Text2PDE: Latent Diffusion Models for Accessible Physics Simulation
Comments Published at ICLR 2025. Github: http://github.com/anthonyzhou-1/ldm_pdes
期刊&会议
ICLR
International Conference on Learning Representations · 会议 · Machine Learning
MME-RealWorld: Could Your Multimodal LLM Challenge High-Resolution Real-World Scenarios that are Difficult for Humans?
Comments Project Page: https://mme-realworld.github.io/; accepted by ICLR 2025
Dissecting Adversarial Robustness of Multimodal LM Agents
Comments ICLR 2025. Also oral at NeurIPS 2024 Open-World Agents Workshop
Towards Realistic Data Generation for Real-World Super-Resolution
Comments accepted by ICLR 2025
A Theoretical Analysis of Self-Supervised Learning for Vision Transformers
Comments Accepted by ICLR 2025
Shot2Story: A New Benchmark for Comprehensive Understanding of Multi-shot Videos
Comments ICLR 2025. Extended annotation with 43K multi-shot videos in total. https://mingfei.info/shot2story for updates and more information
DeepAveragers: Offline Reinforcement Learning by Solving Derived Non-Parametric MDPs
Comments Camera Ready ICLR 2021
Addressing Label Shift in Distributed Learning via Entropy Regularization
Comments Accepted at the International Conference on Learning Representations (ICLR 2025)
Graph-based Document Structure Analysis
Comments Accepted by ICLR 2025. Project page: https://yufanchen96.github.io/projects/GraphDoc
Wavelet-based Positional Representation for Long Context
Comments Accepted to ICLR 2025. 28 pages, 11 figures
Memory Efficient Transformer Adapter for Dense Predictions
Comments This paper is accepted by ICLR 2025
LoRA-X: Bridging Foundation Models with Training-Free Cross-Model Adaptation
Comments Accepted to ICLR 2025
Posterior-Mean Rectified Flow: Towards Minimum MSE Photo-Realistic Image Restoration
Comments Accepted to ICLR 2025. Code and demo are available at https://pmrf-ml.github.io/
Towards Marginal Fairness Sliced Wasserstein Barycenter
Comments Accepted to ICLR 2025, 29 pages, 15 figures, 6 tables
BEEM: Boosting Performance of Early Exit DNNs using Multi-Exit Classifiers as Experts
Comments Published at International Conference on Learning Representations (ICLR) 2025
Ensembles of Low-Rank Expert Adapters
Comments 29 pages, 5 figures, 5 tables; proceedings in ICLR 2025
Reflective Gaussian Splatting
Comments Accepted for ICLR 2025
Multi-modal Agent Tuning: Building a VLM-Driven Agent for Efficient Tool Usage
Comments ICLR 2025, https://mat-agent.github.io/
Large Language Models are Advanced Anonymizers
Comments International Conference on Learning Representations (ICLR 2024)
From Few to Many: Self-Improving Many-Shot Reasoners Through Iterative Optimization and Generation
Comments Expanded version of the ICLR 2025 paper
ProtoSnap: Prototype Alignment for Cuneiform Signs
Comments Accepted to ICLR 2025. Project page: https://tau-vailab.github.io/ProtoSnap/
Homomorphism Counts as Structural Encodings for Graph Learning
Comments Proceedings of the Thirteenth International Conference on Learning Representations (ICLR 202R). Code available at: https://github.com/linusbao/MoSE
Efficient Source-Free Time-Series Adaptation via Parameter Subspace Disentanglement
Comments Accepted at ICLR 2025
The Crucial Role of Samplers in Online Direct Preference Optimization
Comments ICLR accepted
Scrutinize What We Ignore: Reining In Task Representation Shift Of Context-Based Offline Meta Reinforcement Learning
Comments Accept at ICLR 2025
Circuit Transformer: A Transformer That Preserves Logical Equivalence
Comments In ICLR 2025
Group Distributionally Robust Dataset Distillation with Risk Minimization
Comments ICLR 2025
RGB-Event ISP: The Dataset and Benchmark
Comments Accepted by ICLR 2025; 14 pages, 8 figures, 4 tables
Towards Fast, Specialized Machine Learning Force Fields: Distilling Foundation Models via Energy Hessians
Comments Accepted as a conference paper at ICLR 2025. The implementation of our method is available at https://github.com/ASK-Berkeley/MLFF-distill
Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?
Comments Published as a conference paper at ICLR 2025, GitHub: https://github.com/egozverev/Shold-It-Be-Executed-Or-Processed. 10 pages main text, 30 pages in total
