Lijie(Derrick) Yang

TidalDecode. [ICLR 2025] TidalDecode: A Fast and Accurate LLM Decoding with Position Persistent Sparse Attention

LessIsMore. [ICML 2026] Less Is More: Training-Free Sparse Attention with Global Locality for Efficient Reasoning

Blocking_Waived_Estimation. [LCN 2024] solving worst case delay of relatively complicated network architecture with [1] Trajectory Approach; [2] Network Calculus; [3] Compositional Performance Analysis (CPA); and [4] Flow Aggregation and summarize both advantages and disadvantages of each approach and strives to seek out the optimal method under specific scenarios.

WCD_Calculation. This is used for the LORIA research internship and worst-case delay (WCD) calculation.

Xcircuit-GPT-Circuit-Classifier. Jupyter Notebook