← Search

Yuebin XU

2 accepted papers

2026

TD-MoE: Tensor Decomposition for MoE Models

ICLR 2026poster

Mixture-of-Experts (MoE) architectures have demonstrated remarkable capabilities and scalability for large language models, but incur a prohibitive memory footprint due to duplicated expert parameters. Existing compression approaches, particularly those based on low-rank decomposition, typically ope…

Cited by 0SourceScholar
2025

EcoTune: Token-Efficient Multi-Fidelity Hyperparameter Optimization for Large Language Model Inference

EMNLP 2025

Tuning inference hyperparameters, such as temperature and maximum output tokens, on downstream tasks can enhance inference performance. However, directly applying hyperparameter optimization to these hyperparameters is token-expensive. Multi-fidelity optimization improves HPO efficiency with low-fid

Cited by 0SourcePDFScholar