# Deck LLM

> Machine learning for deck analysis using PyTorch neural networks Use when this capability is needed.

- Skill: `tomevault-io/deck-llm` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add tomevault-io/deck-llm`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/deck-llm/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: tomevault-io (https://skillmd.com/u/tomevault-io)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/tomevault-io/deck-llm

---

<!-- REFERENCE: not a workflow skill — documentation for ML utilities, not an instruction to follow -->
# Deck LLM (Language Model)

- **Purpose**
  
  Machine learning utilities for Yu-Gi-Oh! deck analysis and modeling using PyTorch. Provides neural network components, card embeddings, and training examples for deck-based language models.

- **Core Components**

  - **[nn.py](nn.py)** — Neural network utilities and building blocks
    - Custom PyTorch modules and tensor operations
    - Attention mechanisms (Flash Attention support)
    - Embedding functions with Qwen3-Embedding integration
    - PCA dimensionality reduction
    - Various activation functions (SwiGLU, SoftmaxGLU)
    
  - **[ygonn.py](ygonn.py)** — YGO-specific neural network features
    - Card database integration (SQLite cards.cdb)
    - Card embedding generation
    - Deck loading and preprocessing
    - Feature extraction for cards
    - Bindings and aliases management
    
  - **[bindx5c.py](bindx5c.py)** — Bind card ID prediction model
    - Lightning-based training example
    - Predicts 5-card bind combinations
    - Uses attention mechanism for sequence prediction
    
  - **[open.py](open.py)** — Open vocabulary model
    - Alternative model architecture
    - Card feature embedding model
    - Training example for deck generation

- **Requirements**

  - PyTorch with CUDA support
  - Flash Attention (optional, for performance)
  - Lightning AI framework
  - SQLite database (cards.cdb)
  - YGOPro installation with card data
  
- **Usage Examples**

  ```python
  from ygonn import cardsfeat, decks
  
  # Load card features
  features = cardsfeat()
  
  # Load deck data
  deck_tensors = decks()
  
  # Train model (example)
  model = Model()
  trainer = L.Trainer(precision=16)
  trainer.fit(model, DataLoader(decks(), ...))
  ```

- **Integration Skills**

  Used by:
  - `ygo/card-build/SKILL.md` — For ML-assisted card design
  - `ygo/deck-compare/SKILL.md` — For deck pattern recognition
  - `ygopro/ranking.md` — For meta trend prediction
  
- **Customization**

  - Modify embedding models in `ygonn.py`
  - Adjust model architectures in examples
  - Add custom training objectives
  - Integrate with other YGO data sources

- **Performance Notes**

  - Models benefit from GPU acceleration
  - Use mixed precision (FP16) for training
  - Flash Attention significantly speeds up attention layers
  - PCA reduces feature dimensionality for efficiency

---
> Source: [ZisIsNotZis/ygoskill](https://github.com/ZisIsNotZis/ygoskill) — distributed by [TomeVault](https://tomevault.io).
<!-- tomevault:4.0:skill_md:2026-05-22 -->