Axioma KAN System — Axioma KAN 系统

🧠 Axioma KAN 系统 v1.5

Complete KAN lifecycle management for OpenClaw 代理s.

信息 Value Version 1.5.0 状态 ✅ Verified 组件s 4 scripts + AutoRe搜索 integration Tar获取 19 KANs (14 watchdogs + 5 L9 Swarm) auto-trAIned nightly Overview

This 技能 provides complete KAN (Kolmogorov-Arnold Networks) lifecycle management for the Axioma cluster:

创建 new KAN concepts and architectures TrAIn KAN 模型s with PyTorch Assemble KAN 流水线s and connections 监控 KAN 健康 and auto-evolve Integrate T-KAN for memory enhancement Table of Contents Purpose — Overview and goals When to Use — Trigger scenarios Prerequisites — Requirements 工具s — Core scripts Quick 启动 — 获取ting 启动ed KAN Core Concepts — Technical detAIls Error Handling — Troubleshooting ConstrAInts — Limitations Performance — Benchmarks Related Files — File structure References — Resources Support — Help and contact

• Purpose

Axioma KAN 系统 is the cluster's core intelligent infrastructure providing complete KAN lifecycle management:

Function Description Concept Creation De签名 new KAN architectures using kan_创建器.py 模型 TrAIning TrAIn KAN weights and parameters using kan_trAIner.py 流水线 Assembly Connect multiple KANs into 流水线s using kan_assembler.py T-KAN Integration 添加 temporal KAN for memory enhancement 健康 监控ing 监控 KAN performance using kan_健康.py Auto-Optimization Auto-retrAIn degraded KANs using kan_auto_task.py

• When to Use

Trigger Action "创建 a new KAN" 运行 kan_创建器.py "TrAIn a KAN 模型" 运行 kan_trAIner.py "Assemble KAN 流水线" 运行 kan_assembler.py "检查 KAN 健康" 运行 kan_健康.py "Auto-evolve KANs" 运行 kan_auto_task.py "Integrate T-KAN" 运行 kan_assembler.py --integrate-t-kan "创建 AutoRe搜索→KAN 流水线" 运行 autore搜索_task.py

• Prerequisites

Requirement Version 检查 Command 状态 Python >= 3.8 python3 --version [OK] NumPy >= 1.21 python3 -c "导入 numpy; print(numpy.__version__)" [OK] PyTorch >= 1.9 python3 -c "导入 torch; print(torch.__version__)" [OK] Qdrant 运行ning curl -s http://localhost:6333/collections [OK] Ollama 运行ning curl -s http://localhost:11434/API/tags [OK] 技能 directory exists ls <技能-dir>/ [OK] sudo rights Docker sudo docker ps [OK] 3.1 安装ation Commands # 安装 PyTorch (CPU or GPU) pip3 安装 torch>=1.9

# 安装 numpy for data processing pip3 安装 numpy

# 安装 Qdrant 命令行工具ent pip3 安装 qdrant-命令行工具ent

# 验证 all 安装ations python3 -c "导入 torch; 导入 numpy; 导入 qdrant_命令行工具ent; print('All OK')"

Note: PyTorch is required for all KAN operations (trAIning, inference, 模型 manipulation).

3.2 环境 Variables Variable Default Description KAN_模型_DIR 模型s/ Directory for KAN 模型 files QDRANT_HOST localhost Qdrant server host QDRANT_PORT 6333 Qdrant server port OLLAMA_HOST localhost Ollama server host OLLAMA_PORT 11434 Ollama server port

• 工具s

Core Scripts 工具 Path Purpose Example kan_创建器.py scripts/ 创建 new KAN concepts python3 kan_创建器.py --name stc --角色 "emotion" kan_trAIner.py scripts/ TrAIn KAN 模型s python3 kan_trAIner.py --kan stc --epochs 50 kan_assembler.py scripts/ Assemble KAN connections python3 kan_assembler.py --流水线 "stc→syn→w7" kan_健康.py scripts/ 健康 检查 python3 kan_健康.py --kan stc --verbose watchdog_unified_trAIner.py <技能-dir>/scripts/ TrAIn all 19 KANs python3 watchdog_unified_trAIner.py --all --epochs 200 External Integration 工具 Path Purpose 状态 kan_auto_task.py references/auto-task/ 13 KANs auto-optimization [OK] autore搜索_task.py references/auto-task/ Re搜索→Vaccine→KAN 流水线 [OK] l9_l6_bridge.py references/ L9-L6 bridge [OK]

• Quick 启动

5.1 创建 a KAN Concept cd <技能-directory> python3 scripts/kan_创建器.py --name my_watchdog --角色 "监控ing"

Expected 输出:

✅ KAN concept 'my_watchdog' 创建d 📁 Directory: scripts/my_watchdog/ 📋 Config: scripts/my_watchdog/config.json 🧠 模型: scripts/my_watchdog/模型s/my_watchdog_kan.pt

5.2 TrAIn a KAN 模型 # TrAIn specific KAN python3 scripts/kan_trAIner.py --kan stc --epochs 50 --batch-size 32

# 检查 健康 python3 scripts/kan_trAIner.py --检查-健康

Expected 输出:

🔄 TrAIning stc... Epoch 10/50: Loss = 0.0856 Epoch 20/50: Loss = 0.0233 Epoch 50/50: Loss = 0.0175 ✅ stc trAIned and saved!

Python API:

from kan_trAIner 导入 KANTrAIner

trAIner = KANTrAIner(kan_name='stc') trAIner.trAIn(epochs=50, batch_size=32) 健康 = trAIner.检查_健康() print(f'KAN 健康: {健康}')

5.3 Assemble KAN 流水线 # 创建 KAN 流水线 python3 scripts/kan_assembler.py --流水线 "stc→syn→w7" --输出 流水线.json

# Connect two KANs python3 scripts/kan_assembler.py --connect stc --with flx --mode serial

# 列出 all KANs python3 scripts/kan_assembler.py --列出

5.4 健康 检查 # 检查 all KANs cd <技能-directory> python3 scripts/kan_健康.py --all

# 运行 test suite (5/5 tests MUST PASS) python3 tests/test_kan_系统.py

Expected 输出:

╔══════════════════════════