neural-train

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Neural Training

神经训练

Train and manage SONA neural patterns for self-learning.

训练并管理用于自学习的SONA神经模式。

When to use

使用场景

After completing a successful task, use this skill to capture what worked and train the intelligence system so future tasks benefit from learned patterns.

在成功完成一项任务后，使用此技能来捕捉有效的方法并训练智能系统，以便未来的任务能够受益于已学习的模式。

Steps

步骤

Check current neural status — call
```
mcp__claude-flow__neural_status
```
to see active patterns and training state

Start a trajectory — call

mcp__claude-flow__hooks_intelligence_trajectory-start

with the task context

Record steps — for each significant action, call

mcp__claude-flow__hooks_intelligence_trajectory-step

End trajectory — call
```
mcp__claude-flow__hooks_intelligence_trajectory-end
```
with outcome (success/failure)

Train patterns — call

mcp__claude-flow__neural_train

with

--pattern-type coordination --epochs 10

Store patterns — call

mcp__claude-flow__hooks_intelligence_pattern-store

to persist learnings

Verify — call
```
mcp__claude-flow__neural_patterns
```
to confirm patterns were stored

检查当前神经状态 — 调用
```
mcp__claude-flow__neural_status
```
查看活跃模式和训练状态

启动轨迹 — 传入任务上下文调用

mcp__claude-flow__hooks_intelligence_trajectory-start

记录步骤 — 对于每个重要操作，调用

mcp__claude-flow__hooks_intelligence_trajectory-step

结束轨迹 — 传入结果（成功/失败）调用
```
mcp__claude-flow__hooks_intelligence_trajectory-end
```

训练模式 — 使用参数

--pattern-type coordination --epochs 10

调用

mcp__claude-flow__neural_train

存储模式 — 调用

mcp__claude-flow__hooks_intelligence_pattern-store

保存学习成果

验证 — 调用
```
mcp__claude-flow__neural_patterns
```
确认模式已存储

CLI alternative

CLI替代方案

bash

npx @claude-flow/cli@latest neural train --pattern-type coordination --epochs 10
npx @claude-flow/cli@latest neural patterns --list
npx @claude-flow/cli@latest neural status
npx @claude-flow/cli@latest hooks pretrain --model-type moe --epochs 10

bash

npx @claude-flow/cli@latest neural train --pattern-type coordination --epochs 10
npx @claude-flow/cli@latest neural patterns --list
npx @claude-flow/cli@latest neural status
npx @claude-flow/cli@latest hooks pretrain --model-type moe --epochs 10

SONA adaptation

SONA适配

For real-time micro-adaptation (<0.05ms), use:

```
mcp__claude-flow__ruvllm_sona_create
```
to initialize a SONA instance
```
mcp__claude-flow__ruvllm_sona_adapt
```
to adapt weights based on feedback

对于实时微适配（<0.05毫秒），请使用：

```
mcp__claude-flow__ruvllm_sona_create
```
初始化SONA实例
```
mcp__claude-flow__ruvllm_sona_adapt
```
根据反馈调整权重