peters 2 ,
Ricardo Rei 1 , Patrick Fernandes 2,6 , Sweta Agrawal 2 ,
Pierre Colombo 4,5 , José G.C. de Souza... | 00d48011086ed173b8031d77111a1e36262f202003f5f4b88faee40796578766 | 8d8c173a72a48b44435764fd64a67262b8b3563b84632c78e464edfcd0307558 | positioned-clean-v2 | 31 | {"removed_words": 306, "retained_words": 15166, "rules": {"dehyphenated": 11, "joined_hyphenated_wrap": 2, "joined_small_caps": 535, "page_number": 34, "repeated_review_header": 272}, "source_words": 15472} |
EIjJ6ykPnh | EIjJ6ykPnh | D2PO: Discriminator-Guided DPO with Response Evaluation Models | colm | 2,024 | c55a51d31e5d7e487e07dece3cd0f71ac52f3da23cecc88bd3f6f7a56f0389df | D2PO: Discriminator-Guided DPO
with Response Evaluation Models
♡
♠
♣
♢
♡
Prasann Singhal , Nathan Lambert , Scott Niekum , Tanya Goyal , Greg Durrett
♡
♠
The University of Texas at Austin, Allen Institute for Artificial Intelligence
♣
♢
University of Massachusetts Amherst, Princeton University
prasanns@cs.utexas.edu
Ab... | b9d6ad928cfca95b8b2fbcc264eb394d223c313532fd6715e982cf67bd86a78e | 96e9e59a84bfa9146656fc8a7d417411b37d5608d91babc91c0052ba3b8d2dfb | positioned-clean-v2 | 28 | {"removed_words": 180, "retained_words": 11722, "rules": {"dehyphenated": 17, "joined_small_caps": 14, "page_number": 20, "repeated_review_header": 160, "separated_prose_lane": 4}, "source_words": 11902} |
EKBPn7no4y | EKBPn7no4y | StructLM: Towards Building Generalist Models for Structured Knowledge Grounding | colm | 2,024 | 78063e590b17f8a146c9983f5dd307da7a0af2e6d81236f23503a49df47e094d | StructLM: Towards Building Generalist Models for Structured
Knowledge Grounding
Alex Zhuang 1 , ∗ Ge Zhang 1,2,7 , ∗
Tianyu Zheng 2 , Xinrun Du 2 , Junjie Wang 3,4 , Weiming Ren 1,7 ,
Stephen W. Huang 6 , Jie Fu 2,4† , Xiang Yue 2,5 , Wenhu Chen 1,2,7†
1 University of Waterloo, 2 Multimodal Art Projection Research Comm... | 5d96ad8000fe70b3f2691aa2fdaed3db58c93313300a44955e74c68f51e515b7 | 3244147a35c15bfe4e0635d6599df79333ebd87585576ea9abbd22b46ff3f148 | positioned-clean-v2 | 18 | {"removed_words": 198, "retained_words": 10047, "rules": {"dehyphenated": 9, "joined_hyphenated_wrap": 7, "page_number": 22, "repeated_review_header": 176}, "source_words": 10245} |
Ecgev5ZZpq | Ecgev5ZZpq | Evaluating the Adversarial Robustness of Retrieval-Based In-Context Learning for Large Language Models | colm | 2,024 | 63d5568b5b7f21cda1a5d19b85b5d2e20b77faf23d0473f43fdcbdcb754b2e1e | Evaluating the Adversarial Robustness of Retrieval-Based In-Context Learning for Large Language Models
Simon Yu ∗ † , Jie He ∗ , Pasquale Minervini & Jeff Z. Pan †
School of Informatics
University of Edinburgh
{s1967531,j.he,j.z.pan}@ed.ac.uk , pminervi@exseed.ed.ac.uk
Abstract
With the emergence of large language mode... | df9c66bf8712561d2993560f4bd35d7c2432d075d35caf9faf2414ad18ec1f10 | 7395a4122c585d78b6acf2dbf27c1ef65a4f10c08faa71c9566239721d5cc229 | positioned-clean-v2 | 9 | {"removed_words": 270, "retained_words": 15251, "rules": {"dehyphenated": 20, "joined_hyphenated_wrap": 1, "joined_small_caps": 14, "mapped_symbol_piece": 4, "page_number": 30, "repeated_review_header": 240}, "source_words": 15521} |
F2yGbwXJAi | F2yGbwXJAi | Suspicion Agent: Playing Imperfect Information Games with Theory of Mind Aware GPT-4 | colm | 2,024 | d68089a2c707435bb23c9adb2b52aa00768a965fe57d1fb4d19d71b7448a6c51 | Suspicion-Agent: Playing Imperfect Information Games with Theory of Mind Aware GPT-4
Jiaxian Guo ∗ † 1
Bo Yang * 1
Paul Yoo 1
1
Yusuke Iwasawa
Yutaka Matsuo 1
The University of Tokyo 1 Allen Institute for AI 2
Bill Yuchen Lin 2
Abstract
Unlike perfect information games, where all elements are known to every player, imp... | e562bbcaf91e77449740378b24e75b78dac10cdc45f61cae2793892f3ad38682 | efdcf27efd7625a231a41f4bd76dbbb4929b2674137ac0694003bec09468c6b9 | positioned-clean-v2 | 18 | {"removed_words": 261, "retained_words": 18854, "rules": {"dehyphenated": 20, "joined_hyphenated_wrap": 6, "page_number": 29, "repeated_review_header": 232}, "source_words": 19115} |
F7aAhfitX6 | F7aAhfitX6 | Massive Activations in Large Language Models | colm | 2,024 | 79d99d65f353119fd3a5770bc9f2c1bb209b123c33a9859257f8c3940ca4cdbe | Massive Activations in Large Language Models
2k
1k
1k
1k
1k
0
0
0
0
2533
1415
2533
1415
2533
1415
2533
1415
It
2k
t is
be he
st
tim of
es
.
\n
W
e
mm
sum hy
is
wa mer
rm
win and
t
col er
d
?
2k
wa is r
r
Wi \n m
nte
r
col is
d
\n .
2k
Su
Su
mm
e
wa is r
r
Wi m
nte .
r
col is
d
\n .
Mingjie Sun 1 ∗
Xinlei Chen 2
J. Zico... | bb233d08bc7ea3dc28b013504b4af5e7509a2fc68c22c6a9650da3c49185aac9 | b45340b8cf0aa72b0fba10076b3854c4deb05e523c1513a45e2bf972341bef7f | positioned-clean-v2 | 7 | {"removed_words": 247, "retained_words": 11162, "rules": {"dehyphenated": 14, "page_number": 26, "repeated_review_header": 208, "separated_prose_lane": 1, "unmapped_glyph": 13}, "source_words": 11409} |
F9tqgOPXH5 | F9tqgOPXH5 | EnvGen: Generating and Adapting Environments via LLMs for Training Embodied Agents | colm | 2,024 | 71fe0933624bea0aacc646ad3ffb09472b0f2b9b21dd243955e59ec0a4cf8cf9 | EnvGen: Generating and Adapting Environments via LLMs for Training Embodied Agents
Abhay Zala ∗ Jaemin Cho ∗ Han Lin Jaehong Yoon Mohit Bansal
UNC Chapel Hill
{aszala, jmincho, hanlincs, jhyoon, mbansal}@cs.unc.edu
https://envgen-llm.github.io
Abstract
Recent state-of-the-art approaches for embodied learning via intera... | 40b148db908fbc58f70dc844b9f0f54d44f7628ecd603252b65bda225818896b | 51658d065862ee8b590bac8b89360fe62b0939779d2af188fc53cef744282c16 | positioned-clean-v2 | 11 | {"removed_words": 207, "retained_words": 12364, "rules": {"dehyphenated": 15, "joined_hyphenated_wrap": 5, "page_number": 23, "repeated_review_header": 184}, "source_words": 12571} |
FX4fUThO9H | FX4fUThO9H | Model Autophagy Analysis to Explicate Self-consumption within Human-AI Interactions | colm | 2,024 | b112522f6309b9d66886489883d4cab8d44926aee96383f84aecae198173c642 | Model Autophagy Analysis to Explicate Self-consumption
within Human-AI Interactions
Shu Yang ∗ ,1,2,3 , Muhammad Asif Ali ∗ ,1,2 , Lu Yu 5 , Lijie Hu † ,1,2,4 , and Di Wang †,1,2,4
1 Provable Responsible AI and Data Analytics (PRADA) Lab 2 King Abdullah University of Science and Technology 3 University of Macau
4 SDAIA... | d2efecfa9fe7043d564ed657a820b7bfbf96fea0ec53389394b33490744a415f | a2d2ab91dd7ce2774f9bfaa64c34411aa88bf45fde79fe84d3d4bb150e5f059c | positioned-clean-v2 | 26 | {"removed_words": 225, "retained_words": 10649, "rules": {"dehyphenated": 11, "joined_hyphenated_wrap": 5, "joined_small_caps": 14, "page_number": 25, "repeated_review_header": 200}, "source_words": 10874} |
FbhjirzvJG | FbhjirzvJG | Hydra: Sequentially-Dependent Draft Heads for Medusa Decoding | colm | 2,024 | a825079286e433820befe951e51d13f5cedb046d1919d3c3ba6d8b865720625a | Hydra: Sequentially-Dependent Draft Heads for Medusa De-
coding
Zachary Ankner 1,2, ∗ Rishab Parthasarathy 1, ∗ Aniruddha Nrusimha 1
Christopher Rinard 1 Jonathan Ragan-Kelley 1 William Brandon 1
1 MIT 2 MosaicML
Abstract
To combat the memory bandwidth-bound nature of autoregressive LLM inference, previous research has... | b318a7379ef601f8086a2e41a4d6715b3b188aef5cbb482dfc98818c0de57de6 | a921cdf869cca771f349778bb976495054ce1809a2df46c71269178c6e166820 | positioned-clean-v2 | 8 | {"removed_words": 153, "retained_words": 8733, "rules": {"dehyphenated": 14, "joined_hyphenated_wrap": 1, "page_number": 17, "repeated_review_header": 136}, "source_words": 8886} |
FgHpT6u7pk | FgHpT6u7pk | CoCA: Regaining Safety-awareness of Multimodal Large Language Models with Constitutional Calibration | colm | 2,024 | b32dc4c6a53f8bca014862374e781b9835eee5b55aa2494fc965eccc61b4f721 | CoCA: Regaining Safety-awareness of Multimodal Large Language Models with Constitutional Calibration
Jiahui Gao 1 , Renjie Pi 2 , Tianyang Han 3 , Han Wu 1 , Lanqing Hong 1 ,
Lingpeng Kong 4 , Xin Jiang 1 , Zhenguo Li 1
1 Noah’s Ark Lab
2 The Hong Kong University of Science and Technology
3 The Hong Kong Polytechnic Un... | 3a874d1d19f8ce7fbb52b4490407345e6115285f679832b868ce354e527ba336 | 4906868eddf6c2a899024d8077265ea90af40e59f96a86e2bfb8727ddfcf8975 | positioned-clean-v2 | 11 | {"removed_words": 180, "retained_words": 9644, "rules": {"dehyphenated": 10, "joined_hyphenated_wrap": 1, "page_number": 20, "repeated_review_header": 160}, "source_words": 9824} |
Fkr1yVUb9G | Fkr1yVUb9G | CoLLEGe: Concept Embedding Generation for Large Language Models | colm | 2,024 | a1036f53301e389e5587bb285420db2cd5e1c4b132c8e32faaaf7b91c97ec6aa | CoLLEGe: Concept Embedding Generation for Large Language Models
Ryan Teehan, Brenden M. Lake & Mengye Ren
New York University
{rst306,brenden,mengye}@nyu.edu
Abstract
Current language models are unable to quickly learn new concepts on the fly, often requiring a more involved finetuning process to learn robustly. Prompt... | 013a711096961beabaf966d41f3c48223b4bf10d6bd28ce5ab9d340658d21742 | bf6c17cf47977828b5f60f965d1bc2d9f64a95e0dc5d709a77af9e0ecfe24c1c | positioned-clean-v2 | 11 | {"removed_words": 162, "retained_words": 10297, "rules": {"composed_negation": 2, "dehyphenated": 11, "page_number": 18, "repeated_review_header": 144}, "source_words": 10459} |
FmhPg4UJ9K | FmhPg4UJ9K | Counting Like Transformers: Compiling Temporal Counting Logic Into Softmax Transformers | colm | 2,024 | e8d3002c175993e72fdd660ee6f25e4f4c348cb37d2584638fcb75f5b693697f | Counting Like Transformers: Compiling Temporal Counting Logic Into Softmax Transformers
Andy Yang & David Chiang
University of Notre Dame
{ayang4,dchiang}@nd.edu
Abstract
Deriving formal bounds on the expressivity of transformers, as well as studying transformers that are constructed to implement known algorithms, are ... | 3721e71b5f7614ba3c5ade518b1a18b9cc6e4bf4b98e0ac808ce4d2322d79da0 | 2a9889a20f3d896022af0cebfa7202d27ab36d4b779c7ea3ddfb1c6a8e414898 | positioned-clean-v2 | 17 | {"removed_words": 311, "retained_words": 15860, "rules": {"combined_negation": 2, "composed_negation": 3, "dehyphenated": 32, "dropped_unmapped_glyph": 1, "joined_hyphenated_wrap": 6, "mapped_symbol_piece": 330, "page_number": 26, "repeated_review_header": 208, "unmapped_glyph": 75}, "source_words": 16171} |
G8LaO1P0xv | G8LaO1P0xv | A Long Way to Go: Investigating Length Correlations in RLHF | colm | 2,024 | 4673fe116c9f4fa92e0345778d4e3227595e5ed7e389d8698337943e888c1299 | A Long Way to Go: Investigating Length Correlations in RLHF
Prasann Singhal
The University of Texas at Austin
prasanns@cs.utexas.edu
Tanya Goyal
Princeton University
tanyagoyal@princeton.edu
Jiacheng Xu
Salesforce AI
jiacheng.xu@salesforce.com
Greg Durrett
The University of Texas at Austin
gdurrett@cs.utexas.edu
Abstra... | 4d76d3af77c42413c283db700c29889631eb33f47074b742dc6a6c90f4686aa8 | f850797ae3697bb80f7f1d8179abafab9740b2b7f0f03c2c068151c79e748f50 | positioned-clean-v2 | 24 | {"removed_words": 189, "retained_words": 11264, "rules": {"dehyphenated": 15, "joined_hyphenated_wrap": 1, "joined_small_caps": 12, "page_number": 21, "repeated_review_header": 168, "separated_prose_lane": 2}, "source_words": 11453} |
GC4mXVfquq | GC4mXVfquq | JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks | colm | 2,024 | 8bc1778e195d69aca03cabe6c4e9e8b5f9dfd20de50ae0921f9e789dfedb7c2d | JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks
Weidi Luo ∗
The Ohio State University
Columbus, OH 43210, USA
luo.1455@osu.edu
Siyuan Ma ∗
University of Wisconsin-Madison
Madison, WI 53706, USA
siyuan.ma.jasper@outlook.com
Xiaogeng Liu ∗
University of W... | 1647fe011aab652b20bd5fa315bbf69e78f73516b06794d8fe0ad701b0f41df5 | eb94fb0b0148c98e62609c387dc970737dc5c1357d52c9f8dab93dd2de0b14c6 | positioned-clean-v2 | 16 | {"removed_words": 189, "retained_words": 10053, "rules": {"dehyphenated": 5, "joined_hyphenated_wrap": 4, "page_number": 21, "repeated_review_header": 168, "separated_prose_lane": 1}, "source_words": 10242} |
GMalvQu0XL | GMalvQu0XL | RAVEN: In-Context Learning with Retrieval-Augmented Encoder-Decoder Language Models | colm | 2,024 | 57711fad7195e5d0415194bd8b68e7ada874a0decc9eb9b8d801ed1d9512ba5e | RAVEN : In-Context Learning with Retrieval-Augmented
Encoder-Decoder Language Models
Jie Huang 1,2, ∗ Wei Ping 2
Peng Xu 2
Mohammad Shoeybi 2
1
Kevin Chen-Chuan Chang
Bryan Catanzaro 2
1 University of Illinois at Urbana-Champaign
2 NVIDIA
jeffhj@illinois.edu, wping@nvidia.com
Abstract
In this paper, we investigate the ... | be4749f53868ef7b6c52c170ae8649c27c62837e9b1b641444cf887663b535fb | 0349bf8eafc04f09b55f49c0f459fff28d0e8af8daa5905ca254e669d2320664 | positioned-clean-v2 | 16 | {"removed_words": 180, "retained_words": 10449, "rules": {"dehyphenated": 18, "joined_hyphenated_wrap": 10, "joined_small_caps": 139, "page_number": 20, "repeated_review_header": 160, "separated_prose_lane": 1}, "source_words": 10629} |
GqDntYTTbk | GqDntYTTbk | Starling-7B: Improving Helpfulness and Harmlessness with RLAIF | colm | 2,024 | 308c09d9a9b0872fd25629f7f8c1951c6f0aadbf6835d4dc5c2e2d836607acfa | Starling-7B:
Improving Helpfulness and Harmlessness with RLAIF
Banghua Zhu ∗
UC Berkeley
Evan Frick ∗
UC Berkeley
Tianhao Wu ∗
UC Berkeley
Karthik Ganesan
Nexusflow AI
Wei-Lin Chiang
UC Berkeley
Jian Zhang
Nexusflow AI
Hanlin Zhu
UC Berkeley
Jiantao Jiao
UC Berkeley
Abstract
This paper presents Starling-7B, a strong 7B... | 2b36e2a3a5498812073058470807a18966158099ecefb48e0b0fbc7ccb4fb2f0 | aa359abf2a700a3e645866bfa7f549933273e8deced51284a977754267bdfb48 | positioned-clean-v2 | 13 | {"removed_words": 235, "retained_words": 10202, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 8, "joined_hyphenated_wrap": 2, "mapped_symbol_piece": 4, "page_number": 26, "repeated_review_header": 208}, "source_words": 10437} |
H1Edd5d2JP | H1Edd5d2JP | LLM4Causal: Democratized Causal Tools for Everyone via Large Language Model | colm | 2,024 | 49a72e83258f52f0983149c8b15e9b9f163355a1156a55870522164a5104c815 | LLM4Causal: Democratized Causal Tools for Everyone via Large Language Model
Haitao Jiang, Lin Ge, Yuhe Gao , ∗ Jianian Wang ∗ & Rui Song †
Department of Statistics
North Carolina State University
Raleigh, NC 27607, USA
{hjiang24,lge,ygao32,jwang95,rsong}@ncsu.edu
Abstract
Large Language Models (LLMs) have shown their s... | 00ad430497ca9c5911c5c4eb84914c0108845b8390eae87be47d0d10a3779587 | d5ffdc7ba32272f9f5f8ae3c6c887ef514e5241849804e8590093d805e1dd68b | positioned-clean-v2 | 17 | {"removed_words": 225, "retained_words": 12402, "rules": {"dehyphenated": 19, "joined_hyphenated_wrap": 4, "page_number": 25, "repeated_review_header": 200}, "source_words": 12627} |
HDkNbfLQgu | HDkNbfLQgu | Reverse Training to Nurse the Reversal Curse | colm | 2,024 | 67a59b442aaf587b2059a8a9d82395379d03c0861bc5284a6ed11f1b095741a0 | Reverse Training to Nurse the Reversal Curse
Olga Golovneva
FAIR at Meta
olggol@meta.com
Zeyuan Allen-Zhu
FAIR at Meta
Jason Weston
FAIR at Meta
Sainbayar Sukhbaatar
FAIR at Meta
sainbar@meta.com
Abstract
Large language models (LLMs) have a surprising failure: when trained on “A has a feature B”, they do not generalize... | e9fb90caa14798a25421f8fafbac6f4913d688f5219a0385ccd6d8db65462a00 | bb0993c8fcb5cab842e4ac87fa1e082b0d6c7286e3729ddcd32c5452c632f406 | positioned-clean-v2 | 9 | {"removed_words": 144, "retained_words": 9110, "rules": {"dehyphenated": 8, "joined_hyphenated_wrap": 4, "page_number": 16, "repeated_review_header": 128}, "source_words": 9254} |
HLoWN6m4fS | HLoWN6m4fS | Elephants Never Forget: Memorization and Learning of Tabular Data in Large Language Models | colm | 2,024 | 63566f36c3c1fb77f84e5b384aa5de2da10372283fd35b42d0f1e217f3da1421 | Elephants Never Forget: Memorization and Learning of Tabular Data in Large Language Models
Sebastian Bordt
University of Tübingen, Tübingen AI Center
sebastian.bordt@uni-tuebingen.de
Harsha Nori & Vanessa Rodrigues & Besmira Nushi & Rich Caruana
Microsoft Research
{hnori,vanessa.rodrigues,besmira.nushi,rcaruana}@micros... | a7208686639d87a91b3a560d16ea6e9d8ec91fe6d2da59fd50e4c8379f48b434 | 97605a3e68da3299d2d8c5812309f719d69fe730579f7a300b1184ba448792b6 | positioned-clean-v2 | 9 | {"removed_words": 279, "retained_words": 13868, "rules": {"dehyphenated": 14, "joined_hyphenated_wrap": 4, "page_number": 31, "repeated_review_header": 248}, "source_words": 14147} |
HVK6nl3i97 | HVK6nl3i97 | TriForce: Lossless Acceleration of Long Sequence Generation with Hierarchical Speculative Decoding | colm | 2,024 | 3cb90015c6a817718b7bc00f7465ecf59971b4c08bc300f0989b01db9f0aa26b | TRI FORCE : Lossless Acceleration of Long Sequence
Generation with Hierarchical Speculative Decoding
Hanshi Sun 1 , Zhuoming Chen 1 , Xinyu Yang 1 , Yuandong Tian 2 , Beidi Chen 1,2
1 Carnegie Mellon Univeristy
2 Meta AI (FAIR)
{hanshis, zhuominc, xinyuya2, beidic}@andrew.cmu.edu
yuandong@meta.com
Abstract
With large l... | 41070d70f77fc7160f0163ebe76bcbcefeff7103f1ea2e26a38859db37845bc1 | 776f28279f300db0e59894b9197f0c4a2393303cdc48b08a697cdb4f92abcb94 | positioned-clean-v2 | 16 | {"removed_words": 144, "retained_words": 8459, "rules": {"dehyphenated": 17, "joined_hyphenated_wrap": 4, "joined_small_caps": 164, "page_number": 16, "repeated_review_header": 128, "separated_prose_lane": 3}, "source_words": 8603} |
Hi8jKh4HE9 | Hi8jKh4HE9 | What is in Your Safe Data? Identifying Benign Data that Breaks Safety | colm | 2,024 | 975d1c5cb373d61b60b7858c6df9eee5f08ceec247b690988b05006bf9d2054c | What is in Your Safe Data?
Identifying Benign Data that Breaks Safety
Luxi He ∗
Mengzhou Xia ∗
Peter Henderson
Princeton Language and Intelligence (PLI), Princeton University
{luxihe, mengzhou, peter.henderson}@princeton.edu
Abstract
Current Large Language Models (LLMs), even those tuned for safety and alignment, are s... | 0d0aa736c3257bd114925aea28fc85afe1c8057069416a6215b216ea13a5a8e7 | 1c3936bf2d58c0eac67f08f7a4ef36da7406cb2589423160112dc7850f5ec6bf | positioned-clean-v2 | 17 | {"removed_words": 236, "retained_words": 11988, "rules": {"dehyphenated": 19, "dropped_invisible_character": 3, "joined_hyphenated_wrap": 6, "joined_small_caps": 85, "page_number": 26, "repeated_review_header": 208, "unmapped_glyph": 2}, "source_words": 12224} |
Hvq9RtSoHG | Hvq9RtSoHG | Chain-of-Symbol Prompting For Spatial Reasoning in Large Language Models | colm | 2,024 | 957c3a52e5b2d49c58ce440c4f2cb1c6d7c85f21b34ba5ab213d4fb57ede0155 | Chain-of-Symbol Prompting for Spatial Reasoning in Large
Language Models
Hanxu Hu 1,3 , ∗ Hongyuan Lu 2 ∗ , † Huajian Zhang 3 , Yun-Ze Song 1 , Wai Lam 2 , Yue Zhang 1
2 The Chinese University of Hong Kong
3 University of Edinburgh
huhanxu1233@gmail.com
{hylu,wlam}@se.cuhk.edu.hk
v1hzha17@exseed.ed.ac.uk
zhangyue@westl... | d537a560ea8312912b9981aedd796cd56a018333121ccc63818c653ef62dcbc3 | def7d19904a06e524418dc438fd9ad00b4a9c20f280086424bec4e3065c49a4a | positioned-clean-v2 | 10 | {"removed_words": 224, "retained_words": 18404, "rules": {"dehyphenated": 13, "joined_hyphenated_wrap": 2, "joined_small_caps": 72, "page_number": 24, "repeated_review_header": 200}, "source_words": 18628} |
IA8CWtNkUr | IA8CWtNkUr | Early Weight Averaging meets High Learning Rates for LLM Pre-training | colm | 2,024 | 747576d7e779e1522490a48cdbec1c597ec81c756548ffcfeb09a9c8d1c923ac | Early Weight Averaging meets High Learning Rates for LLM
Pre-training
Sunny Sanyal ∗
UT Austin
Atula Neerkaje ∗
UT Austin
Jean Kaddour
University College London
Abhishek Kumar
Google DeepMind
Sujay Sanghavi
UT Austin
Abstract
Training Large Language Models (LLMs) incurs significant cost; hence, any strategy that accele... | 6239ab9f899a2774757ba7aa3f3fb099ac296d673089c8b03b62363a70d903db | cb4cff1b06cc87f2fd276a503786bf2c350a474e5c748f06b11ab926576048f4 | positioned-clean-v2 | 15 | {"removed_words": 181, "retained_words": 10015, "rules": {"combined_negation": 1, "composed_negation": 1, "dehyphenated": 16, "joined_hyphenated_wrap": 2, "page_number": 20, "repeated_review_header": 160, "separated_prose_lane": 4}, "source_words": 10196} |
IBCBMeAhmC | IBCBMeAhmC | Evaluating Language Models for Efficient Code Generation | colm | 2,024 | 74f6a272ba509d403e841848ef0f8d148bc1ffefdc0209c2de4e4667443da59f | Evaluating Language Models for Efficient Code Generation
Jiawei Liu
Yuxiang Wei
Songrun Xie
Yifeng Ding
Junhao Wang
Lingming Zhang
University of Illinois Urbana-Champaign
{jiawei6, lingming}@illinois.edu
Tongji University
Abstract
We introduce Differential Performance Evaluation (DPE), a framework designed to reliably ... | 16aba132459e0be9877a0ddd1b98c96bd8c7a5bc5b8d7c88b7667c8f26512d59 | 2af0a3f7bed0cf53a81ecd89199b1f5e4c5a3f709457defc358a283ce6ab7b09 | positioned-clean-v2 | 12 | {"removed_words": 166, "retained_words": 9120, "rules": {"dehyphenated": 14, "joined_small_caps": 91, "joined_split_letters": 11, "page_number": 18, "repeated_review_header": 144, "unmapped_glyph": 4}, "source_words": 9286} |
INivcBeIDK | INivcBeIDK | AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models | colm | 2,024 | 73bb247a26d30817c5c18fccb3de1415b14367ba760819fa17399ee14ea1e4df | AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models
Warning: This paper contains potentially harmful AI-generated language.
Sicheng Zhu 1 ∗ , Ruiyi Zhang 2 , Bang An 1 , Gang Wu 2 , Joe Barrow 2 , Zichao Wang 2
Furong Huang 1 , Ani Nenkova 2 , Tong Sun 2
1 University of Maryland, College ... | 2fc0700834f14daaa09338aedd123b068974cc2e2c3ae74fe65ae0b3da4efc6c | c153405b448a186915af9aa983e52ae1f35905016a99a68e07d2578049ac9d43 | positioned-clean-v2 | 31 | {"removed_words": 318, "retained_words": 16745, "rules": {"combined_negation": 2, "composed_negation": 2, "dehyphenated": 24, "joined_hyphenated_wrap": 3, "page_number": 35, "repeated_review_header": 280, "separated_prose_lane": 2, "unmapped_glyph": 1}, "source_words": 17063} |
IPZ28ZqD4I | IPZ28ZqD4I | Faithful and Unfaithful Error Recovery in Chain of Thought | colm | 2,024 | 3187c99a13ff61afb5a22add8a18bc161feb6cb5dd8d214471eb34b6a92f7c33 | Dissociation of Faithful and Unfaithful Reasoning in LLMs
Evelyn Yee, Alice Li ∗ , Chenyu Tang*, Yeon Ho Jung † , Ramamohan Paturi ‡ , & Leon Bergen ‡
Laboratory of Emerging Intelligence
University of California, San Diego
{eyee,axl001,cht019,rpaturi,lbergen}@ucsd.edu § ; yjung-24@peddie.org
Abstract
Large language mod... | c85dc24fa72edf408ee43f02466274625703a68c957bd5f2dd689d53c8d04f6c | 7ed4e190d30165a959aa51877e31e18541d605e37cc369ae0ea57ec96a1d3637 | positioned-clean-v2 | 14 | {"removed_words": 315, "retained_words": 9826, "rules": {"dehyphenated": 11, "page_number": 35, "repeated_review_header": 280}, "source_words": 10141} |
IW1PR7vEBf | IW1PR7vEBf | LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders | colm | 2,024 | f22e208370a11de83995bda709da3fb460e03b6c3ad25b95a32d3a943ae6ecc1 | LLM2Vec: Large Language Models Are Secretly Powerful Text
Encoders
Parishad BehnamGhader ∗ , ⋄ Vaibhav Adlakha ∗ , ⋄ ,† Marius Mosbach ⋄
Dzmitry Bahdanau † Nicolas Chapados † Siva Reddy ⋄ ,†,‡
⋄ McGill University, Mila
† ServiceNow Research
‡ Facebook CIFAR AI Chair
{parishad.behnamghader,vaibhav.adlakha,marius.mosbach... | cdf3efdd8e0fefed6c441ef9e1fe2541b2fd6f87ead764a3726b83ef7159fa2a | ad66efd8299092f6c7cd7782846ba97c268dd59f327fff0d87640e5f48b873aa | positioned-clean-v2 | 13 | {"removed_words": 252, "retained_words": 13448, "rules": {"dehyphenated": 31, "joined_hyphenated_wrap": 3, "page_number": 28, "repeated_review_header": 224}, "source_words": 13700} |
JXcXnJJSuL | JXcXnJJSuL | Information-Theoretic Distillation for Reference-less Summarization | colm | 2,024 | c402e71de8f7102af74d62c9728ea934fa039720b99a91edd2a5262e0feb819e | Information-Theoretic Distillation for
Reference-less Summarization
Jaehun Jung † Ximing Lu † Liwei Jiang † Faeze Brahman †‡ Peter West † Pang Wei Koh † Yejin Choi †‡
† Paul G. Allen School of Computer Science & Engineering, University of Washington ‡ Allen Institute for Artificial Intelligence
hoony123@cs.washington.e... | 2792c5b94afa39208653785e4716e4459657b4df2734bd0be398809ace5e1ecf | 0e8a6dc95bf82773f294f23a0ff2169c8c9432a4347c21768ff04a272b88e031 | positioned-clean-v2 | 26 | {"removed_words": 241, "retained_words": 14070, "rules": {"dehyphenated": 30, "joined_hyphenated_wrap": 3, "joined_small_caps": 163, "mapped_symbol_piece": 3, "page_number": 27, "repeated_review_header": 208, "separated_prose_lane": 4, "unmapped_glyph": 6}, "source_words": 14311} |
Jd0bCD12DS | Jd0bCD12DS | Mind the Privacy Unit! User-Level Differential Privacy for Language Model Fine-Tuning | colm | 2,024 | 6cc9ecd5b33cb5c9e14fbaff5cbaeeba3fc981ab798e9d971fc06b96e88421d6 | Mind the Privacy Unit! User-Level Differential Privacy for
Language Model Fine-Tuning
Lynn Chua
Google Research
chualynn@google.com
Badih Ghazi
Google Research
badihghazi@gmail.com
Yangsibo Huang
Princeton University
yangsibo@google.com
Pritish Kamath
Google Research
pritishk@google.com
Ravi Kumar
Google Research
ravi.... | 4554241db36e5a0238127fe537a66c2ab06b2a4ecb8b59207d58eaf96bb6b87b | bcba0434b701405bb516c84738e313e2fdf73ac6536560171200dca6c7182909 | positioned-clean-v2 | 8 | {"removed_words": 218, "retained_words": 10901, "rules": {"dehyphenated": 3, "joined_hyphenated_wrap": 6, "joined_small_caps": 31, "mapped_symbol_piece": 5, "page_number": 22, "repeated_review_header": 176, "unmapped_glyph": 20}, "source_words": 11119} |
K1M3gLW0MX | K1M3gLW0MX | On Fairness of Low-Rank Adaptation of Large Models | colm | 2,024 | 273bfd706ae3a661f1c9769b65d7e92421dc4ab3585164c789c40902a4638dbb | On Fairness of Low-Rank Adaptation of Large Models
Zhoujie Ding ∗ Ken Ziyu Liu ∗ Pura Peetathawatchai
Stanford University
{d1ng,kzliu}@cs.stanford.edu
Berivan Isik
Sanmi Koyejo
Abstract
Low-rank adaptation of large models, particularly LoRA, has gained trac-
tion due to its computational efficiency. This efficiency, co... | d419a72c07c90a8ccab530776f4d5c21e59622e01d0c6bb7f2709cb986f16b87 | 2a5cc0f9b776230cca141b4bcb8066ce43b81949f76472753b73e1e9baeb923f | positioned-clean-v2 | 19 | {"removed_words": 352, "retained_words": 21768, "rules": {"composed_negation": 3, "dehyphenated": 12, "joined_hyphenated_wrap": 6, "page_number": 40, "repeated_review_header": 312}, "source_words": 22120} |
KZd1EErRJ1 | KZd1EErRJ1 | IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations | colm | 2,024 | 874373b8113435105ddf5395a6a00e277c9451f11b2231f691ef54dfe113a3b8 | IsoBench: Benchmarking Multimodal Foundation Models on
Isomorphic Representations
Deqing Fu ψ ∗ Ruohao Guo δ ∗ Ghazal Khalighinejad λ ∗ Ollie Liu ψ ∗ .
Bhuwan Dhingra λ Dani Yogatama ψ
Robin Jia ψ Willie Neiswanger ψ
ψ USC
δ Georgia Tech
λ Duke
Abstract
Current foundation models exhibit impressive capabilities when pro... | 48bbe00b537369f091c2bb9296d1995c1666d6de231a2a0c4d73429b46906309 | ef488039e2b4f1fc498bf149207f9559b41ec427d9c0fb3c3f3b5e681ccaaa52 | positioned-clean-v2 | 9 | {"removed_words": 274, "retained_words": 13818, "rules": {"dehyphenated": 11, "joined_hyphenated_wrap": 6, "mapped_symbol_piece": 9, "page_number": 30, "repeated_review_header": 240, "unmapped_glyph": 4}, "source_words": 14092} |
KidynPuLNW | KidynPuLNW | On Limitations of the Transformer Architecture | colm | 2,024 | 71bb7419b92a0ba42586fe1f11df3ee84743fb9d6c9d58370de6646a6c7e5107 | On Limitations of the Transformer Architecture
Binghui Peng
Department of Computer Science
Columbia University
bp2601@columbia.edu
Srini Narayanan
Google DeepMind
srinin@google.com
Christos Papadimitriou
Department of Computer Science
Columbia University
christos@columbia.edu
Abstract
What are the root causes of halluc... | 25c1e87857028059adc8fee65541ef499399720e4c75057bde251dd59d720f76 | b3e0e5815f3d122ccb2812b1f32e58ed46513794d405e1ef3a580a7a57fa45dd | positioned-clean-v2 | 7 | {"removed_words": 141, "retained_words": 9225, "rules": {"combined_negation": 5, "composed_negation": 5, "dehyphenated": 9, "page_number": 15, "repeated_review_header": 120, "unmapped_glyph": 1}, "source_words": 9366} |
KqK5XcgEhR | KqK5XcgEhR | Empowering Large Language Model Agents through Action Learning | colm | 2,024 | ce54bf4d9ddfcdd5ce0246a8ad0f892845c8158b8ff88053664f071762cf9814 | Empowering Large Language Model Agents through Action
Learning
Haiteng Zhao 1 Chang Ma 3 Guoyin Wang 2 Jing Su 2 Lingpeng Kong 3 Jingjing Xu 2 Zhi-Hong Deng 1 Hongxia Yang 2 1 Peking University 2 ByteDance 3 The University of Hong Kong
{zhaohaiteng, zhdeng}@pku.edu.cn
{cma, lpk}@cs.hku.hk
guoyinwang.duke@gmail.com {suj... | 8556b0db0d74979cdba3cf9541cf67cead93865f72a7e3bcbeb59f2aafaadb59 | b232823fc4ae2015d08bf2e26f2deeb3979b694e701bb4bda604c0d2c7b796cd | positioned-clean-v2 | 7 | {"removed_words": 180, "retained_words": 10633, "rules": {"dehyphenated": 4, "expanded_ligature": 5, "page_number": 20, "repeated_review_header": 160}, "source_words": 10813} |
LFfktMPAci | LFfktMPAci | What makes a good metric? Evaluating automatic metrics for text-to-image consistency | colm | 2,024 | a6de4b3bc4c342674a4a23ac6cde460c273707e0b58586608c51dc1f06109fcf | What makes a good metric?
Evaluating automatic metrics for text-to-image consistency
Candace Ross, Melissa Hall, Adriana Romero Soriano, Adina Williams Meta AI (FAIR Labs)
{ccross,melissahall,adrianars,adinawilliams}@meta.com
Abstract
Language models are increasingly being incorporated as components in larger AI system... | 88b4ce44e4086c057042d75a26a1b161b684cea56a5897b3e22fb8fcda06c788 | 9530e0c3960943ec8cf5a2851ac4c2568d5479a0f2c5ad89b2f5e0abbdf5621c | positioned-clean-v2 | 13 | {"removed_words": 171, "retained_words": 9118, "rules": {"dehyphenated": 12, "joined_hyphenated_wrap": 2, "page_number": 19, "repeated_review_header": 152}, "source_words": 9289} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.