Matthew Smith had to wait almost a year for a double lung transplant
Rank-1 linear, factorized embed, sparse gate, param-free norm, low-rank head, cross-layer sharing
,更多细节参见safew官方版本下载
// 易错点1:边界处理 - 移除所有数字时直接返回"0"
const sb = _apply.call(origAddSB, this, arguments);