sample-scaffolder
This skill is designed to take a skill that has been submitted as a PR and scaffold it into the sample format as an expected standard by the repository.
perf-check
Run a Maestro-style performance assessment for hotspots, regressions, and optimization planning
perf-check
Run a Maestro-style performance assessment for hotspots, regressions, and optimization planning
test-automation-strategy
Design and implement effective test automation with proper pyramid, patterns, and CI/CD integration. Use when building automation frameworks or improving test efficiency.
maya-pythonnurbsuv
使用Maya Python将选中的多边形或细分模型转换为NURBS曲面，并按照指定的U和V方向数量提取曲线。
excelbacktrader
用于将包含自定义时间列（如'bob'）的Excel金融数据读入Backtrader。该技能包含处理时间格式转换、映射OHLCV列，以及通过Pandas倒序解决K线图时间轴反向问题的完整流程。
asymmetric-cost-loss-function-false-negative-cost-0
Defines a custom loss function in TensorFlow/Keras where predicting 1 as 0 (False Negative) has zero cost, while predicting 0 as 1 (False Positive) has a cost of 1.
dual-branch-vit-rgb-event-tracking
Implements a dual-branch Vision Transformer for RGB and Event tracking, processing Template and Search tokens independently per branch and applying late fusion with Dropout regularization.
pytorch-cosineannealinglr-scheduler-integration
Integrates the CosineAnnealingLR learning rate scheduler into the existing training pipeline configuration, allowing dynamic learning rate adjustment based on cosine annealing strategy.
pytorch-ssim-l1-l2
实现一个用于图像复原任务的组合损失函数，包含结构相似性(SSIM)、平均绝对误差(L1)和均方误差(L2)的加权和，并配置对应的Adam优化器和ReduceLROnPlateau学习率调度器。
tensorflow-java-040savedmodel
针对TensorFlow Java 0.4.0版本，实现加载SavedModel模型，将三维double数组转换为Tensor并执行预测返回double数组的逻辑。
ceutrackactor
在CEUTrackActor类中集成正交高秩正规化（Orthogonal High-rank Regularization），通过在损失函数中添加基于注意力矩阵SVD的正则化项，以提升模型的特征区分能力和泛化能力。
matlab-rect-pulse-sampling-recovery
基于MATLAB实现矩形脉冲信号的生成、频谱分析确定fm、理想抽样（2倍fm）、特定增益低通滤波及信号恢复的完整实验流程。
pytorch-dynamic-loss-weighting
指导在PyTorch中实现多任务学习或知识蒸馏场景下的动态Loss权重调整，涵盖可学习标量权重、GradNorm算法及基于不确定性的加权方法，并解决设备一致性、计算图错误及权重约束问题。
tensorflow
针对TensorFlow训练代码进行内存泄漏修复，包括优化数据管道、添加每轮结束后的垃圾回收回调以及修正ModelCheckpoint配置。
vit-module-fusion-and-training-optimization
集成UEP或Counter_Guide等模块到Vision Transformer，处理模块适配（2D转1D）、并行或条件插入逻辑，并针对新增模块调整训练超参数以防止过拟合。