Wnma's Blogs

03-30

TurboQuant 详解

03-06

Qwen3.5 Attention 的变化：Gated DeltaNet 详解

08-20

LIMA 阅读笔记

08-08

Knowledge Distillation in LLM

03-28

LLM 深度的”稀疏性“

01-24

MoE 自动选择专家个数 from Top-k to Top-p

11-17

Tracing Model Outputs to the Training Data

08-13

模仿游戏——机器会思考吗？

12-31

Short Paper Reading