skills.rest
ruvnet
ruvnetCOMMUNITY
Explore all skills by this author →
Website𝕏@ruvGitHub

neural-training

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

70.1k|8.4k|Updated Jun 2, 2025
Search Tags:#agent routing#neural training#knowledge consolidation#pattern learning#mixture of experts#Machine learning#Claude-Flow
One-click install
npx skills add https://github.com/ruvnet/claude-flow --skill neural-training
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: neural-training
Source: https://github.com/ruvnet/claude-flow/tree/main/.agents/skills/neural-training
Command: npx skills add https://github.com/ruvnet/claude-flow --skill neural-training

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Agent systems often repeat mistakes and fail to learn from past executions. This Skill provides pattern training, knowledge consolidation, and adaptive routing so agents retain successful behaviors and improve over time without catastrophic forgetting.

Core Features & Use Cases

  • Pattern Training: Train neural patterns with SONA self-optimizing adaptation and Mixture of Experts routing across 8 experts.
  • Knowledge Consolidation: Use EWC++ to prevent catastrophic forgetting and LoRA-based distillation to extract key learnings.
  • Fast Pattern Retrieval: Search stored patterns via HNSW indexing for 150x-12,500x faster retrieval.
  • Use Case: After completing a batch of agent tasks, run pattern training to store successful strategies, then use prediction and optimization commands to route future tasks based on learned complexity.

Quick Start

Train neural patterns by running npx claude-flow neural train with the moe model type for 10 epochs, then check status with the neural status command.

Frequently Asked Questions about neural-training

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I train neural patterns with claude-flow?▼

Run npx claude-flow neural train with the --model-type moe flag and set the number of epochs, for example --epochs 10. This trains patterns using the Mixture of Experts architecture with 8 experts.

What is EWC++ used for in neural pattern training?▼

EWC++ prevents catastrophic forgetting during continuous learning. It consolidates knowledge so newly trained patterns do not overwrite previously learned successful behaviors.

How do I check the status of neural training?▼

Run npx claude-flow neural status to view the current training state. You can also list stored patterns with npx claude-flow neural patterns --type all.

When should I not use neural pattern training?▼

Skip it for simple tasks, one-off operations, or workflows where no learning is required. The training pipeline adds overhead that only pays off when patterns are reused across tasks.

How does HNSW improve pattern retrieval speed?▼

HNSW indexing accelerates pattern search by 150x to 12,500x compared to baseline lookup. It is used in the retrieve stage of the intelligence pipeline to fetch relevant patterns before judging and distillation.

Related Skills

Explore More Software Engineering Skills→
ste-bahste-bah

run-agent

Assemble a Context Envelope and spawn a Task tool subagent to execute tasks.

Community
Advanced
flagos-aiflagos-ai
18

perf-test-flagos

Benchmark vLLM-served AI models across five workload profiles for throughput, latency, TTFT, and TPOT.

Official
Advanced
OmniNode-aiOmniNode-ai
3

check-agent-performance

Analyze agent routing times, confidence scores, and transformation success rates.

Official
Intermediate
tambo-aitambo-ai
11.2k

cli

Automate Tambo project setup and component management via CLI.

Official
Intermediate