Agent skill · Design & Presentation

plan-mode-advanced

Create and execute advanced execution plans for complex AI model development incorporating 2024-2026 cutting-edge techniques (DeepSeek GRPO, manifold-constrained architectures, geometric scaling). Use when planning large-scale model training, architecture optimization, or multi-stage development workflows requiring state-of-the-art methodologies.

majiayu000github.com/majiayu000GitHub ↗
claude-codeMIT
Install
npx skills add majiayu000/claude-skill-registry --skill plan-mode-advanced-zapabob-so8t --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

Facts
Files in the skill folder: 2
SKILL.md size: 10 KB
Bundled scripts: none
Path: skills/ai-ml/plan-mode-advanced-zapabob-so8t/SKILL.md
Open the folder on GitHub →
Where it comes from
Stars: 534
Language: HTML

Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.

From the SKILL.md

# 高度なPlanモード: 2024-2026最先端手法統合 ## 概要 このスキルは、2024-2026年の最先端AI手法(DeepSeek GRPO、mHC多様体アーキテクチャ、幾何学的スケーリング)を統合した高度な実行計画を作成・実行します。複雑なAIモデル開発において、体系的かつ効率的な計画立案と実行を支援します。 ## 統合手法 ### 1. DeepSeek-R1 GRPO (Group Relative Policy Optimization) **論文**: "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning" (2025) **適用**: - 純粋RLベースの推論能力育成 - 人間の推論トレースなしで創発的推論行動を実現 - マルチステージ訓練: Cold-start SFT → GRPO → Rejection Sampling → All-scenarios RL ### 2. mHC (Manifold-Constrained Hyper-Connections) **論文**: "mHC: Manifold-Constrained Hyper-Connections" (2025) **適用**: - Birkhoff多様体上の二重確率行列制約 - 残差ストリームの安定性確保 - Sinkhorn-Knopp正規化による恒等写像保存 ### 3. 幾何学的スケーリングと動的スケーリング **論文**: "Geometric and Dynamic Scaling in Deep Transformers" (2026) **適用**: - 意味的多様体からのドリフト防止 - 非単調デルタ学習による冗長特徴消去 - 多様体制約付き残差更新 ## 計画作成ワークフロー ### Phase 1: 要件分析と手法選定 ```yaml 計画要件分析: 目標モデル規模: [7B, 13B, 27B, 70B] 対象タスク: [推論, 知識, コード生成, 多言語] 制約条件: [計算リソース, 時間, データ可用性] 最先端手法統合: [GRPO, mHC, 幾何学的スケーリング] ``` ### Phase 2: アーキテクチャ設計計画 #### mHC統合アーキテクチャ ```python class MHCTransformerBlock(nn.Module): def __init__(self, config): super().__init__() # 標準Transformerブロック self.attention = MultiHeadAttention(config) self.mlp = MLP(config)

What's inside
Steps it walks through
  1. 概要
  2. 統合手法
  3. 1. DeepSeek-R1 GRPO (Group Relative Policy Optimization)
  4. 2. mHC (Manifold-Constrained Hyper-Connections)
  5. 3. 幾何学的スケーリングと動的スケーリング
  6. 計画作成ワークフロー
  7. Phase 1: 要件分析と手法選定
  8. Phase 2: アーキテクチャ設計計画
  9. Phase 3: 訓練戦略計画
  10. Phase 4: 評価と検証計画
  11. 実行管理
  12. リアルタイム進捗監視
  13. エラーハンドリングと回復
  14. 高度な最適化手法
Ships with 1 file
  • metadata.json
Commands it runs
python scripts/plan_mode/create_advanced_plan.py \
python scripts/plan_mode/optimize_resources.py \
python scripts/plan_mode/monitor_execution.py \
python scripts/plan_mode/adaptive_optimization.py \
More from claude-skill-registry
All skills →
About this skill
What does the plan-mode-advanced skill do?

Create and execute advanced execution plans for complex AI model development incorporating 2024-2026 cutting-edge techniques (DeepSeek GRPO, manifold-constrained architectures, geometric scaling). Use when planning large-scale model training, architecture optimization, or multi-stage development workflows requiring state-of-the-art methodologies.

How do I install it?

Run `npx skills add majiayu000/claude-skill-registry --skill plan-mode-advanced-zapabob-so8t --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.

Where does this skill come from?

From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.

Is a popular skill a good skill?

Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.

Keep going