【AI】Windows 10 部署 OpenClaw 龙虾

安装node24

下载页

Node.js — 下载 Node.js®

下载地址

https://nodejs.org/dist/v24.14.0/node-v24.14.0-x64.msi

验证安装(powershell)

PS C:\Users\yeqiang> node -v v24.14.0 PS C:\Users\yeqiang> npm.cmd -v 11.9.0

初始化在线模型(modelscope)

获取参考文档(api地址)

GLM-5 · 模型库

API推理介绍 · 文档中心

得到api地址为

https://api-inference.modelscope.cn

获取访问令牌

转到个人中心

个人中心 · 魔搭社区

访问控制->新建访问口令

创建一个长期使用的口令

复制这个口令

获取模型id

访问

api-inference.modelscope.cn/v1/models

找到glm5模型

准确的id是 ZhipuAI/GLM-5

一键安装OpenClaw

PS C:\Users\yeqiang> iwr -useb https://openclaw.ai/install.ps1 | iex OpenClaw Installer [OK] Windows detected [OK] Node.js v24.14.0 found [*] Installing OpenClaw (openclaw@latest)... [OK] OpenClaw installed OpenClaw installed successfully (OpenClaw 2026.3.13 (61d171a))! Installation complete. Your productivity is about to get weird. Starting setup... 🦞 OpenClaw 2026.3.13 (61d171a) — Less clicking, more shipping, fewer "where did that file go" moments. Windows detected - OpenClaw runs great on WSL2! Native Windows might be trickier. Quick setup: wsl --install (one command, one reboot) Guide: https://docs.openclaw.ai/windows ▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄ ██░▄▄▄░██░▄▄░██░▄▄▄██░▀██░██░▄▄▀██░████░▄▄▀██░███░██ ██░███░██░▀▀░██░▄▄▄██░█░█░██░█████░████░▀▀░██░█░█░██ ██░▀▀▀░██░█████░▀▀▀██░██▄░██░▀▀▄██░▀▀░█░██░██▄▀▄▀▄██ ▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀ 🦞 OPENCLAW 🦞 T OpenClaw onboarding | o Security ---------------------------------------------------------------------------------+ | | | Security warning — please read. | | | | OpenClaw is a hobby project and still in beta. Expect sharp edges. | | By default, OpenClaw is a personal agent: one trusted operator boundary. | | This bot can read files and run actions if tools are enabled. | | A bad prompt can trick it into doing unsafe things. | | | | OpenClaw is not a hostile multi-tenant boundary by default. | | If multiple users can message one tool-enabled agent, they share that delegated tool | | authority. | | | | If you’re not comfortable with security hardening and access control, don’t run | | OpenClaw. | | Ask someone experienced to help before enabling tools or exposing it to the internet. | | | | Recommended baseline: | | - Pairing/allowlists + mention gating. | | - Multi-user/shared inbox: split trust boundaries (separate gateway/credentials, ideally | | separate OS users/hosts). | | - Sandbox + least-privilege tools. | | - Shared inboxes: isolate DM sessions (`session.dmScope: per-channel-peer`) and keep | | tool access minimal. | | - Keep secrets out of the agent’s reachable filesystem. | | - Use the strongest available model for any bot with tools or untrusted inboxes. | | | | Run regularly: | | openclaw security audit --deep | | openclaw security audit --fix | | | | Must read: https://docs.openclaw.ai/gateway/security | | | +--------------------------------------------------------------------------------------------+ | * I understand this is personal-by-default and shared/multi-user use requires lock-down. Continue? | Yes / > No

选择yes

o I understand this is personal-by-default and shared/multi-user use requires lock-down. Continue? | Yes | * Onboarding mode | > QuickStart (Configure details later via openclaw configure.) | Manual

选择QuickStart

o Onboarding mode | QuickStart | o QuickStart -------------------------+ | | | Gateway port: 18789 | | Gateway bind: Loopback (127.0.0.1) | | Gateway auth: Token (default) | | Tailscale exposure: Off | | Direct to chat channels. | | | +--------------------------------------+ | * Model/auth provider | OpenAI | Anthropic | Chutes | MiniMax | Moonshot AI (Kimi K2.5) | Google | xAI (Grok) | Mistral AI | Volcano Engine | BytePlus | OpenRouter | Kilo Gateway | Qwen | Z.AI | Qianfan | Alibaba Cloud Model Studio | Copilot | Vercel AI Gateway | OpenCode | Xiaomi | Synthetic | Together AI | Hugging Face | Venice AI | LiteLLM | Cloudflare AI Gateway | > Custom Provider (Any OpenAI or Anthropic compatible endpoint) | Ollama | SGLang | vLLM | Skip for now

选择Custom Provider

o Model/auth provider | Custom Provider | * API Base URL | https://api-inference.modelscope.cn

输入modelscope的api地址

* How do you want to provide this API key? | > Paste API key now (Stores the key directly in OpenClaw config) | Use external secret provider * API Key (leave blank if not required) 粘贴正确的apikey

粘贴正确的apikey

* Endpoint compatibility | OpenAI-compatible | > Anthropic-compatible (Uses /messages) | Unknown (detect automatically)

选择Anthropic-compatible

o Endpoint compatibility | Anthropic-compatible | * Model ID | ZhipuAI/GLM-5█

输入glm5的模型id,回车

| o Verification successful. | * Endpoint ID | custom-api-inference-modelscope-cn█

等待验证完成后,出现模型id,直接回车

* Model alias (optional) | e.g. local, ollama

别名,不管,直接回车

通道选择:

* Select channel (QuickStart) | Telegram (Bot API) | WhatsApp (QR link) | Discord (Bot API) | IRC (Server + Nick) | Google Chat (Chat API) | Slack (Socket Mode) | Signal (signal-cli) | iMessage (imsg) | LINE (Messaging API) | Feishu/Lark (飞书) | Nostr (NIP-04 DMs) | Microsoft Teams (Bot Framework) | Mattermost (plugin) | Nextcloud Talk (self-hosted) | Matrix (plugin) | BlueBubbles (macOS app) | Zalo (Bot API) | Zalo (Personal Account) | Synology Chat (Webhook) | Tlon (Urbit) | > Skip for now (You can add channels later via `openclaw channels add`) —

国内似乎只有飞书可用,直接跳过。

搜索引擎

o Web search ----------------------------------------+ | | | Web search lets your agent look things up online. | | Choose a provider and paste your API key. | | Docs: https://docs.openclaw.ai/tools/web | | | +-----------------------------------------------------+ | * Search provider | > Brave Search (Structured results · country/language/time filters) | Gemini (Google Search) | Grok (xAI) | Kimi (Moonshot) | Perplexity Search | Skip for now

默认第一个Brave Search,回车,后续看是否支持国内引擎

o Search provider | Brave Search | o Brave Search API key | | o Web search ----------------------------------------------------------+ | | | No API key stored — web_search won't work until a key is available. | | Get your key at: https://brave.com/search/api/ | | Docs: https://docs.openclaw.ai/tools/web | | | +-----------------------------------------------------------------------+ | o Skills status -------------+ | | | Eligible: 4 | | Missing requirements: 40 | | Unsupported on this OS: 8 | | Blocked by allowlist: 0 | | | +-----------------------------+ | * Configure skills now? (recommended) | Yes / > No

选择以后配置skill

o Configure skills now? (recommended) | No | o Hooks ------------------------------------------------------------------+ | | | Hooks let you automate actions when agent commands are issued. | | Example: Save session context to memory when you issue /new or /reset. | | | | Learn more: https://docs.openclaw.ai/automation/hooks | | | +--------------------------------------------------------------------------+ | * Enable hooks? | [ ] Skip for now | [ ] 🚀 boot-md | [ ] 📎 bootstrap-extra-files | [ ] 📝 command-logger | [+] 💾 session-memory (Save session context to memory when /new or /reset command is issued)

选择session-memory,方向下,然后空格选中,变成+号状态

o Enable hooks? | 💾 session-memory | o Hooks Configured -----------------+ | | | Enabled 1 hook: session-memory | | | | You can manage hooks later with: | | openclaw hooks list | | openclaw hooks enable <name> | | openclaw hooks disable <name> | | | +------------------------------------+ Config overwrite: C:\Users\yeqiang\.openclaw\openclaw.json (sha256 02f3ade747bef0d4bd5b49bfa30500a426be823f637b7ad9b00581d7825dc20d -> f3268ea6bf0acb2ffc3adce6048a69277bbca05be8be1b275c52a60bbae9db10, backup=C:\Users\yeqiang\.openclaw\openclaw.json.bak) | o Gateway service runtime --------------------------------------------+ | | | QuickStart uses Node for the Gateway service (stable + supported). | | | +----------------------------------------------------------------------+ | o Gateway service install failed. | o Gateway ---------------------------------------------------------------+ | | | Gateway service install failed: schtasks create failed: ����: �ܾ����ʡ� | | | +-------------------------------------------------------------------------+ | o Gateway ---------------------------------------------------------------------------+ | | | Tip: native Windows now falls back to a per-user Startup-folder login item when | | Scheduled Task creation is denied; if install still fails, rerun from an elevated | | PowerShell or skip service install. | | | +-------------------------------------------------------------------------------------+ | o Health check failed: gateway closed (1006 abnormal closure (no close frame)): no close reason Gateway target: ws://127.0.0.1:18789 Source: local loopback Config: C:\Users\yeqiang\.openclaw\openclaw.json Bind: loopback | o Health check help --------------------------------+ | | | Docs: | | https://docs.openclaw.ai/gateway/health | | https://docs.openclaw.ai/gateway/troubleshooting | | | +----------------------------------------------------+ | o Optional apps ------------------------+ | | | Add nodes for extra features: | | - macOS app (system + notifications) | | - iOS app (camera/canvas) | | - Android app (camera/canvas) | | | +----------------------------------------+ | o Control UI -------------------------------------------------------------------------------+ | | | Web UI: http://127.0.0.1:18789/ | | Web UI (with token): | | http://127.0.0.1:18789/#token=3e1e45fc925077d37a0fea7071c8edd34551d1bc2e8d8e63 | | Gateway WS: ws://127.0.0.1:18789 | | Gateway: not detected (gateway closed (1006 abnormal closure (no close frame)): no close | | reason) | | Docs: https://docs.openclaw.ai/web/control-ui | | | +--------------------------------------------------------------------------------------------+ | o Workspace backup ----------------------------------------+ | | | Back up your agent workspace. | | Docs: https://docs.openclaw.ai/concepts/agent-workspace | | | +-----------------------------------------------------------+ | o Security ------------------------------------------------------+ | | | Running agents on your computer is risky — harden your setup: | | https://docs.openclaw.ai/security | | | +-----------------------------------------------------------------+ | o Shell completion -------------------------------------------------------+ | | | Shell completion installed. Restart your shell or run: source ~/.zshrc | | | +--------------------------------------------------------------------------+ | o Dashboard ready ----------------------------------------------------------------+ | | | Dashboard link (with token): | | http://127.0.0.1:18789/#token=3e1e45fc925077d37a0fea7071c8edd34551d1bc2e8d8e63 | | Opened in your browser. Keep that tab to control OpenClaw. | | | +----------------------------------------------------------------------------------+ | o Web search --------------------------------------------------+ | | | Provider Brave Search is selected but no API key was found. | | web_search will not work until a key is added. | | openclaw configure --section web | | | | Get your key at: https://brave.com/search/api/ | | Docs: https://docs.openclaw.ai/tools/web | | | +---------------------------------------------------------------+ | o What now -------------------------------------------------------------+ | | | What now: https://openclaw.ai/showcase ("What People Are Building"). | | | +------------------------------------------------------------------------+ | — Onboarding complete. Dashboard opened; keep that tab to control OpenClaw.

启动失败了

手动启动gateway

PS C:\Users\yeqiang> openclaw.cmd gateway 🦞 OpenClaw 2026.3.13 (61d171a) — The only open-source project where the mascot could eat the competition. 20:50:58 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\yeqiang\.openclaw\canvas) 20:50:59 [heartbeat] started 20:50:59 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s) 20:50:59 [gateway] agent model: custom-api-inference-modelscope-cn/ZhipuAI/GLM-5 20:50:59 [gateway] listening on ws://127.0.0.1:18789, ws://[::1]:18789 (PID 9880) 20:50:59 [gateway] log file: C:\Users\yeqiang\AppData\Local\Temp\openclaw\openclaw-2026-03-15.log 20:50:59 [browser/server] Browser control listening on http://127.0.0.1:18791/ (auth=token) 20:50:59 [hooks:loader] Registered hook: boot-md -> gateway:startup 20:50:59 [hooks:loader] Registered hook: bootstrap-extra-files -> agent:bootstrap 20:50:59 [hooks:loader] Registered hook: command-logger -> command 20:50:59 [hooks:loader] Registered hook: session-memory -> command:new, command:reset 20:50:59 [hooks] loaded 4 internal hook handlers 20:51:10 [gateway] device pairing auto-approved device=93f39242bf125dc222916878d661ace3d43e3eec198023ad54a13830d9ac981c role=operator 20:51:10 [ws] webchat connected conn=e953cb9f-66c4-4cbd-995d-7b8a6ba7241d remote=127.0.0.1 client=openclaw-control-ui webchat v2026.3.13 20:51:16 [agent/embedded] low context window: custom-api-inference-modelscope-cn/ZhipuAI/GLM-5 ctx=16000 (warn<32000) source=modelsConfig 20:51:16 [tools] tools.profile (coding) allowlist contains unknown entries (apply_patch, image). These entries are shipped core tools but unavailable in the current runtime/provider/model/config. 20:51:31 [compaction-safeguard] Compaction safeguard: cancelling compaction with no real conversation messages to summarize. 

访问web

OpenClaw Control

今天就先到这里吧。

参考:

Getting Started - OpenClaw

Read more

扩散模型详解:从DDPM到Stable Diffusion再到DiT的技术演进

扩散模型详解:从DDPM到Stable Diffusion再到DiT的技术演进

1.摘要 扩散模型(Diffusion Models)作为当前最热门的生成模型之一,已彻底改变图像生成领域,本文从DDPM开始,逐步深入到Stable Diffusion和DiT架构。 扩散模型就像是一个"破坏-修复"的过程,想象一下你有一张美丽的图片,然后一点点地给它加上噪声,直到完全看不清原来的图片,然后让AI学会如何一步步把噪声去掉,重新还原出原始图片。这就是扩散模型的基本思路。 2. DDPM:扩散模型的奠基之作(2020年) 2.1 什么是DDPM? DDPM(Denoising Diffusion Probabilistic Models)是扩散模型的开山鼻祖,由OpenAI团队在2020年提出,它的工作原理: 前向过程(加噪声):从一张清晰的图片开始,逐步添加噪声,最终变成完全随机的噪声图。 反向过程(去噪声):训练AI学会如何一步步去除噪声,从随机噪声中重建出原始图片。 2.2 DDPM的模型结构详解 DDPM的核心是一个U-Net网络结构,U-Net详细架构如下图:

【论文阅读】Vision-skeleton dual-modality framework for generalizable assessment of Parkinson’s disease ga

【论文阅读】Vision-skeleton dual-modality framework for generalizable assessment of Parkinson’s disease ga

论文题目:《Vision-skeleton dual-modality framework for generalizable assessment of Parkinson’s disease gait》 论文链接:https://doi.org/10.1016/j.media.2025.103727  代码链接:https://github.com/FJNU-LWP/PD-gait-VSDF 视觉-骨架双模态框架:通过视频实现帕金森病步态的泛化评估 研究背景介绍 帕金森病评估与帕金森病评分量表(MDS-UPDRS) 帕金森病步态评估 研究内容 总体方法流程 关键点视觉 Transformer (KVT) 图像块嵌入 (Patches embedding) 位置与连接嵌入 (Positions and connections embedding) 关键点自注意力 (Keypoints Self-Attention,

74个低空无人机AI算法详解,总体精度达90%,公安执法、消防应急、水利、林业、能源电力、城建、市政、城管、工程、农业、生态

74个低空无人机AI算法详解,总体精度达90%,公安执法、消防应急、水利、林业、能源电力、城建、市政、城管、工程、农业、生态

公安执法 一、人员智能识别与管控 聚焦人员相关的身份、行为、状态识别,核心服务于治安防控、人群管理、突发事件处置,是公安基层执法的核心应用方向: 1. 人员识别/计数:支持复杂场景(人群聚集、遮挡、移动)下的人员精准检测与数量统计,实时反馈人群密度,为大型活动安保、人群聚集风险管控提供数据支撑; 2. 人员异常聚焦识别:识别人员突然聚集、徘徊、逃窜、翻越护栏等异常行为,快速锁定可疑区域,触发执法预警; 3. 打架斗殴识别:精准检测肢体冲突、推搡、殴打等暴力行为,毫秒级触发预警并定位事发位置,助力执法人员快速处置,减少冲突升级; 4. 重点人员监控识别:对接公安重点人员数据库,通过人脸识别算法实现低空移动场景下的重点人员精准匹配与轨迹追踪,支持跨区域、动态化管控; 5. 人员属性识别:识别人员性别、年龄段、衣着特征、是否携带疑似管制器具 / 大件物品等属性信息,

2.2 基于ultrascale 架构FPGA的system manager wizard使用(温压监测)

2.2 基于ultrascale 架构FPGA的system manager wizard使用(温压监测)

Reference: 《PG185》《UG580》 部分文案源于网友博客,AIGC和个人理解,如有雷同纯属抄袭 一、介绍 简述: Xilinx System Management Wizard 是 Vivado 和 Vitis 工具中的一个图形化配置工具,主要用于为 FPGA 设计生成与系统监控和管理相关的 IP 核。这个工具帮助用户配置和集成诸如温度监控、电压监控、时钟监控、外部模拟输入等功能到 FPGA 设计中。它支持AXI4-Lite 与 DRP 接口 主要功能: * 温度和电压监测: * 内建传感器:支持 FPGA 内部温度、VCCINT(核心电压)、VCCAUX(辅助电压)、VCCBRAM(BRAM 电压)等电压和温度监测。通过 SYSMON 进行实时数据采集。