Assessing Contamination in Large Language Models: Introducing the LogProber methodby Nicolas Yax, Pierre-Yves Oudeyer, Stefano…
Selective Preference Optimization via Token-Level Reward Function Estimationby Kailai Yang, Zhiwei Liu, Qianqian Xie, Jimin…
A Law of Next-Token Prediction in Large Language Modelsby Hangfeng He, Weijie J. SuFirst submitted…
HMoE: Heterogeneous Mixture of Experts for Language Modelingby An Wang, Xingwu Sun, Ruobing Xie, Shuaipeng…
A Lightweight Modular Framework for Low-Cost Open-Vocabulary Object Detection Trainingby Bilal Faye, Binta Sow, Hanane…
Tracing Privacy Leakage of Language Models to Training Data via Adjusted Influence Functionsby Jinxin Liu,…
Integrating Multi-Modal Input Token Mixer Into Mamba-Based Decision Models: Decision MetaMambaby Wall KimFirst submitted to…
AdapMoE: Adaptive Sensitivity-based Expert Gating and Management for Efficient MoE Inferenceby Shuzhang Zhong, Ling Liang,…
Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recyclingby Xianzhen Luo,…
Where is the signal in tokenization space?by Renato Lui Geh, Honghua Zhang, Kareem Ahmed, Benjie…