Scholar
Zefan Cai
Google Scholar ID: eyIrttAAAAAJ
Student, Peking University
Citations & Impact
Citations
1,601
H-index
15
i10-index
21
Publications
20
Co-authors
16
list available
Publications
Resume (English only)
Background
- Research interests include LLM efficiency (e.g., KV cache retrieval, off-loading, compression, and other inference optimization topics), LLM architecture (e.g., native sparse attention, test-time training), and LLM memorization (e.g., parametric memory, agent memory). He considers long-context as the most important problem in LLM.
