ICLR 2025 论文和开源项目合集
本仓库旨在收集ICLR最新研究进展，尤其是LLM方面，涉及NLP领域的各个方向，此项目长期不定时更新。
欢迎watch和fork！不过给个star⭐就更好了❤️。
知乎地址：ShuYini
微信公众号: AINLPer（每日更新，欢迎关注）
另外也欢迎大家进入AINLPer星球，每天推送最新、最优质论文，紧跟AIGC大模型前沿进展；另外星球也特设大模型Agent、大模型推理、RAG系统搭建、热门综述、大模型实操、数据集、测试基准、行业发展状况、大厂工作内推等专栏。详细介绍→：https://mp.weixin.qq.com/s/wHnm9ek4ojYTA_2EPLNILw
感兴趣的小伙伴，赶快扫描下方⬇二维码，新用户加入更有50元优惠券🔖，每年仅需49！
💎💎💎💎💎💎💎💎💎💎💎💎💎💎
历年ICLR 论文和开源项目合集
Accept Spotlight
1、How new data permeates LLM knowledge and how to dilute it
2、JudgeLM: Fine-tuned Large Language Models are Scalable Judges
3、MQuAKE-Remastered: Multi-Hop Knowledge Editing Can Only Be Advanced with Reliable Evaluations
4、Physics of Language Models: Part 3.3, Knowledge Capacity Scaling Laws
5、Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence
6、Robustness Reprogramming for Representation Learning
7、LayerDAG: A Layerwise Autoregressive Diffusion Model for Directed Acyclic Graph Generation
8、DailyDilemmas: Revealing Value Preferences of LLMs with Quandaries of Daily Life
9、Learning-Augmented Frequent Directions
10、Quality Measures for Dynamic Graph Generative Models
11、Holistically Evaluating the Environmental Impact of Creating Language Models
12、Mixture-of-Agents Enhances Large Language Model Capabilities
13、Answer, Assemble, Ace: Understanding How LMs Answer Multiple Choice Questions
14、TopoNets: High performing vision and language models with brain-like topography
15、INCLUDE: Evaluating Multilingual Language Understanding with Regional Knowledge
16、Vision Language Models are In-Context Value Learners
17、Wasserstein Distances, Neuronal Entanglement, and Sparsity
18、WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild
19、Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI
20、Multi-session, multi-task neural decoding from distinct cell-types and brain regions
21、Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering
22、AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs
23、BRIGHT: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval
24、Online Reinforcement Learning in Non-Stationary Context-Driven Environments
25、CEB: Compositional Evaluation Benchmark for Fairness in Large Language Models
26、TabWak: A Watermark for Tabular Diffusion Models
27、Generating Freeform Endoskeletal Robots
28、AnalogGenie: A Generative Engine for Automatic Discovery of Analog Circuit Topologies
29、DataEnvGym: Data Generation Agents in Teacher Environments with Student Feedback
30、LoRA3D: Low-Rank Self-Calibration of 3D Geometric Foundation models
31、Exact Certification of (Graph) Neural Networks Against Label Poisoning
32、Test-time Adaptation for Cross-modal Retrieval with Query Shift
33、In Search of Forgotten Domain Generalization
34、TOP-ERL: Transformer-based Off-Policy Episodic Reinforcement Learning
35、Harnessing Diversity for Important Data Selection in Pretraining Large Language Models
36、CausalRivers - Scaling up benchmarking of causal discovery for real-world time-series
37、Provably Reliable Conformal Prediction Sets in the Presence of Data Poisoning
38、Progressive Compositionality in Text-to-Image Generative Models
39、Benchmarking Predictive Coding Networks – Made Simple
40、Can Watermarked LLMs be Identified by Users via Crafted Prompts?
41、On Quantizing Neural Representation for Variable-Rate Video Coding
42、Attention with Markov: A Curious Case of Single-layer Transformers
43、A Second-Order Perspective on Model Compositionality and Incremental Learning
44、Iterative Label Refinement Matters More than Preference Optimization under Weak Supervision
45、EmbedLLM: Learning Compact Representations of Large Language Models
46、Lean-STaR: Learning to Interleave Thinking and Proving
47、NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models
48、DARE the Extreme: Revisiting Delta-Parameter Pruning For Fine-Tuned Models
49、Representative Guidance: Diffusion Model Sampling with Coherence
50、Budgeted Online Continual Learning by Adaptive Layer Freezing and Frequency-based Sampling
51、InverseBench: Benchmarking Plug-and-Play Diffusion Priors for Inverse Problems in Physical Sciences
52、Revisiting Random Walks for Learning on Graphs
53、DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inference
54、Control-oriented Clustering of Visual Latent Representation
55、AIR-BENCH 2024: A Safety Benchmark based on Regulation and Policies Specified Risk Categories
56、Not All LLM-Generated Data Are Equal: Rethinking Data Weighting in Text Classification
57、X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale
58、Token Statistics Transformer: Linear-Time Attention via Variational Rate Reduction
59、Bilinear MLPs enable weight-based mechanistic interpretability
60、Can Large Language Models Understand Symbolic Graphics Programs?
61、AgentTrek: Agent Trajectory Synthesis via Guiding Replay with Web Tutorials
62、Advantage-Guided Distillation for Preference Alignment in Small Language Models
63、Simplifying Deep Temporal Difference Learning
64、SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding
65、Biologically Constrained Barrel Cortex Model Integrates Whisker Inputs and Replicates Key Brain Network Dynamics
66、Broaden your SCOPE! Efficient Multi-turn Conversation Planning for LLMs with Semantic Space
67、The Superposition of Diffusion Models Using the Itô Density Estimator
68、MAGNet: Motif-Agnostic Generation of Molecules from Scaffolds
69、Interleaved Scene Graphs for Interleaved Text-and-Image Generation Assessment
70、IGL-Bench: Establishing the Comprehensive Benchmark for Imbalanced Graph Learning
71、RegMix: Data Mixture as Regression for Language Model Pre-training
72、When Attention Sink Emerges in Language Models: An Empirical View
73、PianoMotion10M: Dataset and Benchmark for Hand Motion Generation in Piano Performance
74、Revisiting text-to-image evaluation with Gecko: on metrics, prompts, and human rating
75、Streamlining Redundant Layers to Compress Large Language Models
76、SynFlowNet: Design of Diverse and Novel Molecules with Synthesis Constraints
77、Rethinking and Improving Autoformalization: Towards a Faithful Metric and a Dependency Retrieval-based Approach
78、Learning Spatiotemporal Dynamical Systems from Point Process Observations
79、Probabilistic Neural Pruning via Sparsity Evolutionary Fokker-Planck-Kolmogorov Equation
80、Uncovering Gaps in How Humans and LLMs Interpret Subjective Language
81、SPA-BENCH: A COMPREHENSIVE BENCHMARK FOR SMARTPHONE AGENT EVALUATION
82、Demystifying the Token Dynamics of Deep Selective State Space Models
83、Knowledge Localization: Mission Not Accomplished? Enter Query Localization!
84、Realistic Evaluation of Deep Partial-Label Learning Algorithms
85、Graph Sparsification via Mixture of Graphs
86、RAG-SR: Retrieval-Augmented Generation for Neural Symbolic Regression
87、MixEval-X: Any-to-any Evaluations from Real-world Data Mixture
88、DEEM: Diffusion models serve as the eyes of large language models for image perception
89、BodyGen: Advancing Towards Efficient Embodiment Co-Design
90、Diffusion Bridge AutoEncoders for Unsupervised Representation Learning
91、Bi-Factorial Preference Optimization: Balancing Safety-Helpfulness in Language Models
92、Grounding Video Models to Actions through Goal Conditioned Exploration
93、Dense Video Object Captioning from Disjoint Supervision
94、RESuM: A Rare Event Surrogate Model for Physics Detector Design
95、DeLLMa: Decision Making Under Uncertainty with Large Language Models
96、Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation
97、A Periodic Bayesian Flow for Material Generation
98、SplatFormer: Point Transformer for Robust 3D Gaussian Splatting
99、Continuous Exposure Learning for Low-light Image Enhancement using Neural ODEs
100、Lumina-T2X: Scalable Flow-based Large Diffusion Transformer for Flexible Resolution Generation
101、Towards Marginal Fairness Sliced Wasserstein Barycenter
102、Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion
103、Following the Human Thread in Social Navigation
104、OS-ATLAS: Foundation Action Model for Generalist GUI Agents
105、Student-Informed Teacher Training
106、Preference Optimization for Reasoning with Pseudo Feedback
107、OmniRe: Omni Urban Scene Reconstruction
108、Multimodality Helps Few-shot 3D Point Cloud Semantic Segmentation
109、Multi-modal Agent Tuning: Building a VLM-Driven Agent for Efficient Tool Usage
110、Multi-Field Adaptive Retrieval
111、On Disentangled Training for Nonlinear Transform in Learned Image Compression
112、MMAU: A Massive Multi-Task Audio Understanding and Reasoning Benchmark
113、ReDeEP: Detecting Hallucination in Retrieval-Augmented Generation via Mechanistic Interpretability
114、Towards General-Purpose Model-Free Reinforcement Learning
115、Accelerating Goal-Conditioned Reinforcement Learning Algorithms and Research
116、Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts
117、SimBa: Simplicity Bias for Scaling Up Parameters in Deep Reinforcement Learning
118、Self-play with Execution Feedback: Improving Instruction-following Capabilities of Large Language Models
119、Diffusion On Syntax Trees For Program Synthesis
120、Vision-RWKV: Efficient and Scalable Visual Perception with RWKV-Like Architectures
121、A CLIP-Powered Framework for Robust and Generalizable Data Selection
122、Fine-tuning with Reserved Majority for Noise Reduction
123、GOLD: Graph Out-of-Distribution Detection via Implicit Adversarial Latent Generation
124、Topograph: An Efficient Graph-Based Framework for Strictly Topology Preserving Image Segmentation
125、Test-time Alignment of Diffusion Models without Reward Over-optimization
126、ImpScore: A Learnable Metric For Quantifying The Implicitness Level of Sentences
127、LoRA-Pro: Are Low-Rank Adapters Properly Optimized?
128、LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models
129、Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding
130、GETS: Ensemble Temperature Scaling for Calibration in Graph Neural Networks
131、Easing Training Process of Rectified Flow Models Via Lengthening Inter-Path Distance
132、Recovering Manifold Structure Using Ollivier Ricci Curvature
133、AutoCGP: Closed-Loop Concept-Guided Policies from Unlabeled Demonstrations
134、Min-K%++: Improved Baseline for Pre-Training Data Detection from Large Language Models
135、MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion
136、Atlas Gaussians Diffusion for 3D Generation
137、Samba: Synchronized Set-of-Sequences Modeling for Multiple Object Tracking
138、Eagle: Exploring The Design Space for Multimodal LLMs with Mixture of Encoders
139、Programming Refusal with Conditional Activation Steering
140、Perm: A Parametric Representation for Multi-Style 3D Hair Modeling
141、CLoSD: Closing the Loop between Simulation and Diffusion for multi-task character control
142、ND-SDF: Learning Normal Deflection Fields for High-Fidelity Indoor Reconstruction
143、D-FINE: Redefine Regression Task of DETRs as Fine-grained Distribution Refinement
144、Mitigating Information Loss in Tree-Based Reinforcement Learning via Direct Optimization
145、Sharpness-Aware Minimization Efficiently Selects Flatter Minima Late In Training
146、DartControl: A Diffusion-Based Autoregressive Motion Model for Real-Time Text-Driven Motion Control
147、ThinK: Thinner Key Cache by Query-Driven Pruning
148、4K4DGen: Panoramic 4D Generation at 4K Resolution
149、RelitLRM: Generative Relightable Radiance for Large Reconstruction Models
150、ConFIG: Towards Conflict-free Training of Physics Informed Neural Networks
151、Severing Spurious Correlations with Data Pruning
152、MotionAura: Generating High-Quality and Motion Consistent Videos using Discrete Diffusion
153、Hymba: A Hybrid-head Architecture for Small Language Models
154、Training-Free Activation Sparsity in Large Language Models
155、Determine-Then-Ensemble: Necessity of Top-k Union for Large Language Model Ensembling
156、CubeDiff: Repurposing Diffusion-Based Image Models for Panorama Generation
157、Poison-splat: Computation Cost Attack on 3D Gaussian Splatting
158、Rare-to-Frequent: Unlocking Compositional Generation Power of Diffusion Models on Rare Concepts with LLM Guidance
159、Scalable and Certifiable Graph Unlearning: Overcoming the Approximation Error Barrier
160、TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters
161、Linear Spherical Sliced Optimal Transport: A Fast Metric for Comparing Spherical Data
162、Enhancing Pre-trained Representation Classifiability can Boost its Interpretability
163、LiveBench: A Challenging, Contamination-Limited LLM Benchmark
164、Stem-OB: Generalizable Visual Imitation Learning with Stem-Like Convergent Observation through Diffusion Inversion
165、Reti-Diff: Illumination Degradation Image Restoration with Retinex-based Latent Diffusion Model
166、MRS: A Fast Sampler for Mean Reverting Diffusion based on ODE and SDE Solvers
167、SRSA: Skill Retrieval and Adaptation for Robotic Assembly Tasks
168、DenseMatcher: Learning 3D Semantic Correspondence for Category-Level Manipulation from a Single Demo
169、Beyond Next Token Prediction: Patch-Level Training for Large Language Models
170、LeFusion: Controllable Pathology Synthesis via Lesion-Focused Diffusion Models
171、LOKI: A Comprehensive Synthetic Data Detection Benchmark using Large Multimodal Models
172、Both Ears Wide Open: Towards Language-Driven Spatial Audio Generation
173、One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt
174、MetaUrban: An Embodied AI Simulation Platform for Urban Micromobility
175、UniMatch: Universal Matching from Atom to Task for Few-Shot Drug Discovery
176、DynamicCity: Large-Scale 4D Occupancy Generation from Dynamic Scenes
177、Instance-dependent Early Stopping
178、Moner: Motion Correction in Undersampled Radial MRI with Unsupervised Neural Representation
179、SVDQuant: Absorbing Outliers by Low-Rank Component for 4-Bit Diffusion Models
Accept Poster
1、Self-supervised contrastive learning performs non-linear system identification
2、Sparse autoencoders reveal selective remapping of visual concepts during adaptation
3、Multi-Label Test-Time Adaptation with Bound Entropy Minimization
4、TabM: Advancing tabular deep learning with parameter-efficient ensembling
5、ToolGen: Unified Tool Retrieval and Calling via Generation
6、Activation Gradient based Poisoned Sample Detection Against Backdoor Attacks
7、Video Action Differencing
8、Optimal Transport for Time Series Imputation
9、RaSA: Rank-Sharing Low-Rank Adaptation
10、Offline Model-Based Optimization by Learning to Rank
11、From Search to Sampling: Generative Models for Robust Algorithmic Recourse
12、Monte Carlo Planning with Large Language Model for Text-Based Game Agents
13、Robust Root Cause Diagnosis using In-Distribution Interventions
14、Boosting Neural Combinatorial Optimization for Large-Scale Vehicle Routing Problems
15、SimulPL: Aligning Human Preferences in Simultaneous Machine Translation
16、SCOPE: A Self-supervised Framework for Improving Faithfulness in Conditional Text Generation
17、Clique Number Estimation via Differentiable Functions of Adjacency Matrix Permutations
18、ADAM: An Embodied Causal Agent in Open-World Environments
19、Beware of Calibration Data for Pruning Large Language Models
20、Efficient Residual Learning with Mixture-of-Experts for Universal Dexterous Grasping
21、Herald: A Natural Language Annotated Lean 4 Dataset
22、HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere
23、DPLM-2: A Multimodal Diffusion Protein Language Model
24、Language Imbalance Driven Rewarding for Multilingual Self-improving
25、Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web Navigation
26、Distribution-Free Data Uncertainty for Neural Network Regression
27、LASeR: Towards Diversified and Generalizable Robot Design with Large Language Models
28、SOO-Bench: Benchmarks for Evaluating the Stability of Offline Black-Box Optimization
29、Stealthy Shield Defense: A Conditional Mutual Information-Based Approach against Black-Box Model Inversion Attacks
30、Be More Diverse than the Most Diverse: Optimal Mixtures of Generative Models via Mixture-UCB Bandit Algorithms
31、Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count
32、One-for-All Few-Shot Anomaly Detection via Instance-Induced Prompt Learning
33、URLOST: Unsupervised Representation Learning without Stationarity or Topology
34、Charting the Design Space of Neural Graph Representations for Subgraph Matching
35、Distilling Dataset into Neural Field
36、SIM: Surface-based fMRI Analysis for Inter-Subject Multimodal Decoding from Movie-Watching Experiments
37、SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models
38、Group-robust Sample Reweighting for Subpopulation Shifts via Influence Functions
39、GotenNet: Rethinking Efficient 3D Equivariant Graph Neural Networks
40、Efficient Automated Circuit Discovery in Transformers using Contextual Decomposition
41、How to Evaluate Reward Models for RLHF
42、An Efficient Framework for Crediting Data Contributors of Diffusion Models
43、You Only Prune Once: Designing Calibration-Free Model Compression With Policy Learning
44、SageAttention: Accurate 8-Bit Attention for Plug-and-play Inference Acceleration
45、FreDF: Learning to Forecast in the Frequency Domain
46、Have the VLMs Lost Confidence? A Study of Sycophancy in VLMs
47、Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Pass
48、Ensembling Diffusion Models via Adaptive Feature Aggregation
49、Model merging with SVD to tie the Knots
50、Making Text Embedders Few-Shot Learners
51、PolyhedronNet: Representation Learning for Polyhedra with Surface-attributed Graph
52、SafeDiffuser: Safe Planning with Diffusion Probabilistic Models
53、Searching for Optimal Solutions with LLMs via Bayesian Optimization
54、Adversarial Generative Flow Network for Solving Vehicle Routing Problems
55、LongMamba: Enhancing Mamba’s Long-Context Capabilities via Training-Free Receptive Field Enlargement
56、Lawma: The Power of Specialization for Legal Annotation
57、Stiefel Flow Matching for Moment-Constrained Structure Elucidation
58、No Preference Left Behind: Group Distributional Preference Optimization
59、Robust Weight Initialization for Tanh Neural Networks with Fixed Point Analysis
60、Agent S: An Open Agentic Framework that Uses Computers Like a Human
61、Semi-Parametric Retrieval via Binary Bag-of-Tokens Index
62、PN-GAIL: Leveraging Non-optimal Information from Imperfect Demonstrations
63、Hyperbolic Genome Embeddings
64、RTop-K: Ultra-Fast Row-Wise Top-K Selection for Neural Network Acceleration on GPUs
65、Attributing Culture-Conditioned Generations to Pretraining Corpora
66、Unveiling the Secret Recipe: A Guide For Supervised Fine-Tuning Small LLMs
67、DelTA: An Online Document-Level Translation Agent Based on Multi-Level Memory
68、Model Editing as a Robust and Denoised variant of DPO: A Case Study on Toxicity
69、A Common Pitfall of Margin-based Langu
README 内容过长，已截取前 64 KB。完整内容请前往 GitHub 仓库查看。
评论