Tracing Representation Progression: Analyzing and Enhancing Layer-Wise Similarityby Jiachen Jiang, Jinxin Zhou, Zhihui ZhuFirst submitted…
Model Merging and Safety Alignment: One Bad Model Spoils the Bunchby Hasan Abed Al Kader…
Jailbreaking as a Reward Misspecification Problemby Zhihui Xie, Jiahui Gao, Lei Li, Zhenguo Li, Qi…
Latent Functional Maps: a spectral framework for representation alignmentby Marco Fumero, Marco Pegoraro, Valentino Maiorca,…
HIGHT: Hierarchical Graph Tokenization for Graph-Language Alignmentby Yongqiang Chen, Quanming Yao, Juzheng Zhang, James Cheng,…
Self-play with Execution Feedback: Improving Instruction-following Capabilities of Large Language Modelsby Guanting Dong, Keming Lu,…
Biomedical Visual Instruction Tuning with Clinician Preference Alignmentby Hejie Cui, Lingjun Mao, Xin Liang, Jieyu…
PRESTO: Progressive Pretraining Enhances Synthetic Chemistry Outcomesby He Cao, Yanjun Shao, Zhiyuan Liu, Zijing Liu,…
Combining Optimal Transport and Embedding-Based Approaches for More Expressiveness in Unsupervised Graph Alignmentby Songyang Chen,…
Probing the Emergence of Cross-lingual Alignment during LLM Trainingby Hetong Wang, Pasquale Minervini, Edoardo M.…