forum id stringlengths 10 10 | title stringlengths 31 125 | scores listlengths 3 6 | text stringlengths 52.4k 300k |
|---|---|---|---|
WWXjMYZxfH | MA-RLHF: Reinforcement Learning from Human Feedback with Macro Actions | [
6,
8,
8,
6,
3
] | Published as a conference paper at ICLR 2025
MA-RLHF: REINFORCEMENT LEARNING FROM HU-
MAN FEEDBACK WITH MACRO ACTIONS
Yekun Chai∗ Haoran Sun∗ Huang Fang Shuohuan Wang Yu Sun Hua Wu
Baidu Inc.
{chaiyekun,fanghuang,wangshuohuan}@baidu.com
sunhaoran0402@gmail.com
ABSTRACT
Reinforcement learning from human feedback (RL... |
iZeQBqJamf | Language models scale reliably with over-training and on downstream tasks | [
6,
6,
6,
8
] | Published as a conference paper at ICLR 2025
LANGUAGE MODELS SCALE RELIABLY WITH OVER-
TRAINING AND ON DOWNSTREAM TASKS
Samir Yitzhak Gadre1,2, Georgios Smyrnis3, Vaishaal Shankar4, Suchin Gururangan5,
Mitchell Wortsman5, Rulin Shao5, Jean Mercat2, Alex Fang5, Jeffrey Li5, Sedrick Keh2, Rui
Xin5, Marianna Nezhurina6,... |
DKkQtRMowq | Improving Data Efficiency via Curating LLM-Driven Rating Systems | [
6,
6,
6,
5
] | Published as a conference paper at ICLR 2025
IMPROVING DATA EFFICIENCY VIA CURATING
LLM-DRIVEN RATING SYSTEMS
Jiaheng Wei† 4 Ankit Parag Shah2 Zhaowei Zhu3 Yaxuan Wang1
Jinlong Pang∗ 1
Chen Qian1 Yang Liu1 Yujia Bao2 Wei Wei2
1University of California, Santa Cruz
3BIAI, ZJUT & D5Data.ai 4The Hong Kong University of ... |
p6ncr0eTKE | Task-Adaptive Pretrained Language Models via Clustered-Importance Sampling | [
6,
8,
6,
6
] | Published as a conference paper at ICLR 2025
TASK-ADAPTIVE PRETRAINED LANGUAGE MODELS
VIA CLUSTERED IMPORTANCE SAMPLING
David Grangier, Simin Fan, Skyler Seto, Pierre Ablin
Apple
ABSTRACT
Specialist language models (LMs) focus on a specific task or domain on which
they often outperform generalist LMs of the same si... |
qIN5VDdEOr | Do LLMs ``know'' internally when they follow instructions? | [
6,
8,
6,
5,
5
] | Published as a conference paper at ICLR 2025
DO LLMS “KNOW” INTERNALLY WHEN THEY FOLLOW
INSTRUCTIONS?
Juyeon Heo1,* Christina Heinze-Deml2 Oussama Elachqar2 Kwan Ho Ryan Chan3,* Shirley Ren2
Udhay Nallasamy2 Andy Miller2 Jaya Narain2
1University of Cambridge
jh2324@cam.ac.uk jnarain@apple.com
3University of Pennsylv... |
fL4qWkSmtM | What is Wrong with Perplexity for Long-context Language Modeling? | [
8,
6,
6,
6,
8
] | Published as a conference paper at ICLR 2025
WHAT IS WRONG WITH PERPLEXITY FOR LONG-
CONTEXT LANGUAGE MODELING?
Lizhe Fang1∗ Yifei Wang2∗ Zhaoyang Liu3 Chenheng Zhang1
Stefanie Jegelka4,5
1 State Key Lab of General Artificial Intelligence,
Jinyang Gao3 Bolin Ding3 Yisen Wang1,6†
School of Intelligence Science and T... |
4NRjdISWby | LoCA: Location-Aware Cosine Adaptation for Parameter-Efficient Fine-Tuning | [
6,
6,
6,
5,
6,
6
] | Published as a conference paper at ICLR 2025
LOCA: LOCATION-AWARE COSINE ADAPTATION FOR
PARAMETER-EFFICIENT FINE-TUNING
Zhekai Du†,‡∗, Yinjie Min⋄, Jingjing Li†(cid:66), Ke Lu†, Changliang Zou⋄, Liuhua Peng‡
Tingjin Chu‡, Mingming Gong‡,⋆
† University of Electronic Science and Technology of China ‡ The University of ... |
l2zFn6TIQi | Controlling Language and Diffusion Models by Transporting Activations | [
8,
6,
8,
8
] | Published as a conference paper at ICLR 2025
CONTROLLING LANGUAGE AND DIFFUSION MODELS
BY TRANSPORTING ACTIVATIONS
Pau Rodr´ıguez∗ Arno Blaas Michal Klein Luca Zappella Nicholas Apostoloff
Marco Cuturi Xavier Suau∗
pau.rodriguez,ablaas,michal klein,lzappella,napostoloff,
{
m cuturi,xsuaucuadros
Apple
@apple.com
}
A... |
Yk87CwhBDx | Can Large Language Models Understand Symbolic Graphics Programs? | [
6,
8,
8
] | Published as a conference paper at ICLR 2025
CAN LARGE LANGUAGE MODELS UNDERSTAND
SYMBOLIC GRAPHICS PROGRAMS?
Zeju Qiu1,† Weiyang Liu1,2,†,* Haiwen Feng1,† Zhen Liu1,‡ Tim Z. Xiao1,‡ Katherine M. Collins2,‡
Joshua B. Tenenbaum3 Adrian Weller2 Michael J. Black1 Bernhard Schölkopf1
1Max Planck Institute for Intelligent... |
1EnpStvBU8 | Feast Your Eyes: Mixture-of-Resolution Adaptation for Multimodal Large Language Models | [
6,
6,
5,
6,
6
] | Published as a conference paper at ICLR 2025
FEAST YOUR EYES: MIXTURE-OF-RESOLUTION
ADAPTATION FOR MULTIMODAL LARGE
LANGUAGE MODELS
Gen Luo1,2, Yiyi Zhou1, Yuxin Zhang1, Xiawu Zheng1, Xiaoshuai Sun1, Rongrong Ji1(cid:0)
1Key Laboratory of Multimedia Trusted Perception and Efficient Computing,
Ministry of Education of... |
rQ7fz9NO7f | Multimodal Large Language Models for Inverse Molecular Design with Retrosynthetic Planning | [
8,
6,
6,
8
] | Published as a conference paper at ICLR 2025
MULTIMODAL LARGE LANGUAGE MODELS FOR IN-
VERSE MOLECULAR DESIGN WITH RETROSYNTHETIC
PLANNING
Gang Liu1∗, Michael Sun2∗, Wojciech Matusik2, Meng Jiang1,
1University of Notre Dame
2MIT CSAIL
{gliu7, mjiang2}@nd.edu,
chenjie@us.ibm.com
{msun415, wojciech}@csail.mit.edu,
3 M... |
SnDmPkOJ0T | REEF: Representation Encoding Fingerprints for Large Language Models | [
10,
8,
6,
8
] | Published as a conference paper at ICLR 2025
REEF: REPRESENTATION ENCODING FINGERPRINTS
FOR LARGE LANGUAGE MODELS
Jie Zhang1,2⋆, Dongrui Liu1⋆, Chen Qian1,3, Linfeng Zhang4, Yong Liu3, Yu Qiao1, Jing Shao1†
1 Shanghai Artificial Intelligence Laboratory 2 University of Chinese Academy of Sciences
3 Renmin University o... |
hXm0Wu2U9K | Correcting the Mythos of KL-Regularization: Direct Alignment without Overoptimization via Chi-Squared Preference Optimization | [
6,
8,
6,
6,
6
] | Published as a conference paper at ICLR 2025
CORRECTING THE MYTHOS OF KL-REGULARIZATION:
DIRECT ALIGNMENT WITHOUT OVEROPTIMIZATION
VIA χ2-PREFERENCE OPTIMIZATION
Audrey Huang* Wenhao Zhan† Tengyang Xie‡
Wen Sun§ Akshay Krishnamurthy⋄ Dylan J. Foster⋄
Jason D. Lee†
*University of Illinois Urbana-Champaign
‡Universit... |
uZFXpPrwSh | Zero-shot Model-based Reinforcement Learning using Large Language Models | [
5,
8,
6,
8
] | Published as a conference paper at ICLR 2025
ZERO-SHOT MODEL-BASED REINFORCEMENT LEARN-
ING USING LARGE LANGUAGE MODELS
Abdelhakim Benechehab†12, Youssef Attia El Hili1, Ambroise Odonnat13, Oussama Zekri‡4,
Albert Thomas1, Giuseppe Paolo1, Maurizio Filippone5, Ievgen Redko1, Bal´azs K´egl1
1 Huawei Noah’s Ark Lab, Pa... |
gp32jvUquq | Basis Sharing: Cross-Layer Parameter Sharing for Large Language Model Compression | [
8,
5,
8,
5
] | Published as a conference paper at ICLR 2025
BASIS SHARING: CROSS-LAYER PARAMETER SHARING
FOR LARGE LANGUAGE MODEL COMPRESSION
Jingcun Wang
Technical University of Darmstadt
jingcun.wang@tu-darmstadt.de
Yu-Guang Chen
National Central University
andyygchen@ee.ncu.edu.tw
Ing-Chao Lin
National Cheng Kung University
ic... |
wUtCieKuQU | Towards Effective Evaluations and Comparisons for LLM Unlearning Methods | [
3,
8,
5,
6
] | Published as a conference paper at ICLR 2025
TOWARDS EFFECTIVE EVALUATIONS AND COMPAR-
ISONS FOR LLM UNLEARNING METHODS
Qizhou Wang1∗ Bo Han1,2† Puning Yang1
Tongliang Liu3 Masashi Sugiyama2,4
Jianing Zhu1
1TMLR Group, Department of Computer Science, Hong Kong Baptist University
2RIKEN Center for Advanced Intellige... |
RzUvkI3p1D | Concept-ROT: Poisoning Concepts in Large Language Models with Model Editing | [
6,
5,
6,
6
] | Published as a conference paper at ICLR 2025
CONCEPT-ROT: POISONING CONCEPTS IN LARGE
LANGUAGE MODELS WITH MODEL EDITING
Keltin Grimes, Marco Christiani, David Shriver & Marissa Connor
Software Engineering Institute
Carnegie Mellon University
Pittsburgh, PA 15213, USA
{kgrimes,mchristiani,dlshriver,mconnor}@sei.cmu.e... |
acxHV6werE | VibeCheck: Discover and Quantify Qualitative Differences in Large Language Models | [
3,
6,
6,
6
] | Published as a conference paper at ICLR 2025
VIBECHECK: DISCOVER & QUANTIFY QUALITATIVE
DIFFERENCES IN LARGE LANGUAGE MODELS
Lisa Dunlap
UC Berkeley
Krishna Mandal
UC Berkeley
Trevor Darrell
UC Berkeley
Jacob Steinhardt
UC Berkeley
Joseph Gonzalez
UC Berkeley
ABSTRACT
Large language models (LLMs) often exhibit ... |
HfWcFs7XLR | Agents' Room: Narrative Generation through Multi-step Collaboration | [
6,
5,
8
] | Published as a conference paper at ICLR 2025
AGENTS’ ROOM: NARRATIVE GENERATION THROUGH
MULTI-STEP COLLABORATION
Fantine Huot, Reinald Kim Amplayo, Jennimaria Palomaki, Alice Shoshana Jakobovits,
Elizabeth Clark & Mirella Lapata
Google DeepMind
{fantinehuot,reinald,jpalomaki,jakobovits,eaclark,lapata}@google.com
ABS... |
W9FZEQj3vv | Variational Best-of-N Alignment | [
6,
6,
3,
8
] | Published as a conference paper at ICLR 2025
VARIATIONAL BEST-OF-N ALIGNMENT
Afra Amini Tim Vieira Elliott Ash Ryan Cotterell
ETH Z¨urich
{afra.amini, ryan.cotterell}@inf.ethz.ch
tim.f.vieira@gmail.com ashe@ethz.ch
ABSTRACT
Best-of-N (BoN ) is a popular and effective algorithm for aligning language models
to human ... |
HPSAkIHRbb | BingoGuard: LLM Content Moderation Tools with Risk Levels | [
6,
8,
6,
6
] | Published as a conference paper at ICLR 2025
BINGOGUARD: LLM CONTENT MODERATION TOOLS
WITH RISK LEVELS
Fan Yin1 ∗
Philippe Laban3 †
Xiangyu Peng2
Yilun Zhou2
Yixin Mao2
Vaibhav Vats2
Linnea Ross2
Divyansh Agarwal2
Caiming Xiong2
Chien-Sheng Wu2
1University of California, Los Angeles, 2Salesforce, 3Microsoft... |
vkakKdznFS | Text4Seg: Reimagining Image Segmentation as Text Generation | [
6,
5,
8
] | Published as a conference paper at ICLR 2025
TEXT4SEG: REIMAGINING IMAGE SEGMENTATION AS
TEXT GENERATION
Mengcheng Lan, Chaofeng Chen, Yue Zhou
S-Lab, Nanyang Technological University
lanm0002@e.ntu.edu.sg
{chaofeng.chen,yue.zhou}@ntu.edu.sg
Jiaxing Xu, Yiping Ke
CCDS, Nanyang Technological University
jiaxing003@e.n... |
cqsw28DuMW | TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models | [
8,
8,
6
] | Published as a conference paper at ICLR 2025
TAID: TEMPORALLY ADAPTIVE INTERPOLATED DIS-
TILLATION FOR EFFICIENT KNOWLEDGE TRANSFER IN
LANGUAGE MODELS
Makoto Shing1, Kou Misaki1, Han Bao2, Sho Yokoi345, Takuya Akiba1
1Sakana AI, 2Kyoto University, 3NINJAL, 4Tohoku University, 5RIKEN
{mkshing,kou.misaki,takiba}@sakana... |
J1J5eGJsKZ | ToolDial: Multi-turn Dialogue Generation Method for Tool-Augmented Language Models | [
8,
6,
6
] | Published as a conference paper at ICLR 2025
TOOLDIAL: MULTI-TURN DIALOGUE GENERATION
METHOD FOR TOOL-AUGMENTED LANGUAGE MODELS
Jeonghoon Shim1, Gyuhyeon Seo1, Cheongsu Lim2, Yohan Jo1∗
1Graduate School of Data Science, Seoul National University
2Department of Industrial and Management Engineering, Korea University
j... |
7igPXQFupX | CoTFormer: A Chain of Thought Driven Architecture with Budget-Adaptive Computation Cost at Inference | [
6,
5,
6,
6
] | Published as a conference paper at ICLR 2025
COTFORMER: A CHAIN-OF-THOUGHT DRIVEN AR-
CHITECTURE WITH BUDGET-ADAPTIVE COMPUTATION
COST AT INFERENCE
Amirkeivan Mohtashami∗
EPFL
Matteo Pagliardini∗
EPFL
Martin Jaggi
EPFL
ABSTRACT
Scaling language models to larger and deeper sizes has led to significant boosts
in pe... |
m4eXBo0VNc | An Engorgio Prompt Makes Large Language Model Babble on | [
6,
8,
6,
3
] | Published as a conference paper at ICLR 2025
AN ENGORGIO PROMPT MAKES LARGE LANGUAGE
MODEL BABBLE ON
Jianshuo Dong1, Ziyuan Zhang1, Qingjie Zhang1, Tianwei Zhang2,
Hao Wang1, Hewu Li1, Qi Li1, Chao Zhang1, Ke Xu1, and Han Qiu1∗
1Tsinghua University,
2Nanyang Technological University
dongjs23@mails.tsinghua.edu.cn,... |
TqYjhJrp9m | Zero-shot forecasting of chaotic systems | [
6,
6,
8
] | Published as a conference paper at ICLR 2025
ZERO-SHOT FORECASTING OF CHAOTIC SYSTEMS
Yuanzhao Zhang
Santa Fe Institute
Santa Fe, NM, USA
William Gilpin∗
Department of Physics
University of Texas at Austin
Austin, TX, USA
ABSTRACT
Time-series forecasting is a challenging problem that traditionally requires spe-
ci... |
mNVR9jJYqK | DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing | [
6,
6,
8,
5
] | Published as a conference paper at ICLR 2025
DRESSING UP LLM: EFFICIENT STYLIZED QUESTION-
ANSWERING VIA STYLE SUBSPACE EDITING
Xinyu Ma1, Yifeng Xu1, Yang Lin1, Tianlong Wang3, Xu Chu1,2,3, Xin Gao1,
Junfeng Zhao1, Yasha Wang1,3˚
1 School of Computer Science, Peking University
2 Center on Frontiers of Computing Stud... |
eLLBILFRsA | UniDetox: Universal Detoxification of Large Language Models via Dataset Distillation | [
6,
5,
6,
8
] | Published as a conference paper at ICLR 2025
UNIDETOX: UNIVERSAL DETOXIFICATION OF LARGE
LANGUAGE MODELS VIA DATASET DISTILLATION
Huimin Lu 1 ∗ Masaru Isonuma 1,2,3
1The University of Tokyo
2The University of Edinburgh
Junichiro Mori 1,4
Ichiro Sakata 1
4RIKEN AIP
3NII
ABSTRACT
We present UNIDETOX, a universall... |
AV7OXVlAyi | Mitigating Modality Prior-Induced Hallucinations in Multimodal Large Language Models via Deciphering Attention Causality | [
5,
5,
6,
6
] | Published as a conference paper at ICLR 2025
MITIGATING MODALITY PRIOR-INDUCED HALLUCI-
NATIONS IN MULTIMODAL LARGE LANGUAGE MOD-
ELS VIA DECIPHERING ATTENTION CAUSALITY
Guanyu Zhou1 Yibo Yan1,2 Xin Zou1 Kun Wang3 Aiwei Liu1,4 Xuming Hu1,2,∗
1The Hong Kong University of Science and Technology (Guangzhou)
2The Hong K... |
SFN6Wm7YBI | TorchTitan: One-stop PyTorch native solution for production ready LLM pretraining | [
5,
6,
6,
6,
6,
10
] | Published as a conference paper at ICLR 2025
TORCHTITAN: ONE-STOP PYTORCH NATIVE SOLU-
TION FOR PRODUCTION READY LLM PRETRAINING
Wanchao Liang1, Tianyu Liu1∗, Less Wright1, Will Constable1, Andrew Gu1
Chien-Chin Huang1, Iris Zhang1, Wei Feng1, Howard Huang1, Junjie Wang1
Sanket Purandare2†, Gokul Nadathur1, Stratos I... |
wHLMsM1SrP | Needle Threading: Can LLMs Follow Threads Through Near-Million-Scale Haystacks? | [
6,
5,
6,
8
] | Published as a conference paper at ICLR 2025
NEEDLE THREADING: CAN LLMS FOLLOW THREADS
THROUGH NEAR-MILLION-SCALE HAYSTACKS?
Jonathan Roberts♦
♦University of Cambridge
https://needle-threading.github.io/
Kai Han♠
Samuel Albanie
♠The University of Hong Kong
ABSTRACT
As the context limits of Large Language Models ... |
yR47RmND1m | Understanding and Enhancing Safety Mechanisms of LLMs via Safety-Specific Neuron | [
6,
8,
6,
8,
3
] | Published as a conference paper at ICLR 2025
UNDERSTANDING AND ENHANCING SAFETY MECHA-
NISMS OF LLMS VIA SAFETY-SPECIFIC NEURON
Yiran Zhao1† Wenxuan Zhang2 Yuxi Xie1 Anirudh Goyal3
Kenji Kawaguchi1 Michael Qizhe Shieh1†
1 National University of Singapore
3 Google DeepMind
2 Singapore University of Technology and Des... |
l32IrJtpOP | Enhancing Graph Of Thought: Enhancing Prompts with LLM Rationales and Dynamic Temperature Control | [
6,
6,
5,
8
] | Published as a conference paper at ICLR 2025
ENHANCING GRAPH OF THOUGHT:
ENHANCING PROMPTS WITH LLM RATIONALES AND
DYNAMIC TEMPERATURE CONTROL
Sunguk Shin and Youngjoon Kim∗
Korea University
Seoul, Republic of Korea
{ssw1419, acorn421}@korea.ac.kr
ABSTRACT
We introduce Enhancing Graph of Thoughts (EGoT), a method d... |
60Vd7QOXlM | Privacy Auditing of Large Language Models | [
6,
6,
6,
8
] | Published as a conference paper at ICLR 2025
PRIVACY AUDITING OF LARGE LANGUAGE MODELS
Ashwinee Pandap∗ Xinyu Tangp∗ Milad Nasrg Christopher A. Choquette-Choog Prateek Mittalp
pPrinceton University, gGoogle DeepMind, ∗Equal contribution
ABSTRACT
Current techniques for privacy auditing of large language models (LLMs... |
3GTtZFiajM | Justice or Prejudice? Quantifying Biases in LLM-as-a-Judge | [
6,
5,
8,
8
] | Published as a conference paper at ICLR 2025
JUSTICE OR PREJUDICE?
QUANTIFYING BIASES IN LLM-AS-A-JUDGE
Jiayi Ye♢, ∗, Yanbo Wang△, ∗, Yue Huang♠, ∗, Dongping Chen♣, Qihui Zhang♡
Nuno Moniz♠, Tian Gao⋆, Werner Geyer⋆, Chao Huang▲, Pin-Yu Chen⋆, Nitesh V. Chawla♠
Xiangliang Zhang♠, †
♠University of Notre Dame △MBZUAI ♣... |
fq24pEb8SL | Weighted-Reward Preference Optimization for Implicit Model Fusion | [
6,
6,
6
] | Published as a conference paper at ICLR 2025
WEIGHTED-REWARD PREFERENCE OPTIMIZATION FOR
IMPLICIT MODEL FUSION
Ziyi Yang∗ Fanqi Wan∗ Longguang Zhong Tianyuan Shi Xiaojun Quan†
School of Computer Science and Engineering, Sun Yat-sen University, China
yangzy39@mail2.sysu.edu.cn, quanxj3@mail.sysu.edu.cn
ABSTRACT
Whil... |
oWdzUpOlkX | AgentOccam: A Simple Yet Strong Baseline for LLM-Based Web Agents | [
8,
5,
6
] | Published as a conference paper at ICLR 2025
AGENTOCCAM: A SIMPLE YET STRONG BASELINE
FOR LLM-BASED WEB AGENTS
Ke Yang†∗, Yao Liu♢, Sapana Chaudhary♢, Rasool Fakoor♢, Pratik Chaudhari♢, George
Karypis♢, Huzefa Rangwala♢
University of Illinois Urbana-Champaign†, Amazon♢
key4@illinois.edu, {yaoliuai,chausapa,fakoor,rhu... |
4FWAwZtd2n | Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Parameters for Reasoning | [
8,
6,
8,
8
] | Published as a conference paper at ICLR 2025
SCALING LLM TEST-TIME COMPUTE OPTIMALLY CAN BE MORE
EFFECTIVE THAN SCALING PARAMETERS FOR REASONING
Charlie Snell*, Jaehoon Lee§, Kelvin Xu§†, Aviral Kumar#§†
ABSTRACT
Enabling LLMs to improve their outputs by using more test-time compute is a crit-
ical step towards bui... |
MbX0t1rUlp | MLPs Learn In-Context on Regression and Classification Tasks | [
6,
8,
8,
6,
3
] | Published as a conference paper at ICLR 2025
MLPS LEARN IN-CONTEXT ON
REGRESSION AND CLASSIFICATION TASKS
William L. Tong & Cengiz Pehlevan
School of Engineering and Applied Sciences
Center for Brain Sciences
Kempner Institute for the Study of Artificial and Natural Intelligence
Harvard University, Cambridge, MA 0213... |
GGlpykXDCa | MMQA: Evaluating LLMs with Multi-Table Multi-Hop Complex Questions | [
8,
8,
8
] | Published as a conference paper at ICLR 2025
MMQA: EVALUATING LLMS WITH MULTI-TABLE
MULTI-HOP COMPLEX QUESTIONS
Jian Wu1∗ Linyi Yang2∗ Dongyuan Li4∗ Yuliang Ji5 Manabu Okumura1 Yue Zhang3†
1Institute of Science Tokyo 2University College London 3School of Engineering, Westlake Univeristy
4The University of Tokyo 5Nan... |
J5sUOvlLbQ | LiNeS: Post-training Layer Scaling Prevents Forgetting and Enhances Model Merging | [
6,
6,
6,
5
] | Published as a conference paper at ICLR 2025
LINES: POST-TRAINING LAYER SCALING PREVENTS
FORGETTING AND ENHANCES MODEL MERGING
Ke Wang∗
EPFL
k.wang@epfl.ch
Nikolaos Dimitriadis∗
EPFL
nikolaos.dimitriadis@epfl.ch
Alessandro Favero
EPFL
alessandro.favero@epfl.ch
Guillermo Ortiz-Jimenez
Google DeepMind
gortizj@google... |
gI0kPklUKS | Bilinear MLPs enable weight-based mechanistic interpretability | [
8,
8,
8,
6
] | Published as a conference paper at ICLR 2025
BILINEAR MLPS ENABLE WEIGHT-BASED
MECHANISTIC INTERPRETABILITY
Michael T. Pearce∗
Independent
pearcemt@
alumni.stanford.edu
Thomas Dooms∗
University of Antwerp
thomas.dooms@
uantwerpen.be
Alice Rigg
Independent
rigg.alice0@
gmail.com
Jose Oramas
University of Antwerp, s... |
whaO3482bs | ChroKnowledge: Unveiling Chronological Knowledge of Language Models in Multiple Domains | [
6,
6,
6,
6
] | Published as a conference paper at ICLR 2025
CHROKNOWLEDGE: UNVEILING CHRONOLOGICAL
KNOWLEDGE OF LANGUAGE MODELS IN MULTIPLE
DOMAINS
Yein Park1, Chanwoong Yoon1, Jungwoo Park1,3,
Donghyeon Lee1,3, Minbyul Jeong2∗, Jaewoo Kang1,3∗
Korea University1 Upstage AI2 AIGEN Sciences3
{522yein, cwyoon99, jungwoo-park, dong9733... |
BWS5gVjgeY | Number Cookbook: Number Understanding of Language Models and How to Improve It | [
6,
6,
8,
6
] | Published as a conference paper at ICLR 2025
NUMBER COOKBOOK: NUMBER UNDERSTANDING OF
LANGUAGE MODELS AND HOW TO IMPROVE IT
Haotong Yang123 Yi Hu12 Shijia Kang12 Zhouchen Lin1234∗ Muhan Zhang23∗
1 School of Intelligence Science and Technology, Peking University
2 Institution for Artificial Intelligence, Peking Univer... |
41uZB8bDFh | Durable Quantization Conditioned Misalignment Attack on Large Language Models | [
6,
6,
6
] | Published as a conference paper at ICLR 2025
DURABLE QUANTIZATION CONDITIONED MISALIGN-
MENT ATTACK ON LARGE LANGUAGE MODELS
Peiran Dong∗
Department of Computing
Hong Kong Polytechnic University
peiran.dong@connect.polyu.hk
Haowei Li∗
School of Cyber Science and Engineering
Wuhan University
haowei.li@whu.edu.cn
Son... |
HN8V0flwJF | World Model on Million-Length Video And Language With Blockwise RingAttention | [
8,
3,
6,
6
] | Published as a conference paper at ICLR 2025
WORLD MODEL ON MILLION-LENGTH VIDEO AND
LANGUAGE WITH BLOCKWISE RINGATTENTION
Hao Liu∗ Wilson Yan∗ Matei Zaharia Pieter Abbeel
UC Berkeley
ABSTRACT
Enabling long-context understanding remains a key challenge in scaling existing
sequence models – a crucial component in d... |
IDxZhXrpNf | SOAP: Improving and Stabilizing Shampoo using Adam for Language Modeling | [
5,
6,
6,
8
] | Published as a conference paper at ICLR 2025
SOAP: IMPROVING AND STABILIZING SHAMPOO US-
ING ADAM FOR LANGUAGE MODELING
Nikhil Vyas∗
Harvard University
Depen Morwani∗
Harvard University
Rosie Zhao†
Harvard University
Itai Shapira†
Harvard University
David Brandfonbrener
Kempner Institute at Harvard University
Sh... |
IjduZQK8gM | From Attention to Activation: Unraveling the Enigmas of Large Language Models | [
5,
6,
6
] | Preprint
FROM ATTENTION TO ACTIVATION: UNRAVELING THE
ENIGMAS OF LARGE LANGUAGE MODELS
∗ Chengcheng Ma2
Prannay Kaul1
1Huawei Noah’s Ark Lab, London, UK
2Institute of Automation, Chinese Academy of Sciences (CASIA)
Ismail Elezi1
† Jiankang Deng1
CURRENT TRANSFORMER MODELS
OUR TRANSFORMER MODELS
(a)
(b)
Figure... |
uZ5K4HeNwd | Beyond Autoregression: Fast LLMs via Self-Distillation Through Time | [
6,
8,
8,
6
] | Published as a conference paper at ICLR 2025
BEYOND AUTOREGRESSION:
FAST LLMS VIA SELF-DISTILLATION THROUGH TIME
Justin Deschenaux, Caglar Gulcehre
School of Computer and Communication Sciences
CLAIRE, EPFL
Lausanne, Switzerland
{justin.deschenaux, caglar.gulcehre}@epfl.ch
ABSTRACT
Autoregressive (AR) Large Languag... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.