Towards Verifiable Transformers: Solver-Checkable Circuit Explanations
迈向可验证的Transformer:求解器可检查的电路解释
机构 * Independent Researcher(独立研究者)
专题命中 代码与定理证明 :reasoning(abstract);分类 cs.LG
AI总结 提出Verifiable Transformers框架,通过将任务局部Transformer电路转化为有界、求解器可检查的声明,实现电路属性的形式化验证。
Comments 23 pages. v2: adds GPT-2-scale verified distillation (three-edge verified quote circuit), LayerNorm removal for sparsemax models, and gated localization protocols