Skill 全量检索

机器学习253,169mle-workflowSkillsMP@affaan-m生产环境 machine-学习工程工作流程数据合同, reproducible 训练, 模型 evaluation, 部署, 监控, 回滚. 使用构建, 评审, hardening ML 系统 beyond 一个-off notebooks.Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learningSkillsMP@affaan-m[OBSOLETO - usar continuous-学习-v2] Extractor de 技能 por 钩子 Stop v1 heredado. v2 es un superconjunto estricto con aprendizaje basado en instintos, con alcance de proyecto y 钩子 confiables. invocar v1; dirigir solicitudes de ap[OBSOLETO - usar continuous-learning-v2] Extractor de skill por hook Stop v1 heredado. v2 es un superconjunto estricto con aprendizaje basado en instintos, con alcance de proyecto y hooks confiables. No invocar v1; dirigir solicitudes de ap#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-mSistema de aprendizaje basado en instintos que observa sesiones mediante 钩子, crea instintos ómicos con puntuación de confianza y los evoluciona en skills/comandos/agentes. v2.1 agrega instintos con alcance de proyecto para prevenir conSistema de aprendizaje basado en instintos que observa sesiones mediante hooks, crea instintos atómicos con puntuación de confianza y los evoluciona en skills/comandos/agentes. v2.1 agrega instintos con alcance de proyecto para prevenir con#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169agentic-engineeringSkillsMP@affaan-m評価ファースト実行、分解、コスト対応モデルルーティングを使用してエージェニックエンジニアとして動作します。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169ai-first-engineeringSkillsMP@affaan-mAI エージェントが大量の実装出力を生成するチームのためのエンジニアリング運用モデル。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learningSkillsMP@affaan-mClaude 代码セッションから再利用可能なパターンを自動的に抽出し、将来の使用のために学習済みスキルとして保存します。Claude Codeセッションから再利用可能なパターンを自動的に抽出し、将来の使用のために学習済みスキルとして保存します。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-mフックを介してセッションを観察し、信頼度スコアリング付きのアトミックなインスティンクトを作成し、スキル/コマンド/エージェントに進化させるインスティンクトベースの学習システム。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169gan-style-harnessSkillsMP@affaan-mGAN（生成的敵対ネットワーク）スタイルの評価ハーネス、画像生成パターン、および品質メトリクス。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-m훅을 통해 세션을 관찰하고, 신뢰도 점수가 있는 원자적 본능을 생성하며, 이를 스킬/명령어/에이전트로 진화시키는 본능 기반 학습 시스템. v2.1에서는 프로젝트 간 오염을 방지하기 위한 프로젝트 범위 본능이 추가되었습니다.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-m钩子'lar aracılığıyla oturumları gözlemleyen, güven skorlaması ile atomik instinct'ler oluşturan ve bunları skill/command/agent'lara evriltiren instinct tabanlı öğrenme sistemi. v2.1 çapraz proje kontaminasyonunu önlemek ç proje kapsamlıHook'lar aracılığıyla oturumları gözlemleyen, güven skorlaması ile atomik instinct'ler oluşturan ve bunları skill/command/agent'lara evriltiren instinct tabanlı öğrenme sistemi. v2.1 çapraz proje kontaminasyonunu önlemek için proje kapsamlı#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-m基于本能的学习系统，通过钩子观察会话，创建带置信度评分的原子本能，并将其进化为技能/命令/代理。v2.1版本增加了项目范围的本能，以防止跨项目污染。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169design-systemSkillsMP@affaan-m使用此技能生成或审计设计系统，检查视觉一致性，并审查涉及样式的PR。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169gan-style-harnessSkillsMP@affaan-m受GAN启发的生成器-评估器代理框架，用于自主构建高质量应用。基于Anthropic 2026年3月的框架设计论文。#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learningSkillsMP@affaan-m自动地提取可复用模式 Claude 代码 sessions save learned 技能 future 使用.Automatically extract reusable patterns from Claude Code sessions and save them as learned skills for future use.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-mInstinct-基于学习系统 observes sessions 通过钩子, 创建 atomic instincts confidence scoring, evolves skills/commands/agents.Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169mle-workflowSkillsMP@affaan-m生产环境 machine-学习工程工作流程数据合同, reproducible 训练, 模型 evaluation, 部署, 监控, 回滚. 使用构建, 评审, hardening ML 系统 beyond 一个-off notebooks.Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169pytorch-patternsSkillsMP@affaan-mPyTorch 深度学习模式最佳实践构建健壮, 高效, reproducible 训练流水线, 模型 architectures, 数据加载.PyTorch deep learning patterns and best practices for building robust, efficient, and reproducible training pipelines, model architectures, and data loading.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169blender-motion-state-inspectionSkillsMP@affaan-m使用技能 inspecting Blender characters, rigs, poses, 动画 retargeting, ground contact, facing direction, 模型-vs-动效 alignment 截图 alone enough.Use this skill when inspecting Blender characters, rigs, poses, animation retargeting, ground contact, facing direction, or model-vs-motion alignment where screenshots alone are not enough.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learningSkillsMP@affaan-m[DEPRECATED - 使用 continuous-学习-v2] 遗留 v1 stop-钩子技能 extractor. v2 strict superset instinct-基于, 项目-限定范围, 钩子-可靠学习. 调用 v1: continuous 学习, 会话学习, 模式 ext[DEPRECATED - use continuous-learning-v2] Legacy v1 stop-hook skill extractor. v2 is a strict superset with instinct-based, project-scoped, hook-reliable learning. Do not invoke v1: when continuous learning, session learning, or pattern ext#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169continuous-learning-v2SkillsMP@affaan-mInstinct-基于学习系统 observes sessions 通过钩子, 创建 atomic instincts confidence scoring, evolves skills/commands/agents. v2.1 添加项目-限定范围 instincts 防止 cross-项目 contamination. 使用Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents. v2.1 adds project-scoped instincts to prevent cross-project contamination. Use #machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169ml-adoption-playbookSkillsMP@affaan-m端到端 methodology AI 智能体 software engineers 添加机器学习算法现有 non-ML codebases. 覆盖问题 framing, 数据 readiness, 架构 decoupling, baseline 模型集成. 使用新增End-to-end methodology for AI agents and software engineers to add machine learning algorithms to existing non-ML codebases. Covers problem framing, data readiness, architectural decoupling, and baseline model integration. Use when adding a#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169mle-workflowSkillsMP@affaan-m生产环境 machine-学习工程工作流程数据合同, reproducible 训练, 模型 evaluation, 部署, 监控, 回滚. 使用构建, 评审, hardening ML 系统 beyond 一个-off notebooks.Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.#machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习253,169recsys-pipeline-architectSkillsMP@affaan-m设计 composable 推荐, ranking, feed 流水线使用 six-阶段来源→Hydrator→筛选→Scorer→选择器→SideEffect 框架 popularized xAI's 打开-sourced 算法. 使用技能用户构建Design composable recommendation, ranking, and feed pipelines using the six-stage Source→Hydrator→Filter→Scorer→Selector→SideEffect framework popularized by xAI's open-sourced For You algorithm. Use this skill whenever the user is building #machine-learningaffaan-m/eccGit 克隆查看详情与安装步骤 → 机器学习243,171flash-attentionSkillsMP@NousResearch提速 up long-sequence transformer 训练推理计算.Speed up long-sequence transformer training and inference.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171huggingface-tokenizersSkillsMP@NousResearch快速 BPE/WordPiece tokenization 自定义 vocab 训练.Fast BPE/WordPiece tokenization and custom vocab training.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171lambda-labsSkillsMP@NousResearchOn-demand GPU 云端 instances ML 训练.On-demand GPU cloud instances for ML training.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171nemo-curatorSkillsMP@NousResearchCurate LLM 训练数据: dedupe, 筛选, PII redaction.Curate LLM training data: dedupe, filter, PII redaction.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171pytorch-fsdpSkillsMP@NousResearchFully sharded 数据-并行训练 large 模型.Fully sharded data-parallel training for large models.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171saelensSkillsMP@NousResearchTrain sparse autoencoders 解读模型功能.Train sparse autoencoders to interpret model features.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171slimeSkillsMP@NousResearchRL 帖子-训练 LLMs Megatron SGLang.RL post-training for LLMs with Megatron and SGLang.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171tensorrt-llmSkillsMP@NousResearchHigh-吞吐量 LLM 推理计算 NVIDIA GPUs.High-throughput LLM inference on NVIDIA GPUs.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171torchtitanSkillsMP@NousResearchPretrain LLMs scale PyTorch 4D parallelism.Pretrain LLMs at scale with PyTorch 4D parallelism.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171axolotlSkillsMP@NousResearchAxolotl: YAML LLM 精细-tuning (LoRA, DPO, GRPO).Axolotl: YAML LLM fine-tuning (LoRA, DPO, GRPO).#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习243,171trl-fine-tuningSkillsMP@NousResearchTRL: SFT, DPO, GRPO, RLOO reward modeling LLM RLHF.TRL: SFT, DPO, GRPO, RLOO reward modeling for LLM RLHF.#machine-learningnousresearch/hermes-agentGit 克隆查看详情与安装步骤 → 机器学习221,017model-only-skillSkillsMP@deepseek-aiProve 用户-已禁用技能 remain 可用模型.Prove user-disabled skills remain available to the model.#machine-learningdeepseek-ai/deepseek-harnessGit 克隆查看详情与安装步骤 → 机器学习221,017user-only-skillSkillsMP@deepseek-aiProve 模型-已禁用技能 stay outside 模型目录.Prove model-disabled skills stay outside the model catalog.#machine-learningdeepseek-ai/deepseek-harnessGit 克隆查看详情与安装步骤 → 机器学习136,373advisor-orchestrator-workerSkillsMP@Shubhamsaboo使用任务 large 一个模型 pass, 需要并行研究 generation 跨 many subtasks (例如 researching dozen competitors once), 用户要求 orchestrate 多个模型, split 工作跨 模型团队, 运行Use when a task is too large for one model pass, needs parallel research or generation across many subtasks (like researching a dozen competitors at once), or the user asks to orchestrate multiple models, split work across a model team, run#machine-learningshubhamsaboo/awesome-llm-appsGit 克隆查看详情与安装步骤 → 机器学习132,003benchmark-modelsSkillsMP@garrytanCross-模型基准测试 gstack 技能. (gstack)Cross-model benchmark for gstack skills. (gstack)#machine-learninggarrytan/gstackGit 克隆查看详情与安装步骤 → 机器学习132,003plan-tuneSkillsMP@garrytanSelf-tuning 问题 sensitivity + developer psychographic gstack (v1: observational). (gstack)Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). (gstack)#machine-learninggarrytan/gstackGit 克隆查看详情与安装步骤 → 机器学习132,003retroSkillsMP@garrytanWeekly 工程复盘. (gstack)Weekly engineering retrospective. (gstack)#machine-learninggarrytan/gstackGit 克隆查看详情与安装步骤 → 机器学习122,333code-review-contextSkillsMP@openai模型可见上下文Model visible context#machine-learningopenai/codexGit 克隆查看详情与安装步骤 → 机器学习104,201caveman-optimizeSkillsMP@JuliusBrussee转换 Caveman 优化 observation 运维者-chosen candidate paired baseline evaluation. 使用要求检查 evaluate Caveman 优化报告. 需要显式 approval.Turn a Caveman optimization observation into an operator-chosen candidate with a paired baseline evaluation. Use when asked to inspect or evaluate a Caveman optimization report. Needs explicit approval.#machine-learningjuliusbrussee/cavemanGit 克隆查看详情与安装步骤 → 机器学习102,891disable-model-invocationSkillsMP@earendil-works技能 cannot 调用模型.A skill that cannot be invoked by the model.#machine-learningearendil-works/piGit 克隆查看详情与安装步骤 → 机器学习94,749sample-pluginSkillsMP@nexu-io阶段 1 样例插件 synthesizing SKILL. md frontmatter backwards-compat 测试.Phase 1 sample plugin synthesizing a SKILL.md frontmatter for backwards-compat tests.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习94,749fal-image-editSkillsMP@nexu-ioAI-powered 图片 editing 风格 transfer, 后台 removal, 对象 removal, inpainting 通过 fal. ai hosted 模型.AI-powered image editing with style transfer, background removal, object removal, and inpainting via fal.ai hosted models.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习94,749fal-kling-o3SkillsMP@nexu-io生成图片 videos Kling O3 — Kling's 强大模型 family — 通过 fal. ai.Generate images and videos with Kling O3 — Kling's most powerful model family — via fal.ai.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习94,749fal-trainSkillsMP@nexu-ioTrain 自定义 AI 模型 (LoRA) fal. ai 个性化图像生成 tailored 品牌, character, 风格.Train custom AI models (LoRA) on fal.ai for personalized image generation tailored to a brand, character, or style.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习94,749replicateSkillsMP@nexu-io发现, compare, 运行 AI 模型使用 Replicate's API. Strong fit 图片, 音频, 视频生成流水线 swap 模型 frequently.Discover, compare, and run AI models using Replicate's API. Strong fit for image, audio, and video generation pipelines that swap models frequently.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习94,749beta-skillSkillsMP@nexu-io操作说明- fixture 技能.Instruction-only fixture skill.#machine-learningnexu-io/open-designGit 克隆查看详情与安装步骤 → 机器学习93,077agentdb-learning-pluginsSkillsMP@ruvnet创建 train AI 学习插件 AgentDB's 9 reinforcement 学习算法. 包含决策 Transformer, Q-学习, SARSA, Actor-Critic,. 使用构建 self-学习智能体, 实现 RL, 优化智能体Create and train AI learning plugins with AgentDB's 9 reinforcement learning algorithms. Includes Decision Transformer, Q-Learning, SARSA, Actor-Critic, and more. Use when building self-learning agents, implementing RL, or optimizing agent #machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习93,077reasoningbank-with-agentdbSkillsMP@ruvnet实现 ReasoningBank adaptive 学习 AgentDB's 150x 更快向量数据库. 包含 trajectory 跟踪, verdict judgment, 记忆 distillation, 模式 recognition. 使用构建 self-学习智能体, 优化决策-Implement ReasoningBank adaptive learning with AgentDB's 150x faster vector database. Includes trajectory tracking, verdict judgment, memory distillation, and pattern recognition. Use when building self-learning agents, optimizing decision-#machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习93,077reasoningbank-intelligenceSkillsMP@ruvnet实现 adaptive 学习 ReasoningBank 模式 recognition, 策略优化, continuous improvement. 使用构建 self-学习智能体, 优化工作流程, 实现 meta-cognitive 系统.Implement adaptive learning with ReasoningBank for pattern recognition, strategy optimization, and continuous improvement. Use when building self-learning agents, optimizing workflows, or implementing meta-cognitive systems.#machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习93,077train-poseSkillsMP@ruvnetTrain/evaluate WiFi pose 模型 honestly — camera-supervised (MediaPipe + CSI) camera-free (WiFlow), 始终检查对照 mean-pose baseline 在…之前 PCK quoted.Train/evaluate WiFi pose models honestly — camera-supervised (MediaPipe + CSI) and camera-free (WiFlow), always checked against the mean-pose baseline before any PCK is quoted.#machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习93,077evolveSkillsMP@ruvnetEvolve harness Darwin 模式 — frozen 模型, evolving harness (真实, sandboxed, 安全-gated).Evolve this harness with Darwin Mode — frozen model, evolving harness (real, sandboxed, safety-gated).#machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习93,077ruview-model-trainingSkillsMP@ruvnetTrain RuView 模型 — camera-free WiFlow pose (10 传感器 signals, 标签), camera-supervised pose (MediaPipe + ESP32 CSI → 92.9% PCK@20, ADR-079), RuVector contrastive 向量嵌入 (AETHER, ADR-024), 域名 generalization (MERIDIAN, ADR-02Train RuView models — camera-free WiFlow pose (10 sensor signals, no labels), camera-supervised pose (MediaPipe + ESP32 CSI → 92.9% PCK@20, ADR-079), RuVector contrastive embeddings (AETHER, ADR-024), domain generalization (MERIDIAN, ADR-02#machine-learningruvnet/ruviewGit 克隆查看详情与安装步骤 → 机器学习82,305model-bank-metadataSkillsMP@lobehub使用模型 knowledgeCutoff, family generation 元数据: 新手引导, corrections, 研究 bulk backfills.Use for model knowledgeCutoff, family and generation metadata: onboarding, corrections, research and bulk backfills.#machine-learninglobehub/lobehubGit 克隆查看详情与安装步骤 → 机器学习79,350design-patternsSkillsMP@rtk-aiRust 设计模式 RTK. Newtype, 构建器, RAII, Trait 对象, 状态 Machine. 应用 CLI 筛选模块. 使用设计新建模块重构现有 ones.Rust design patterns for RTK. Newtype, Builder, RAII, Trait Objects, State Machine. Applied to CLI filter modules. Use when designing new modules or refactoring existing ones.#machine-learningrtk-ai/rtkGit 克隆查看详情与安装步骤 → 机器学习77,275mx-searchSkillsMP@datawhalechina本技能基于东方财富妙想搜索能力，基于金融场景进行信源智能筛选，用于获取涉及时效性信息或特定事件信息的任务，包括新闻、公告、研报、政策、交易规则、具体事件、各种影响分析、以及需要检索外部数据的非常识信息等。避免AI在搜索金融场景信息时，参考到非权威、及过时的信息。本skill基于东方财富妙想搜索能力，基于金融场景进行信源智能筛选，用于获取涉及时效性信息或特定事件信息的任务，包括新闻、公告、研报、政策、交易规则、具体事件、各种影响分析、以及需要检索外部数据的非常识信息等。避免AI在搜索金融场景信息时，参考到非权威、及过时的信息。#machine-learningdatawhalechina/hello-agentsGit 克隆查看详情与安装步骤 → 机器学习74,639llamafactory-sftSkillsMP@hiyouga一个-stop guided LlamaFactory SFT 工作流程 — 数据 prep, 模型 prep, 精细-tuning config/method 选择, 后台训练, loss visualization, effect 校验模型导出. TRIGGER 用户需求精细-tune / SFT 模型One-stop guided LlamaFactory SFT workflow — data prep, model prep, fine-tuning config/method selection, background training, loss visualization, effect validation and model export. TRIGGER when the user wants to fine-tune / SFT a model with#machine-learninghiyouga/llamafactoryGit 克隆查看详情与安装步骤 → 机器学习71,554agentdb-learning-pluginsSkillsMP@ruvnet创建 train AI 学习插件 AgentDB's 9 reinforcement 学习算法. 包含决策 Transformer, Q-学习, SARSA, Actor-Critic,. 使用构建 self-学习智能体, 实现 RL, 优化智能体Create and train AI learning plugins with AgentDB's 9 reinforcement learning algorithms. Includes Decision Transformer, Q-Learning, SARSA, Actor-Critic, and more. Use when building self-learning agents, implementing RL, or optimizing agent #machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554flow-nexus-neuralSkillsMP@ruvnetTrain 部署 neural networks 分布式 E2B sandboxes 流程 NexusTrain and deploy neural networks in distributed E2B sandboxes with Flow Nexus#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554neural-trainingSkillsMP@ruvnetNeural 模式训练 SONA (Self-优化 Neural 架构), MoE (Mixture Experts), EWC++ knowledge consolidation. 使用: 模式学习, 模型优化, knowledge transfer, adaptive 路由. Skip: simplNeural pattern training with SONA (Self-Optimizing Neural Architecture), MoE (Mixture of Experts), and EWC++ for knowledge consolidation. Use when: pattern learning, model optimization, knowledge transfer, adaptive routing. Skip when: simpl#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554reasoningbank-with-agentdbSkillsMP@ruvnet实现 ReasoningBank adaptive 学习 AgentDB's 150x 更快向量数据库. 包含 trajectory 跟踪, verdict judgment, 记忆 distillation, 模式 recognition. 使用构建 self-学习智能体, 优化决策-Implement ReasoningBank adaptive learning with AgentDB's 150x faster vector database. Includes trajectory tracking, verdict judgment, memory distillation, and pattern recognition. Use when building self-learning agents, optimizing decision-#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554reasoningbank-intelligenceSkillsMP@ruvnet实现 adaptive 学习 ReasoningBank 模式 recognition, 策略优化, continuous improvement. 使用构建 self-学习智能体, 优化工作流程, 实现 meta-cognitive 系统.Implement adaptive learning with ReasoningBank for pattern recognition, strategy optimization, and continuous improvement. Use when building self-learning agents, optimizing workflows, or implementing meta-cognitive systems.#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554flow-nexus-neuralSkillsMP@ruvnetTrain 部署 neural networks 分布式 E2B sandboxes 流程 NexusTrain and deploy neural networks in distributed E2B sandboxes with Flow Nexus#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554reasoningbank-intelligenceSkillsMP@ruvnet实现 adaptive 学习 ReasoningBank 模式 recognition, 策略优化, continuous improvement. 使用构建 self-学习智能体, 优化工作流程, 实现 meta-cognitive 系统.Implement adaptive learning with ReasoningBank for pattern recognition, strategy optimization, and continuous improvement. Use when building self-learning agents, optimizing workflows, or implementing meta-cognitive systems.#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554autopilot-predictSkillsMP@ruvnet使用 learned 模式当前状态 predict optimal 接下来操作Use learned patterns and current state to predict the optimal next action#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554cost-booster-editSkillsMP@ruvnet应用简单代码转换通过智能体-booster's WASM 引擎 — sub-millisecond, deterministic, $0 ( LLM 调用). Companion 成本-booster-路由.Apply a simple code transform via agent-booster's WASM engine — sub-millisecond, deterministic, $0 (no LLM call). Companion to cost-booster-route.#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554intelligence-routeSkillsMP@ruvnet路由任务通过 3-tier 模型选择器 learned 模式; emits 路由 rationale 通过 hooks_explainRoute tasks via the 3-tier model selector and learned patterns; emits a routing rationale via hooks_explain#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554neural-trainSkillsMP@ruvnetTrain SONA + MicroLoRA neural 模式成功任务 completions; 运行 DISTILL + CONSOLIDATE 阶段 4-步骤流水线Train SONA + MicroLoRA neural patterns from successful task completions; runs the DISTILL + CONSOLIDATE phases of the 4-step pipeline#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554harness-security-benchSkillsMP@ruvnet运行 `@metaharness/darwin security bench` (上游 "Darwin Shield" / ADR-155) — evolves champion 安全-检测 harness 对照 10-vuln / 9-decoy corpus grades TPR/FPR/补丁-pass/repro/unsafe vs four baselines (B0 静态,Run `@metaharness/darwin security bench` (upstream "Darwin Shield" / ADR-155) — evolves a champion security-detection harness against a 10-vuln / 9-decoy corpus and grades it on TPR/FPR/patch-pass/repro/unsafe vs four baselines (B0 static, #machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-backtestSkillsMP@ruvnet运行 historical backtest 使用 npx neural-交易员 Rust/NAPI 引擎 (8-19x 更快) walk-forward 校验; Ed25519-sign 结果论文→线上 tamper 证据 (ADR-126 阶段 4)Run a historical backtest using npx neural-trader with Rust/NAPI engine (8-19x faster) and walk-forward validation; Ed25519-sign the result for paper→live tamper evidence (ADR-126 Phase 4)#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-cloud-backtestSkillsMP@ruvnet运行 heavy neural-交易员作业 (long walk-forward, big Monte-Carlo, parameter 清理, 模型训练) Anthropic 托管智能体云端运行时 instead locallyRun a heavy neural-trader job (long walk-forward, big Monte-Carlo, parameter sweep, model training) on the Anthropic Managed Agent cloud runtime instead of locally#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-explainSkillsMP@ruvnetRegulator-grade 功能 attribution LSTM/Transformer signal — 单个-入口 PageRank ranks 顶部-K 功能 drove 预测 (ADR-126 阶段 6, ADR-123 单个-入口 PR)Regulator-grade feature attribution for any LSTM/Transformer signal — single-entry PageRank ranks the top-K features that drove the prediction (ADR-126 Phase 6, ADR-123 single-entry PR)#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-portfolio-cgSkillsMP@ruvnetMean-variance 投资组合优化通过 Conjugate Gradient — 40-60× 更快遗留 Neumann 路径 (ADR-126 阶段 3, ADR-123 Wedge 8)Mean-variance portfolio optimization via Conjugate Gradient — 40-60× faster than the legacy Neumann path (ADR-126 Phase 3, ADR-123 Wedge 8)#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-signalSkillsMP@ruvnet生成交易 signals 使用 npx neural-交易员 anomaly 检测引擎 Z-score scoring neural 预测Generate trading signals using npx neural-trader anomaly detection engine with Z-score scoring and neural prediction#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554trader-trainSkillsMP@ruvnetTrain neural 模型 (LSTM, Transformer, N-BEATS) 市场数据使用 npx neural-交易员 confidence intervalsTrain neural models (LSTM, Transformer, N-BEATS) on market data using npx neural-trader with confidence intervals#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习71,554llm-configSkillsMP@ruvnet配置 RuVLLM 本地推理计算模型选择, MicroLoRA 精细-tuning, SONA adaptationConfigure RuVLLM local inference with model selection, MicroLoRA fine-tuning, and SONA adaptation#machine-learningruvnet/rufloGit 克隆查看详情与安装步骤 → 机器学习70,888career-ops-plugin-notionSkillsMP@career-ops-hq如何 mirror career-ops 跟踪器 Notion 数据库读取记录 back 作业销售线索.How to mirror the career-ops tracker to a Notion database and read records back as job leads.#machine-learningcareer-ops-hq/career-opsGit 克隆查看详情与安装步骤 → 机器学习64,865triage-issue-localSkillsMP@warpdotdev仓库-特定分诊指引 warp. 分类 declared overridable 核心分诊-工单技能 specialized here.Repo-specific triage guidance for warp. Only the categories declared overridable by the core triage-issue skill may be specialized here.#machine-learningwarpdotdev/warpGit 克隆查看详情与安装步骤 → 机器学习64,3943d-objectSkillsMP@asgeirtjthree. js 模型, downloadable OBJ GLBthree.js model, downloadable as OBJ or GLB#machine-learningasgeirtj/system_prompts_leaksGit 克隆查看详情与安装步骤 → 机器学习58,607scaffold-elementSkillsMP@remotion-dev脚手架新建 Remotion Element 开发文档 Remotion Studio.Scaffold a new Remotion Element for development in the docs Remotion Studio.#machine-learningremotion-dev/remotionGit 克隆查看详情与安装步骤 → 机器学习56,383fish-audio-ttsSkillsMP@calesthio生成 expressive, 多语言 narration fish. 音频 (S1 / S2-generation 模型) reuse cloned voices 通过 reference_id. 使用用户 prefers fish. audio/Fish 音频 TTS, 需求特定 playground 人声模型, 需要 high-emGenerate expressive, multilingual narration with fish.audio (S1 / S2-generation models) and reuse cloned voices via reference_id. Use when the user prefers fish.audio/Fish Audio TTS, wants a specific playground voice model, or needs high-em#machine-learningcalesthio/openmontageGit 克隆查看详情与安装步骤 → 机器学习56,383flux-best-practicesSkillsMP@calesthio全面指南 BFL FLUX 图像生成模型. 覆盖 prompting, T2I, I2I, 结构化 JSON, hex colors, 字体排印, 多-参考 editing, 模型-特定最佳实践 FLUX.2 FLUX.1 families.Comprehensive guide for BFL FLUX image generation models. Covers prompting, T2I, I2I, structured JSON, hex colors, typography, multi-reference editing, and model-specific best practices for FLUX.2 and FLUX.1 families.#machine-learningcalesthio/openmontageGit 克隆查看详情与安装步骤 → 机器学习49,057cli-hub-matrix-3d-cadSkillsMP@HKUDS能力-基于多-tool matrix 3D modeling, CAD, point 云, 渲染, GPU 调试, fabrication. 覆盖 mesh/parametric/photogrammetry 路径 idea printed 部分游戏-就绪 asset.Capability-based multi-tool matrix for 3D modeling, CAD, point clouds, rendering, GPU debugging, and fabrication. Covers mesh/parametric/photogrammetry and the path from idea to printed part or game-ready asset.#machine-learninghkuds/cli-anythingGit 克隆查看详情与安装步骤 → 机器学习49,057cli-hub-matrix-image-designSkillsMP@HKUDS能力-基于多-tool matrix 图片 graphic 设计: AI generation, raster/vector editing, UI mockups, 示意图, upscaling, photo 函数库, 发布.Capability-based multi-tool matrix for image and graphic design: AI generation, raster/vector editing, UI mockups, diagrams, upscaling, photo library, and publishing.#machine-learninghkuds/cli-anythingGit 克隆查看详情与安装步骤 → 机器学习49,057cli-anything-live2dSkillsMP@HKUDS检查, 校验, 编辑, lint, diff, 批量-管理, 部署 Live2D Cubism 模型 (. model3. json) 命令行.Inspect, validate, edit, lint, diff, batch-manage, and deploy Live2D Cubism models (.model3.json) from the command line.#machine-learninghkuds/cli-anythingGit 克隆查看详情与安装步骤 → 机器学习49,057cli-anything-live2dSkillsMP@HKUDS检查, 校验, 编辑, lint, diff, 批量-管理, 部署 Live2D Cubism 模型 (. model3. json) 命令行.Inspect, validate, edit, lint, diff, batch-manage, and deploy Live2D Cubism models (.model3.json) from the command line.#machine-learninghkuds/cli-anythingGit 克隆查看详情与安装步骤 → 机器学习48,134pixijs-blend-modesSkillsMP@pixijs使用技能 compositing 展示对象 blend modes PixiJS v8. 覆盖标准 modes (normal, 添加, multiply, 屏幕, erase, min, max), 先进 modes 通过 pixi. js/先进-blend-modes (color-burn, overlay, hard-light,.),Use this skill when compositing display objects with blend modes in PixiJS v8. Covers standard modes (normal, add, multiply, screen, erase, min, max), advanced modes via pixi.js/advanced-blend-modes (color-burn, overlay, hard-light, etc.), #machine-learningpixijs/pixijsGit 克隆查看详情与安装步骤 → 机器学习48,134pixijs-filtersSkillsMP@pixijs使用技能 applying 可视化 effects PixiJS v8 容器通过筛选流水线. 覆盖构建- 筛选 (AlphaFilter, BlurFilter, ColorMatrixFilter, DisplacementFilter, NoiseFilter), 自定义筛选.() GLSL/WGSL, 选项Use this skill when applying visual effects to PixiJS v8 containers via the filter pipeline. Covers built-in filters (AlphaFilter, BlurFilter, ColorMatrixFilter, DisplacementFilter, NoiseFilter), custom Filter.from() with GLSL/WGSL, options#machine-learningpixijs/pixijsGit 克隆查看详情与安装步骤 → 机器学习48,134pixijs-mathSkillsMP@pixijs使用技能处理 coordinates, vectors, matrices, shapes, hit testing, 布局 rectangles PixiJS v8. 覆盖 Point/ObservablePoint, Matrix (2D affine, decompose, 应用, applyInverse), shapes (Rectangle, Circle, Ellipse, PUse this skill when working with coordinates, vectors, matrices, shapes, hit testing, or layout rectangles in PixiJS v8. Covers Point/ObservablePoint, Matrix (2D affine, decompose, apply, applyInverse), shapes (Rectangle, Circle, Ellipse, P#machine-learningpixijs/pixijsGit 克隆查看详情与安装步骤 → 机器学习48,134pixijs-scene-meshSkillsMP@pixijs使用技能渲染自定义 geometry PixiJS v8. 覆盖 Mesh MeshGeometry (positions, uvs, 索引, topology), MeshSimple 按-frame vertex 动画, MeshPlane subdivided deformation, MeshRope 路径-以下文本Use this skill when rendering custom geometry in PixiJS v8. Covers Mesh with MeshGeometry (positions, uvs, indices, topology), MeshSimple for per-frame vertex animation, MeshPlane for subdivided deformation, MeshRope for path-following text#machine-learningpixijs/pixijsGit 克隆查看详情与安装步骤 → 机器学习44,353captions-overlaySkillsMP@heygen-comOverlay doctrine 嵌入式-配文工作流程 — 配文 MODEL (drop / rail / embed) 规则配文 OVERLAY composited 顶部影视, reserved 底部 band shift 内容 up 避免. 加载添加Overlay doctrine for the embedded-captions workflow — the caption MODEL (drop / rail / embed) and the rule that captions are an OVERLAY composited on top of the film, never a reserved bottom band you shift content up to avoid. Load when add#machine-learningheygen-com/hyperframesGit 克隆查看详情与安装步骤 → 机器学习44,353seam-craftSkillsMP@heygen-com渲染-correctness doctrine scene-to-scene seams HyperFrames 上线 videos — prerequisites 制作过渡动画 composite correctly 主分支时间线. 加载 assembling 主分支时间线 / index. html, white flasRender-correctness doctrine for scene-to-scene seams in HyperFrames launch videos — the prerequisites that make transitions composite correctly on the master timeline. Load when assembling the master timeline / index.html, when a white flas#machine-learningheygen-com/hyperframesGit 克隆查看详情与安装步骤 → 机器学习43,904aeonSkillsMP@K-Dense-AI技能用于时间 series 机器学习任务包括分类, 回归分析, 聚类, forecasting, anomaly 检测, segmentation, similarity 搜索. 使用处理 temporal 数据, sequential 模式This skill should be used for time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search. Use when working with temporal data, sequential patterns#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904arborSkillsMP@K-Dense-AIAutonomously 改进真实产物 (代码, 训练食谱, 智能体 harness, 数据流水线, 提示词) 对照 objective evaluator, 使用假设 Tree Refinement (HTR) Arbor 论文. 使用 someone 需求 iteratAutonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iterat#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904hypogenicSkillsMP@K-Dense-AI计划审计使用 ChicagoHAI HypoGeniC/HypoRefine LLM-assisted 假设 generation labeled 文本数据集. 使用 `hypogenic` 软件包, 任务 configs, 假设 banks, HypoBench 数据集— 手册假设Plans and audits use of ChicagoHAI HypoGeniC/HypoRefine for LLM-assisted hypothesis generation from labeled text datasets. Use for the `hypogenic` package, its task configs, hypothesis banks, or HypoBench datasets—not for manual hypothesis #machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904pymcSkillsMP@K-Dense-AIBayesian modeling PyMC. 构建 hierarchical 模型, MCMC (NUTS), variational 推理计算, LOO/WAIC comparison, posterior 检查, probabilistic programming 推理计算.Bayesian modeling with PyMC. Build hierarchical models, MCMC (NUTS), variational inference, LOO/WAIC comparison, posterior checks, for probabilistic programming and inference.#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904scikit-learnSkillsMP@K-Dense-AI机器学习 Python scikit-learn. 使用处理 supervised 学习 (分类, 回归分析), unsupervised 学习 (聚类, dimensionality reduction), 模型 evaluation, hyperparameter tuning, preprocessing, bMachine learning in Python with scikit-learn. Use when working with supervised learning (classification, regression), unsupervised learning (clustering, dimensionality reduction), model evaluation, hyperparameter tuning, preprocessing, or b#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904scikit-survivalSkillsMP@K-Dense-AI构建, evaluate, 审计正确-censored competing-风险 survival 工作流程 scikit-survival, 包括 leakage-安全 preprocessing, 模型选择, probability 预测, censoring-aware 指标.Build, evaluate, and audit right-censored or competing-risk survival workflows with scikit-survival, including leakage-safe preprocessing, model selection, probability prediction, and censoring-aware metrics.#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904scvi-toolsSkillsMP@K-Dense-AI深度 generative 模型单个-cell omics. 使用 need probabilistic 批量 correction (scVI), transfer 学习, differential expression uncertainty, 多-弹窗集成 (TOTALVI, MultiVI). Best 先进 modeling, bDeep generative models for single-cell omics. Use when you need probabilistic batch correction (scVI), transfer learning, differential expression with uncertainty, or multi-modal integration (TOTALVI, MultiVI). Best for advanced modeling, b#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904timesfm-forecastingSkillsMP@K-Dense-AIZero-shot 时间 series forecasting Google's TimesFM foundation 模型. 使用 univariate 时间 series (销售, 传感器, energy, vitals, weather) 无需训练自定义模型. 支持 CSV/DataFrame/数组输入 point forecastsZero-shot time series forecasting with Google's TimesFM foundation model. Use for any univariate time series (sales, sensors, energy, vitals, weather) without training a custom model. Supports CSV/DataFrame/array inputs with point forecasts#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904torch-geometricSkillsMP@K-Dense-AIPyTorch Geometric (PyG) 图谱 neural networks — node/link/graph 分类, 消息 passing (GCN, GAT, GraphSAGE, GIN), heterogeneous 图谱, neighbor sampling, 自定义数据集. 使用处理 torch_geometric, genPyTorch Geometric (PyG) for graph neural networks — node/link/graph classification, message passing (GCN, GAT, GraphSAGE, GIN), heterogeneous graphs, neighbor sampling, and custom datasets. Use when working with torch_geometric, not for gen#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904transformersSkillsMP@K-Dense-AIHugging Face Transformers 加载中心模型, running 流水线推理计算, 文本生成, Trainer 精细-tuning NLP, 视觉, 音频, multimodal 任务. 使用处理 AutoModel, 流水线, tokenizers, TrainingArgumentsHugging Face Transformers for loading Hub models, running pipeline inference, text generation, and Trainer fine-tuning on NLP, vision, audio, and multimodal tasks. Use when working with AutoModel, pipelines, tokenizers, or TrainingArguments#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,904umap-learnSkillsMP@K-Dense-AI使用 UMAP-learn nonlinear dimensionality reduction, 2D/3D 向量嵌入, 聚类 preprocessing, supervised semi-supervised UMAP, DensMAP, AlignedUMAP, Parametric UMAP 工作流程.Use UMAP-learn for nonlinear dimensionality reduction, 2D/3D embeddings, clustering preprocessing, supervised or semi-supervised UMAP, DensMAP, AlignedUMAP, and Parametric UMAP workflows.#machine-learningk-dense-ai/scientific-agent-skillsGit 克隆查看详情与安装步骤 → 机器学习43,720rebuildSkillsMP@ray-projectRebuild Ray 来源 — determines 正确构建模式基于 changedRebuild Ray from source — determines the right build mode based on what changed#machine-learningray-project/rayGit 克隆查看详情与安装步骤 → 机器学习40,300filters-and-postfxSkillsMP@phaserjs使用技能 applying 可视化筛选帖子-处理 effects Phaser 4. 覆盖 bloom, blur, glow, color matrix, barrel distortion, displacement, 自定义 shaders, 筛选流水线. 触发: 筛选, 帖子-处理, shadUse this skill when applying visual filters or post-processing effects in Phaser 4. Covers bloom, blur, glow, color matrix, barrel distortion, displacement, custom shaders, and the filter pipeline. Triggers on: filter, post-processing, shad#machine-learningphaserjs/phaserGit 克隆查看详情与安装步骤 → 机器学习40,300geometry-and-mathSkillsMP@phaserjs使用技能使用 Phaser 4 math geometry 实用工具. 覆盖 vectors, rectangles, circles, triangles, polygons, random number generation, angles, distance, interpolation, snapping. 触发: Vector2, Rectangle, Circle, mathUse this skill when using Phaser 4 math and geometry utilities. Covers vectors, rectangles, circles, triangles, polygons, random number generation, angles, distance, interpolation, and snapping. Triggers on: Vector2, Rectangle, Circle, math#machine-learningphaserjs/phaserGit 克隆查看详情与安装步骤 → 机器学习40,300tweensSkillsMP@phaserjs使用技能 animating 房产时间 Phaser 4. 覆盖 tweens, tween chains, easing 函数, stagger, yoyo, repeat, callbacks, number tweens, TweenManager. 触发: tween, ease, animate,. tweens. 添加, tweenUse this skill when animating properties over time in Phaser 4. Covers tweens, tween chains, easing functions, stagger, yoyo, repeat, callbacks, number tweens, and the TweenManager. Triggers on: tween, ease, animate, this.tweens.add, tween #machine-learningphaserjs/phaserGit 克隆查看详情与安装步骤 → 机器学习40,300v4-new-featuresSkillsMP@phaserjs使用技能学习新建功能, 游戏对象, 组件, 渲染能力新增 Phaser 4. 覆盖筛选, RenderNodes, CaptureFrame, Gradient, Noise, SpriteGPULayer, TilemapGPULayer, Lighting 组件, 渲染Use this skill when learning about new features, game objects, components, and rendering capabilities added in Phaser 4. Covers Filters, RenderNodes, CaptureFrame, Gradient, Noise, SpriteGPULayer, TilemapGPULayer, Lighting component, Render#machine-learningphaserjs/phaserGit 克隆查看详情与安装步骤 → 机器学习39,485spark-training-gotchasSkillsMP@wshobsonPreflight diagnose ten known 失败 modes ML 训练 NVIDIA DGX Spark. 使用训练运行 DGX Spark fails 启动, OOMs below 128GB 限量, slows down mid-运行, 在…之前多-hour 训练作业 GB10.Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485vector-index-tuningSkillsMP@wshobson优化 vector 索引 performance 延迟, recall, 记忆. 使用 tuning HNSW parameters, selecting quantization 策略, scaling vector 搜索基础设施.Optimize vector index performance for latency, recall, and memory. Use when tuning HNSW parameters, selecting quantization strategies, or scaling vector search infrastructure.#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485checkpoint-promotionSkillsMP@wshobsonGate 精细-tuned checkpoints drift budgets, paired comparison, forgetting 检查在…之前 promotion. 使用在…之后训练运行产出 checkpoint, deciding 是否 tuned 模型 ships, promoted 模型需要 re-gatingGate fine-tuned checkpoints with drift budgets, paired comparison, and forgetting checks before promotion. Use after a training run produces a checkpoint, when deciding whether a tuned model ships, or when a promoted model needs re-gating a#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485eval-harness-firstSkillsMP@wshobson构建 evaluation harness gates 精细-tuning 运行 — golden 集合, 按-失败-模式 graders, judge calibration, 基础-模型 baselines. 使用启动精细-tuning effort, converting 链路追踪 eval 集合, calBuild the evaluation harness that gates every fine-tuning run — golden sets, per-failure-mode graders, judge calibration, and base-model baselines. Use when starting a fine-tuning effort, when converting traces into an eval set, or when cal#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485finetuning-method-selectionSkillsMP@wshobsonDecide 是否精细-tune, 路由正确方法 (SFT, DPO/ORPO/KTO, GRPO/RLVR, continued pretraining) 基础模型. 使用启动精细-tuning effort, unsure 是否 RAG prompting suffice, chDecide whether to fine-tune at all, and route to the right method (SFT, DPO/ORPO/KTO, GRPO/RLVR, continued pretraining) and base model. Use when starting any fine-tuning effort, when unsure whether RAG or prompting would suffice, or when ch#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485grpo-rlvr-trainingSkillsMP@wshobsonTrain 推理 verifiable-任务行为 GRPO reinforcement 学习 verifiable rewards (RLVR). 使用任务成功 algorithmically checkable (math, 代码, tool 调用, 结构化输出), 设计 GRPO reward funTrain reasoning and verifiable-task behavior with GRPO and reinforcement learning from verifiable rewards (RLVR). Use when task success is algorithmically checkable (math, code, tool calls, structured output), when designing GRPO reward fun#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485lora-qlora-recipesSkillsMP@wshobson配置 LoRA QLoRA supervised 精细-tuning 当前 best-practice hyperparameters. 使用编写评审 LoRA/QLoRA 训练配置, 选择 rank/alpha/target 模块, deciding LoRA, QLoRA, 完整 finConfigure LoRA and QLoRA supervised fine-tuning with current best-practice hyperparameters. Use when writing or reviewing a LoRA/QLoRA training configuration, choosing rank/alpha/target modules, or deciding between LoRA, QLoRA, and full fin#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485preference-optimizationSkillsMP@wshobsonAlign 精细-tuned 模型 preference 数据使用 DPO, ORPO, KTO, SimPO. 使用 preference pairs thumbs-up/down feedback exist, 选择 preference-优化方法, DPO 运行需要 hyperparameters debuAlign a fine-tuned model with preference data using DPO, ORPO, KTO, or SimPO. Use when preference pairs or thumbs-up/down feedback exist, when choosing between preference-optimization methods, or when a DPO run needs hyperparameters or debu#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485quantized-exportSkillsMP@wshobson导出 promoted 精细-tuned 模型正确部署格式化 — merged safetensors, LoRA-, GGUF imatrix, FP8. 使用在…之后 checkpoint passes promotion, 选择 quantization 格式化目标设备, exporExport a promoted fine-tuned model in the right deployment format — merged safetensors, LoRA-only, GGUF with imatrix, or FP8. Use after a checkpoint passes promotion, when choosing a quantization format for a target device, or when an expor#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485trace-to-training-dataSkillsMP@wshobson转换 evaluation 链路追踪生产环境日志 SFT 示例 preference pairs. 使用 graded 链路追踪失败示例 exist need become 训练数据, applying rejection sampling 模型输出, 构建 DPOConvert evaluation traces and production logs into SFT examples and preference pairs. Use when graded traces or failure examples exist and need to become training data, when applying rejection sampling to model outputs, or when building DPO#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485vision-sftSkillsMP@wshobson精细-tune 视觉-语言模型 (VLMs) supervised 学习图片+文本数据. 使用 adapting VLM 可视化域名任务, 配置 frozen-视觉-tower LoRA, 调试 VLM 精细-tune trains 无需学习.Fine-tune vision-language models (VLMs) with supervised learning on image+text data. Use when adapting a VLM to a visual domain or task, configuring frozen-vision-tower LoRA, or debugging a VLM fine-tune that trains without learning.#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485ml-pipeline-workflowSkillsMP@wshobson构建端到端 MLOps 流水线数据 preparation 模型训练, 校验, 生产环境部署. 使用创建 ML 流水线, 实现 MLOps practices, automating 模型训练部署工作流程.Build end-to-end MLOps pipelines from data preparation through model training, validation, and production deployment. Use when creating ML pipelines, implementing MLOps practices, or automating model training and deployment workflows.#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,485recsys-pipeline-architectSkillsMP@wshobson设计 composable 推荐, ranking, feed 流水线使用 six-阶段来源→Hydrator→筛选→Scorer→选择器→SideEffect 框架 popularized xAI's 打开-sourced X 算法. 使用构建系统 picks "thDesign composable recommendation, ranking, and feed pipelines using the six-stage Source→Hydrator→Filter→Scorer→Selector→SideEffect framework popularized by xAI's open-sourced X For You algorithm. Use when building any system that picks "th#machine-learningwshobson/agentsGit 克隆查看详情与安装步骤 → 机器学习39,049self-improveSkillsMP@Yeachan-Heo自主 evolutionary 代码 improvement 引擎 tournament 选择Autonomous evolutionary code improvement engine with tournament selection#machine-learningyeachan-heo/oh-my-claudecodeGit 克隆查看详情与安装步骤 → 机器学习39,049traceSkillsMP@Yeachan-Heo证据-driven 链路追踪 lane orchestrates competing tracer hypotheses Claude 构建- 团队模式Evidence-driven tracing lane that orchestrates competing tracer hypotheses in Claude built-in team mode#machine-learningyeachan-heo/oh-my-claudecodeGit 克隆查看详情与安装步骤 → 机器学习38,754phoenix-evalsSkillsMP@github构建运行 evaluators AI/LLM 应用使用 Phoenix.Build and run evaluators for AI/LLM applications using Phoenix.#machine-learninggithub/awesome-copilotGit 克隆查看详情与安装步骤 → 机器学习38,754power-bi-model-design-reviewSkillsMP@github全面 Power BI 数据模型设计评审提示词 evaluating 模型架构, relationships, 优化 opportunities.Comprehensive Power BI data model design review prompt for evaluating model architecture, relationships, and optimization opportunities.#machine-learninggithub/awesome-copilotGit 克隆查看详情与安装步骤 → 机器学习38,754qdrant-model-migrationSkillsMP@github指南向量嵌入模型迁移 Qdrant 无需停机. 使用 someone 要求 '如何 switch 向量嵌入模型', '如何迁移 vectors', '如何更新新建模型', 'zero-停机模型变更', '如何 re-embed 数据', 'Guides embedding model migration in Qdrant without downtime. Use when someone asks 'how to switch embedding models', 'how to migrate vectors', 'how to update to a new model', 'zero-downtime model change', 'how to re-embed my data', or 'can #machine-learninggithub/awesome-copilotGit 克隆查看详情与安装步骤 → 机器学习38,754tm7-threat-modelSkillsMP@github创建有效 Microsoft Threat Modeling Tool (. tm7) 文件 compatible Microsoft Threat Modeling Tool v7.3+. 使用技能要求创建, 生成, 修改. tm7 威胁建模文件, performing STRIDE threat 模式Creates valid Microsoft Threat Modeling Tool (.tm7) files compatible with the Microsoft Threat Modeling Tool v7.3+. Use this skill whenever asked to create, generate, or modify a .tm7 threat model file, or when performing STRIDE threat mode#machine-learninggithub/awesome-copilotGit 克隆查看详情与安装步骤 → 机器学习37,406grpo-finetuneSkillsMP@patchy631精细-tune 模型 GRPO Fireworks-托管 GPUs plain-English 任务描述数据集. 使用技能用户需求精细-tune, RL-tune, GRPO-train 模型自有数据 — 说明 things 例如 "trainFine-tune a model with GRPO on Fireworks-managed GPUs from a plain-English task description and a dataset. Use this skill whenever the user wants to fine-tune, RL-tune, or GRPO-train a model on their own data — or says things like "train a #machine-learningpatchy631/ai-engineering-hubGit 克隆查看详情与安装步骤 → 机器学习37,406hugging-face-model-trainerSkillsMP@patchy631技能用于用户需要 train 精细-tune 语言模型使用 TRL (Transformer Reinforcement 学习) Hugging Face 作业基础设施. 覆盖 SFT, DPO, GRPO reward modeling 训练方法, 以及 GGUF converThis skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conver#machine-learningpatchy631/ai-engineering-hubGit 克隆查看详情与安装步骤 → 机器学习35,842compute-mamba-ratioSkillsMP@sgl-projectCompute optimal --mamba-完整-记忆-ratio ( --max-mamba-缓存-size pin) 混合 attention + linear-attention (Mamba / GDN / KDA) 模型's two serving 记忆 pools, workload serving 配置. 使用用户要求 raCompute the optimal --mamba-full-memory-ratio (or --max-mamba-cache-size pin) for a hybrid attention + linear-attention (Mamba / GDN / KDA) model's two serving memory pools, from the workload and serving config. Use when a user asks what ra#machine-learningsgl-project/sglangGit 克隆查看详情与安装步骤 → 机器学习35,842sglang-diffusion-modelopt-quantSkillsMP@sgl-project使用 quantizing diffusion DiT NVIDIA ModelOpt 制作生成 FP8 NVFP4 checkpoint loadable, verifiable, benchmarkable SGLang Diffusion.Use when quantizing a diffusion DiT with NVIDIA ModelOpt and making the resulting FP8 or NVFP4 checkpoint loadable, verifiable, and benchmarkable in SGLang Diffusion.#machine-learningsgl-project/sglangGit 克隆查看详情与安装步骤 → 机器学习35,842sglang-diffusion-performanceSkillsMP@sgl-project使用选择 fastest SGLang Diffusion flags 模型, GPU, VRAM 预算.Use when choosing the fastest SGLang Diffusion flags for a model, GPU, and VRAM budget.#machine-learningsgl-project/sglangGit 克隆查看详情与安装步骤 → 机器学习35,034ctf-sandboxSkillsMP@zhaoxuya520Thin PRIMARY CTF / AWD / 靶场多-type orchestration. Hands off sidecar CTF-沙箱-Orchestrator. 使用用户说明 CTF, AWD, 靶场, 比赛题特定 pwn/APK/IDA 路由 already won.Thin PRIMARY for CTF / AWD / 靶场 multi-type orchestration. Hands off to the sidecar CTF-Sandbox-Orchestrator. Use when the user says CTF, AWD, 靶场, or 比赛题 and no more specific pwn/APK/IDA route already won.#machine-learningzhaoxuya520/reverse-skillGit 克隆查看详情与安装步骤 → 机器学习34,964llmfit-advisorSkillsMP@AlexsJones检测本地硬件 (RAM, CPU, GPU/VRAM) 推荐 best-fit 本地 LLM 模型 optimal quantization, 提速 estimates, fit scoring.Detect local hardware (RAM, CPU, GPU/VRAM) and recommend the best-fit local LLM models with optimal quantization, speed estimates, and fit scoring.#machine-learningalexsjones/llmfitGit 克隆查看详情与安装步骤 → 机器学习34,450model-integrationSkillsMP@huggingface使用新增新建模型流水线 diffusers, 设置项 up 文件结构新建模型, converting 流水线模块化格式化, converting weights 新建版本 already-支持模型.Use when adding a new model or pipeline to diffusers, setting up file structure for a new model, converting a pipeline to modular format, or converting weights for a new version of an already-supported model.#machine-learninghuggingface/diffusersGit 克隆查看详情与安装步骤 → 机器学习34,264add-model-priceSkillsMP@langfuse使用 editing worker/src/constants/default-model-prices. json, packages/shared/src/server/llm/types. ts, pricing tiers, tokenizer IDs, matchPattern regexes OpenAI, Anthropic, Bedrock, Vertex, Azure, Gemini 模型 pricing.Use when editing worker/src/constants/default-model-prices.json, packages/shared/src/server/llm/types.ts, pricing tiers, tokenizer IDs, or matchPattern regexes for OpenAI, Anthropic, Bedrock, Vertex, Azure, or Gemini model pricing.#machine-learninglangfuse/langfuseGit 克隆查看详情与安装步骤 → 机器学习33,024candlestickSkillsMP@HKUDSCandlestick 模式 recognition 引擎, pure pandas vectorized 实现 15 classic candlestick 模式 (5 单个-candle + 5 double-candle + 4 triple-candle + 1 trend confirmation), 生成 composite signal bullish/bearisCandlestick pattern recognition engine, pure pandas vectorized implementation of 15 classic candlestick patterns (5 single-candle + 5 double-candle + 4 triple-candle + 1 trend confirmation), generating a composite signal from bullish/bearis#machine-learninghkuds/vibe-tradingGit 克隆查看详情与安装步骤 → 机器学习33,024elliott-waveSkillsMP@HKUDSElliott Wave Theory signal 引擎. 检测 swing points Zigzag, matches 5-wave impulse 3-wave corrective structures, 校验 Fibonacci wave relationships, 生成 trend-顶部 / correction-完整 signals. PureElliott Wave Theory signal engine. Detects swing points through Zigzag, matches 5-wave impulse and 3-wave corrective structures, validates them with Fibonacci wave relationships, and generates trend-top / correction-complete signals. Pure i#machine-learninghkuds/vibe-tradingGit 克隆查看详情与安装步骤 → 机器学习33,024harmonicSkillsMP@HKUDSHarmonic 模式 signal 引擎. 识别 XABCD five-point structures Gartley/Bat/Butterfly/Crab 基于 Fibonacci geometry, 生成交易 signals PRZ (Potential Reversal Zone).Harmonic Patterns signal engine. Identifies XABCD five-point structures such as Gartley/Bat/Butterfly/Crab based on Fibonacci geometry, and generates trading signals in the PRZ (Potential Reversal Zone).#machine-learninghkuds/vibe-tradingGit 克隆查看详情与安装步骤 → 机器学习33,024ichimokuSkillsMP@HKUDSIchimoku Kinko Hyo five-line 系统 signal 引擎. standalone Japanese technical-分析 school 生成交易 signals Tenkan/Kijun crossovers, 云端 position, Chikou confirmation. Pure pandas 实现.Ichimoku Kinko Hyo five-line system signal engine. A standalone Japanese technical-analysis school that generates trading signals from Tenkan/Kijun crossovers, cloud position, and Chikou confirmation. Pure pandas implementation.#machine-learninghkuds/vibe-tradingGit 克隆查看详情与安装步骤 → 机器学习33,024ml-strategySkillsMP@HKUDSMachine-学习 predictive 策略基于 sklearn walk-forward 训练, 功能工程, signal generation. Suitable OHLCV 数据.Machine-learning predictive strategy based on sklearn walk-forward training, feature engineering, and signal generation. Suitable for any OHLCV data.#machine-learninghkuds/vibe-tradingGit 克隆查看详情与安装步骤 → 机器学习32,416feynman-learningSkillsMP@THU-MAIC转换 concept, 课时, slide 演示文稿, 来源 material Feynman 学习 cycle learners explain 首先, expose smallest gap, rebuild explanation Socratic 提示词, strip jargon, stress-测试 analogies, tranTurn any concept, lesson, slide deck, or source material into a Feynman learning cycle in which learners explain first, expose the smallest gap, rebuild the explanation through Socratic prompts, strip jargon, stress-test analogies, and tran#machine-learningthu-maic/openmaicGit 克隆查看详情与安装步骤 → 机器学习32,390detecting-data-and-model-poisoningSkillsMP@mukul975识别 poisoned 训练数据 backdoored ML 模型跨 流水线使用 IBM's Adversarial Robustness Toolbox (activation 聚类, spectral 签名, 触发 reconstruction), Cleanlab 标签-质量工单, supply-chaiIdentify poisoned training data and backdoored ML models across the pipeline using IBM's Adversarial Robustness Toolbox (activation clustering, spectral signatures, trigger reconstruction), Cleanlab for label-quality issues, and supply-chai#machine-learningmukul975/anthropic-cybersecurity-skillsGit 克隆查看详情与安装步骤 → 机器学习32,390detecting-deepfake-audio-in-vishing-attacksSkillsMP@mukul975检测 AI-生成 deepfake 音频用于人声钓鱼攻击 (vishing) extracting spectral 功能 (MFCC, spectral centroid, spectral contrast, zero-crossing rate) classifying 样例机器学习模型, 支持批量 audiDetect AI-generated deepfake audio used in voice phishing (vishing) by extracting spectral features (MFCC, spectral centroid, spectral contrast, zero-crossing rate) and classifying samples with machine learning models, supporting batch audi#machine-learningmukul975/anthropic-cybersecurity-skillsGit 克隆查看详情与安装步骤 → 机器学习32,390implementing-cisa-zero-trust-maturity-modelSkillsMP@mukul975Assess, gap-分析, progressively 实现 CISA Zero Trust Maturity 模型 v2.0 跨 five pillars (Identity, 设备, Networks, 应用 & Workloads, 数据) 三个 cross-cutting 能力 (Visibility/Analytics, AutomatioAssess, gap-analyze, and progressively implement the CISA Zero Trust Maturity Model v2.0 across five pillars (Identity, Devices, Networks, Applications & Workloads, Data) and three cross-cutting capabilities (Visibility/Analytics, Automatio#machine-learningmukul975/anthropic-cybersecurity-skillsGit 克隆查看详情与安装步骤 → 机器学习32,284timesfm-forecastingSkillsMP@google-researchZero-shot 时间 series forecasting Google's TimesFM foundation 模型. 使用技能 forecasting ANY univariate 时间 series — 销售, 传感器 readings, 股票 prices, energy demand, 患者 vitals, weather, scientific measurementsZero-shot time series forecasting with Google's TimesFM foundation model. Use this skill when forecasting ANY univariate time series — sales, sensor readings, stock prices, energy demand, patient vitals, weather, or scientific measurements #machine-learninggoogle-research/timesfmGit 克隆查看详情与安装步骤 → 机器学习30,777gws-modelarmor-sanitize-responseSkillsMP@googleworkspaceGoogle 模型 Armor: Sanitize 模型响应模型 Armor 模板.Google Model Armor: Sanitize a model response through a Model Armor template.#machine-learninggoogleworkspace/cliGit 克隆查看详情与安装步骤 → 机器学习30,563nemo-curatorSkillsMP@davila7GPU-accelerated 数据 curation LLM 训练. 支持 text/image/video/audio. 功能 fuzzy deduplication (16× 更快), 质量筛选 (30+ heuristics), 语义 deduplication, PII redaction, NSFW 检测. Scales 跨 GPUs RGPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with R#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563training-llms-megatronSkillsMP@davila7Trains 大语言模型 (2B-462B parameters) 使用 NVIDIA Megatron-核心先进 parallelism 策略. 使用训练模型 >1B parameters, need maximum GPU 效率 (47% MFU H100), 需要 tensor/pipeline/sequence/cTrains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/c#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563pytorch-fsdpSkillsMP@davila7专家指引 Fully Sharded 数据并行训练 PyTorch FSDP - parameter sharding, mixed precision, CPU offloading, FSDP2Expert guidance for Fully Sharded Data Parallel training with PyTorch FSDP - parameter sharding, mixed precision, CPU offloading, FSDP2#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563ray-trainSkillsMP@davila7分布式训练 orchestration 跨集群. Scales PyTorch/TensorFlow/HuggingFace laptop 1000s 节点. 构建- hyperparameter tuning Ray Tune, fault tolerance, elastic scaling. 使用训练 massive 模型跨Distributed training orchestration across clusters. Scales PyTorch/TensorFlow/HuggingFace from laptop to 1000s of nodes. Built-in hyperparameter tuning with Ray Tune, fault tolerance, elastic scaling. Use when training massive models across#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563knowledge-distillationSkillsMP@davila7Compress 大语言模型使用 knowledge distillation 教师学生模型. 使用 deploying smaller 模型 retained performance, transferring GPT-4 能力开源模型, reducing 推理计算成本. CovCompress large language models using knowledge distillation from teacher to student models. Use when deploying smaller models with retained performance, transferring GPT-4 capabilities to open-source models, or reducing inference costs. Cov#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563long-contextSkillsMP@davila7Extend 上下文 Windows transformer 模型使用 RoPE, YaRN, ALiBi, position interpolation techniques. 使用处理 long 文档 (32k-128k+ 词元), extending pre-训练好的模型 beyond original 上下文 limits, 实现Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques. Use when processing long documents (32k-128k+ tokens), extending pre-trained models beyond original context limits, or implementing#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563model-mergingSkillsMP@davila7Merge 多个精细-tuned 模型使用 mergekit combine 能力无需 retraining. 使用创建 specialized 模型 blending 域名-特定 expertise (math + 编码 + 聊天), improving performance beyond 单个模型, exMerge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or ex#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563model-pruningSkillsMP@davila7Reduce LLM size 加速推理计算使用 pruning techniques 例如 Wanda SparseGPT. 使用 compressing 模型无需 retraining, achieving 50% sparsity minimal accuracy loss, enabling 更快推理计算硬件 acceleraReduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or enabling faster inference on hardware accelera#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563moe-trainingSkillsMP@davila7Train Mixture Experts (MoE) 模型使用 DeepSpeed HuggingFace. 使用训练 large-scale 模型 limited compute (5× 成本 reduction vs dense 模型), 实现 sparse architectures 例如 Mixtral 8x7B DeepSeek-V3, scalTrain Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scal#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563speculative-decodingSkillsMP@davila7加速 LLM 推理计算使用 speculative decoding, Medusa 多个 heads, lookahead decoding techniques. 使用优化推理计算提速 (1.5-3.6× speedup), reducing 延迟实时应用, deploying 模型 limAccelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with lim#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563nemo-evaluator-sdkSkillsMP@davila7Evaluates LLMs 跨 100+ 基准测试 18+ harnesses (MMLU, HumanEval, GSM8K, 安全, VLM) 多-后端执行. 使用 needing 可扩展 evaluation 本地 Docker, Slurm HPC, 云端平台. NVIDIA's enterprise-grade platEvaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing scalable evaluation on local Docker, Slurm HPC, or cloud platforms. NVIDIA's enterprise-grade plat#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563axolotlSkillsMP@davila7专家指引精细-tuning LLMs Axolotl - YAML configs, 100+ 模型, LoRA/QLoRA, DPO/KTO/ORPO/GRPO, multimodal 支持Expert guidance for fine-tuning LLMs with Axolotl - YAML configs, 100+ models, LoRA/QLoRA, DPO/KTO/ORPO/GRPO, multimodal support#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563peft-fine-tuningSkillsMP@davila7Parameter-高效精细-tuning LLMs 使用 LoRA, QLoRA, 25+ 方法. 使用精细-tuning large 模型 (7B-70B) limited GPU 记忆, need train <1% parameters minimal accuracy loss, 多-适配器 serviParameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter servi#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563unslothSkillsMP@davila7专家指引快速精细-tuning Unsloth - 2-5x 更快训练, 50-80% less 记忆, LoRA/QLoRA 优化Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563sglangSkillsMP@davila7快速结构化 generation serving LLMs RadixAttention prefix 缓存. 使用 JSON/正则表达式输出, constrained decoding, agentic 工作流程 tool 调用, need 5× 更快推理计算 vLLM prefix sharing. PowFast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Pow#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563lambda-labs-gpu-cloudSkillsMP@davila7Reserved on-demand GPU 云端 instances ML 训练推理计算. 使用 need 专属 GPU instances 简单 SSH 访问, 持久化 filesystems, high-performance 多-Node 集群 large-scale 训练.Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563transformer-lens-interpretabilitySkillsMP@davila7提供指引 mechanistic interpretability 研究使用 TransformerLens 检查 manipulate transformer internals 通过 HookPoints activation 缓存. 使用 reverse-工程模型算法, studying attention patteProvides guidance for mechanistic interpretability research using TransformerLens to inspect and manipulate transformer internals via HookPoints and activation caching. Use when reverse-engineering model algorithms, studying attention patte#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563implementing-llms-litgptSkillsMP@davila7实现 trains LLMs 使用 Lightning AI's LitGPT 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral). 使用 need 整洁模型 implementations, 教育 understanding architectures, 生产环境精细-tuning wiImplements and trains LLMs using Lightning AI's LitGPT with 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral). Use when need clean model implementations, educational understanding of architectures, or production fine-tuning wi#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563mamba-architectureSkillsMP@davila7状态-space 模型 O(n) complexity vs Transformers' O(n²). 5× 更快推理计算, million-词元 sequences, KV 缓存. Selective SSM 硬件-aware 设计. Mamba-1 (d_state=16) Mamba-2 (d_state=128, 多-head). 模型 130M-2.8BState-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563rwkv-architectureSkillsMP@davila7RNN+Transformer 混合 O(n) 推理计算. Linear 时间, infinite 上下文, KV 缓存. Train 例如 GPT (并行), infer 例如 RNN (sequential). Linux Foundation AI 项目. 生产环境 Windows, Office, NeMo. RWKV-7 (March 2025). 模型 upRNN+Transformer hybrid with O(n) inference. Linear time, infinite context, no KV cache. Train like GPT (parallel), infer like RNN (sequential). Linux Foundation AI project. Production at Windows, Office, NeMo. RWKV-7 (March 2025). Models up#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563clipSkillsMP@davila7OpenAI's 模型 connecting 视觉语言. 启用 zero-shot 图片分类, 图片-文本匹配, cross-弹窗 retrieval. 训练好的 400M 图片-文本 pairs. 使用图片搜索, 内容 moderation, 视觉-语言任务OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pairs. Use for image search, content moderation, or vision-language tasks with#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563segment-anything-modelSkillsMP@davila7Foundation 模型图片 segmentation zero-shot transfer. 使用 need segment 对象图片使用 points, boxes, masks 提示词, 自动地生成对象 masks 图片.Foundation model for image segmentation with zero-shot transfer. Use when you need to segment any object in images using points, boxes, or masks as prompts, or automatically generate all object masks in an image.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563quantizing-models-bitsandbytesSkillsMP@davila7Quantizes LLMs 8-bit 4-bit 50-75% 记忆 reduction minimal accuracy loss. 使用 GPU 记忆 limited, need fit larger 模型, 需要更快推理计算. 支持 INT8, NF4, FP4 格式, QLoRA 训练, 8-bit optimizQuantizes LLMs to 8-bit or 4-bit for 50-75% memory reduction with minimal accuracy loss. Use when GPU memory is limited, need to fit larger models, or want faster inference. Supports INT8, NF4, FP4 formats, QLoRA training, and 8-bit optimiz#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563gptqSkillsMP@davila7帖子-训练 4-bit quantization LLMs minimal accuracy loss. 使用 deploying large 模型 (70B, 405B) 消费方 GPUs, need 4× 记忆 reduction <2% perplexity degradation, 更快推理计算 (3-4× speedup) vsPost-training 4-bit quantization for LLMs with minimal accuracy loss. Use for deploying large models (70B, 405B) on consumer GPUs, when you need 4× memory reduction with <2% perplexity degradation, or for faster inference (3-4× speedup) vs #machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563grpo-rl-trainingSkillsMP@davila7专家指引 GRPO/RL 精细-tuning TRL 推理任务-特定模型训练Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563miles-rl-trainingSkillsMP@davila7提供指引 enterprise-grade RL 训练使用 miles, 生产环境-就绪 fork slime. 使用训练 large MoE 模型 FP8/INT4, needing train-推理计算 alignment, 需要 speculative RL maximum 吞吐量.Provides guidance for enterprise-grade RL training using miles, a production-ready fork of slime. Use when training large MoE models with FP8/INT4, needing train-inference alignment, or requiring speculative RL for maximum throughput.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563openrlhf-trainingSkillsMP@davila7High-performance RLHF 框架 Ray+vLLM acceleration. 使用 PPO, GRPO, RLOO, DPO 训练 large 模型 (7B-70B+). 构建 Ray, vLLM, ZeRO-3. 2× 更快 DeepSpeedChat 分布式架构 GPU 资源 sharing.High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563simpo-trainingSkillsMP@davila7简单 Preference 优化 LLM alignment. 参考-free alternative DPO better performance (+6.4 points AlpacaEval 2.0). 参考模型所需, 高效 DPO. 使用 preference alignment 需要更简单,Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model needed, more efficient than DPO. Use for preference alignment when want simpler,#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563slime-rl-trainingSkillsMP@davila7提供指引 LLM 帖子-训练 RL 使用 slime, Megatron+SGLang 框架. 使用训练 GLM 模型, 实现自定义数据 generation 工作流程, needing tight Megatron-LM 集成 RL scaling.Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563fine-tuning-with-trlSkillsMP@davila7精细-tune LLMs 使用 reinforcement 学习 TRL - SFT 操作说明 tuning, DPO preference alignment, PPO/GRPO reward 优化, reward 模型训练. 使用 need RLHF, align 模型 preferences, train huFine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward model training. Use when need RLHF, align model with preferences, or train from hu#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563verl-rl-trainingSkillsMP@davila7提供指引训练 LLMs reinforcement 学习使用 verl (Volcano 引擎 RL). 使用实现 RLHF, GRPO, PPO, 其他 RL 算法 LLM 帖子-训练 scale 灵活基础设施 backends.Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563prompt-engineering-patternsSkillsMP@davila7主分支先进提示词工程 techniques maximize LLM performance, 可靠性, controllability.Master advanced prompt engineering techniques to maximize LLM performance, reliability, and controllability.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563rag-engineerSkillsMP@davila7专家构建 Retrieval-Augmented Generation 系统. Masters 向量嵌入模型, vector 数据库, chunking 策略, retrieval 优化 LLM 应用. 使用: 构建 RAG, vector 搜索, 向量嵌入, 语义搜索,Expert in building Retrieval-Augmented Generation systems. Masters embedding models, vector databases, chunking strategies, and retrieval optimization for LLM applications. Use when: building RAG, vector search, embeddings, semantic search,#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563sentence-transformersSkillsMP@davila7框架状态-of-the-art sentence, 文本, 图片向量嵌入. 提供 5000+ pre-训练好的模型语义 similarity, 聚类, retrieval. 支持多语言, 域名-特定, multimodal 模型. 使用生成 embFramework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Supports multilingual, domain-specific, and multimodal models. Use for generating emb#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563constitutional-aiSkillsMP@davila7Anthropic's 方法训练 harmless AI self-improvement. Two-阶段 approach - supervised 学习 self-critique/revision, RLAIF (RL AI Feedback). 使用安全 alignment, reducing harmful 输出无需 human lAnthropic's method for training harmless AI through self-improvement. Two-phase approach - supervised learning with self-critique/revision, then RLAIF (RL from AI Feedback). Use for safety alignment, reducing harmful outputs without human l#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563llamaguardSkillsMP@davila7Meta's 7-8B specialized moderation 模型 LLM input/output 筛选. 6 安全分类 - violence/hate, sexual 内容, weapons, substances, self-harm, criminal 规划. 94-95% accuracy. 部署 vLLM, HuggingFace, Sagemaker. InteMeta's 7-8B specialized moderation model for LLM input/output filtering. 6 safety categories - violence/hate, sexual content, weapons, substances, self-harm, criminal planning. 94-95% accuracy. Deploy with vLLM, HuggingFace, Sagemaker. Inte#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563huggingface-tokenizersSkillsMP@davila7快速 tokenizers 已优化研究生产环境. Rust-基于实现 tokenizes 1GB <20 seconds. 支持 BPE, WordPiece, Unigram 算法. Train 自定义 vocabularies, 跟踪 alignments, 处理 padding/truncation. IntegratFast tokenizers optimized for research and production. Rust-based implementation tokenizes 1GB in <20 seconds. Supports BPE, WordPiece, and Unigram algorithms. Train custom vocabularies, track alignments, handle padding/truncation. Integrat#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563sentencepieceSkillsMP@davila7语言-independent tokenizer treating 文本 raw Unicode. 支持 BPE Unigram 算法. 快速 (50k sentences/sec), 轻量 (6MB 记忆), deterministic vocabulary. 用于 T5, ALBERT, XLNet, mBART. Train raw 文本无需 pre-Language-independent tokenizer treating text as raw Unicode. Supports BPE and Unigram algorithms. Fast (50k sentences/sec), lightweight (6MB memory), deterministic vocabulary. Used by T5, ALBERT, XLNet, mBART. Train on raw text without pre-#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563angularSkillsMP@davila7现代 Angular (v20+) 专家深度 knowledge Signals, Standalone 组件, Zoneless 应用, SSR/Hydration, 响应式模式.Modern Angular (v20+) expert with deep knowledge of Signals, Standalone Components, Zoneless applications, SSR/Hydration, and reactive patterns.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563golang-proSkillsMP@davila7主分支 Go 1.21+ 现代模式, 先进 concurrency, 性能优化, 生产环境-就绪微服务.Master Go 1.21+ with modern patterns, advanced concurrency, performance optimization, and production-ready microservices.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563haskell-proSkillsMP@davila7专家 Haskell engineer specializing 先进 type 系统, pureExpert Haskell engineer specializing in advanced type systems, pure#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563java-proSkillsMP@davila7主分支 Java 21+ 现代功能例如 virtual 线程, 模式匹配, Spring Boot 3. x. 专家最新 Java 生态包括 GraalVM, 项目 Loom, 云端-原生模式.Master Java 21+ with modern features like virtual threads, pattern matching, and Spring Boot 3.x. Expert in the latest Java ecosystem including GraalVM, Project Loom, and cloud-native patterns.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 → 机器学习30,563rust-proSkillsMP@davila7主分支 Rust 1.75+ 现代异步模式, 先进 type 系统功能, 生产环境-就绪系统 programming.Master Rust 1.75+ with modern async patterns, advanced type system features, and production-ready systems programming.#machine-learningdavila7/claude-code-templatesGit 克隆查看详情与安装步骤 →