# Neural Training

> > Neural 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: simple tasks, no learning required, one-off operations.

- Skill: `thedixitjain/neural-training` (Agent Skill)
- Install (CLI): `npx skillmds add thedixitjain/neural-training`
- Raw SKILL.md: https://api.skillmd.com/api/skills/thedixitjain/neural-training/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: thedixitjain (https://skillmd.com/u/thedixitjain)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/thedixitjain/neural-training

---



# Neural Training Skill

## Purpose
Train and optimize neural patterns using SONA, MoE, and EWC++ systems.

## When to Trigger
- Training new patterns
- Optimizing agent routing
- Knowledge consolidation
- Pattern recognition tasks

## Intelligence Pipeline

1. **RETRIEVE** — Fetch relevant patterns via HNSW (150x-12,500x faster)
2. **JUDGE** — Evaluate with verdicts (success$failure)
3. **DISTILL** — Extract key learnings via LoRA
4. **CONSOLIDATE** — Prevent catastrophic forgetting via EWC++

## Components

| Component | Purpose | Performance |
|-----------|---------|-------------|
| SONA | Self-optimizing adaptation | <0.05ms |
| MoE | Expert routing | 8 experts |
| HNSW | Pattern search | 150x-12,500x |
| EWC++ | Prevent forgetting | Continuous |
| Flash Attention | Speed | 2.49x-7.47x |

## Commands

### Train Patterns
```bash
npx claude-flow neural train --model-type moe --epochs 10
```

### Check Status
```bash
npx claude-flow neural status
```

### View Patterns
```bash
npx claude-flow neural patterns --type all
```

### Predict
```bash
npx claude-flow neural predict --input "task description"
```

### Optimize
```bash
npx claude-flow neural optimize --target latency
```

## Best Practices
1. Use pretrain hook for batch learning
2. Store successful patterns after completion
3. Consolidate regularly to prevent forgetting
4. Route based on task complexity

---

**Source:** [`ruvnet/ruflo`](https://github.com/ruvnet/ruflo) → `.agents/skills/neural-training/SKILL.md`