LatentQA: Teaching LLMs to Decode Activations Into Natural Languageby Alexander Pan, Lijie Chen, Jacob SteinhardtFirst…
Test-Time Alignment via Hypothesis Reweightingby Yoonho Lee, Jonathan Williams, Henrik Marklund, Archit Sharma, Eric Mitchell,…
GPD-1: Generative Pre-training for Drivingby Zixun Xie, Sicheng Zuo, Wenzhao Zheng, Yunpeng Zhang, Dalong Du,…
ALoRE: Efficient Visual Adaptation via Aggregating Low Rank Expertsby Sinan Du, Guosheng Zhang, Keyao Wang,…
Forking Paths in Neural Text Generationby Eric Bigelow, Ari Holtzman, Hidenori Tanaka, Tomer UllmanFirst submitted…
Distributed Gradient Descent with Many Local Steps in Overparameterized Modelsby Heng Zhu, Harsh Vardhan, Arya…
Efficient Online Reinforcement Learning Fine-Tuning Need Not Retain Offline Databy Zhiyuan Zhou, Andy Peng, Qiyang…
Efficient Diversity-Preserving Diffusion Alignment via Gradient-Informed GFlowNetsby Zhen Liu, Tim Z. Xiao, Weiyang Liu, Yoshua…
MoDULA: Mixture of Domain-Specific and Universal LoRA for Multi-Task Learningby Yufei Ma, Zihan Liang, Huangyu…
Towards Graph Foundation Models: A Study on the Generalization of Positional and Structural Encodingsby Billy…