Wenxuan Zeng | Meng Li's Homepagehttps://mengli.me/authors/wenxuan_zeng_2023g/Wenxuan ZengWowchemy (https://wowchemy.com)en-ushttps://mengli.me/media/icon_hu_ad1e0b0e8ea3c209.pngWenxuan Zenghttps://mengli.me/authors/wenxuan_zeng_2023g/OptiPrime: Optimizing Private Inference through protocol-hardware codesignhttps://mengli.me/publication/micro-2026-optiprime/Sun, 05 Jul 2026 00:00:00 +0000https://mengli.me/publication/micro-2026-optiprime/Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapsehttps://mengli.me/publication/icml-2026-attn/Fri, 01 May 2026 00:00:00 +0000https://mengli.me/publication/icml-2026-attn/MPCache: MPC-Friendly KV Cache Eviction for Efficient Private LLM Inferencehttps://mengli.me/publication/neurips-2025-mpcache/Sun, 14 Dec 2025 00:00:00 +0000https://mengli.me/publication/neurips-2025-mpcache/H2EAL: Hybrid-Bonding Architecture with Hybrid Sparse Attention for Efficient Long-Context LLM Inferencehttps://mengli.me/publication/iccad-2025-h2eal/Mon, 27 Oct 2025 00:00:00 +0000https://mengli.me/publication/iccad-2025-h2eal/UniCAIM: A Unified CAM/CIM Architecture with Static-Dynamic KV Cache Pruning for Efficient Long-Context LLM Inferencehttps://mengli.me/publication/dac-2025-unicaim/Tue, 25 Feb 2025 00:00:00 +0000https://mengli.me/publication/dac-2025-unicaim/FlexHE: A flexible Kernel Generation Framework for Homomorphic Encryption-Based Private Inferencehttps://mengli.me/publication/iccad-2024-flexhe/Sun, 27 Oct 2024 00:00:00 +0000https://mengli.me/publication/iccad-2024-flexhe/PrivQuant: Communication-Efficient Private Inference with Quantized Network/Protocol Co-Optimizationhttps://mengli.me/publication/iccad-2024-privquant/Sun, 27 Oct 2024 00:00:00 +0000https://mengli.me/publication/iccad-2024-privquant/CoPriv: Network/Protocol Co-Optimization for Communication-Efficient Private Inferencehttps://mengli.me/publication/neurips-2023-copriv/Thu, 14 Dec 2023 00:00:00 +0000https://mengli.me/publication/neurips-2023-copriv/MPCViT: Searching for Accurate and Efficient MPC-Friendly Vision Transformer with Heterogeneous Attentionhttps://mengli.me/publication/iccv-2023-mpcvit/Tue, 24 Oct 2023 00:00:00 +0000https://mengli.me/publication/iccv-2023-mpcvit/