发表机构
Faculty of Computing, Harbin Institute of Technology（哈尔滨工业大学计算机学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出目标无关微干预方法，通过分支标准化干预构建L-State，利用直接和算子读出预测跨家族语言模型的训练响应，显著降低MSE并提升符号平衡准确率。
AI 中文摘要
基准分数描述了检查点当前能做什么，但并不能决定它在下一次训练阶段中将如何响应。我们通过从同一检查点分支四个简短、标准化、目标无关的微干预，并在一个共同的能力空间中记录其效果，来衡量这一缺失的状态。结合当前能力，这些响应构成了L-State；其脉冲块支持灵活的直接读出和保持结构的算子读出。在平滑局部动力学下，算子构造允许端到端的跨家族界限，并显式考虑源家族和目标家族的坐标异质性。在三家族留一家族开发中，两种脉冲读出相对于仅使用能力，将源标准化均方误差降低了39.4%，同时分离了最佳响应和方向估计。在密封的GLM-4-9B上，直接读出和算子读出分别将均方误差降低了71.8%和78.3%，算子读出将符号平衡准确率从0.366提高到0.754。在密封的Granite-3.1-8B上，直接读出达到均方根误差0.544，开发拟合的动作选择器达到0.554，而仅使用能力时为1.172。一项五家族审计发现，算子坐标随动作和家族而变化，并且对这些偏差进行建模可改善回顾性保持轨迹预测。因此，目标无关干预暴露了当前能力所遗漏的训练响应信息，直接读出和结构化读出覆盖了互补的迁移机制。
英文摘要
Benchmark scores describe what a checkpoint can do now, but they do not determine how it will respond to the next training episode. We measure this missing state by branching four short, standardized, target-independent micro-interventions from the same checkpoint and recording their effects in a common capability space. Together with current capability, these responses form L-State; its pulse block supports a flexible direct readout and a structure-preserving operator readout. Under smooth local dynamics, the operator construction admits an end-to-end cross-family bound with explicit source- and target-family coordinate heterogeneity. In three-family leave-one-family-out development, both pulse readouts reduce source-standardized MSE by 39.4% relative to capability alone, while separating the best response and direction estimates. On sealed GLM-4-9B, the direct and operator readouts reduce MSE by 71.8% and 78.3%, respectively, and the operator readout raises sign balanced accuracy from 0.366 to 0.754. On sealed Granite-3.1-8B, the direct readout reaches RMSE 0.544 and a development-fitted action-wise selector reaches 0.554, compared with 1.172 for capability alone. A five-family audit finds that the operator coordinate varies by action and family, and that modeling these deviations improves retrospective held-trajectory prediction. Target-independent interventions therefore expose training-response information that current capability misses, with direct and structured readouts covering complementary transfer regimes.
Comments24 pages, 2 figures
