| <!doctype html> |
| <html lang="en-US" data-theme="light"> |
| <head> |
| <meta charset="utf-8" /> |
| <meta name="viewport" content="width=device-width,initial-scale=1" /> |
| <meta name="generator" content="VuePress 2.0.0-rc.26" /> |
| <meta name="theme" content="VuePress Theme Hope 2.0.0-rc.99" /> |
| <style> |
| :root { |
| --vp-c-bg: #fff; |
| } |
| |
| [data-theme="dark"] { |
| --vp-c-bg: #1b1b1f; |
| } |
| |
| html, |
| body { |
| background: var(--vp-c-bg); |
| } |
| </style> |
| <script> |
| const userMode = localStorage.getItem("vuepress-theme-hope-scheme"); |
| const systemDarkMode = |
| window.matchMedia && |
| window.matchMedia("(prefers-color-scheme: dark)").matches; |
| |
| if (userMode === "dark" || (userMode !== "light" && systemDarkMode)) { |
| document.documentElement.setAttribute("data-theme", "dark"); |
| } |
| </script> |
| <script type="application/ld+json">{"@context":"https://schema.org","@type":"Article","headline":"AINode","image":["https://iotdb.apache.org/img/AINode-0-en.png","https://iotdb.apache.org/img/AInode1.png","https://iotdb.apache.org/img/AInode2.png","https://iotdb.apache.org/img/ainode-upgrade-state-apache-en.png"],"dateModified":"2026-02-04T04:01:34.000Z","author":[]}</script><meta property="og:url" content="https://iotdb.apache.org/UserGuide/latest/AI-capability/AINode_Upgrade_apache.html"><meta property="og:site_name" content="IoTDB Website"><meta property="og:title" content="AINode"><meta property="og:description" content="AINode AINode is a native IoTDB node that supports the registration, management, and invocation of time-series related models. It includes industry-leading self-developed time-s..."><meta property="og:type" content="article"><meta property="og:image" content="https://iotdb.apache.org/img/AINode-0-en.png"><meta property="og:locale" content="en-US"><meta property="og:locale:alternate" content="zh-CN"><meta property="og:updated_time" content="2026-02-04T04:01:34.000Z"><meta property="article:modified_time" content="2026-02-04T04:01:34.000Z"><link rel="alternate" hreflang="zh-cn" href="https://iotdb.apache.org/zh/UserGuide/latest/AI-capability/AINode_Upgrade_apache.html"><script async src="https://widget.kapa.ai/kapa-widget.bundle.js" data-website-id="2d37bfdd-8d98-40ba-9223-9d4f81bfb327" data-project-name="Apache IoTDB" data-project-color="#FFFFFF" data-button-z-index="1999" data-button-padding="4px" data-button-border-radius="4px" data-button-image-height="24px" data-button-image-width="20px" data-button-text-color="#9E2878" data-project-logo="https://iotdb.apache.org/img/logo.svg" data-button-position-right="16px" data-button-position-bottom="8px" data-button-height="56px" data-button-width="48px" data-button-text="Ask" data-modal-override-open-selector="#custom-ask-ai-button" data-modal-image-width="150px" data-modal-title="AI Docs" data-modal-title-color="#9E2878" data-deep-thinking-button-active-bg-color="#F6F7F8" data-deep-thinking-button-active-text-color="#9E2878" data-deep-thinking-button-active-hover-text-color="#9E2878" data-modal-disclaimer="This is a custom LLM for Apache IoTDB with access to all [documentation](iotdb.apache.org/docs/), [GitHub Open Issues, PRs and READMEs](github.com/apache/iotdb). |
| |
| If you encounter <Error in verifying browser for feedback submission. Captcha token could not be obtained.> please ensure that you can access Google services." data-user-analytics-fingerprint-enabled="true" data-consent-required="true" data-consent-screen-disclaimer="By clicking <I agree, let's chat>, you consent to the use of the AI assistant in accordance with kapa.ai's [Privacy Policy](https://www.kapa.ai/content/privacy-policy). This service uses reCAPTCHA, which requires your consent to Google's [Privacy Policy](https://policies.google.com/privacy) and [Terms of Service](https://policies.google.com/terms). By proceeding, you explicitly agree to both kapa.ai's and Google's privacy policies."></script><link rel="icon" href="/favicon.ico"><meta name="Description" content="Apache IoTDB: Time Series Database for IoT"><meta name="Keywords" content="TSDB, time series, time series database, IoTDB, IoT database, IoT data management, 时序数据库, 时间序列管理, IoTDB, 物联网数据库, 实时数据库, 物联网数据管理, 物联网数据"><meta name="baidu-site-verification" content="wfKETzB3OT"><meta name="google-site-verification" content="mZWAoRY0yj_HAr-s47zHCGHzx5Ju-RVm5wDbPnwQYFo"><script type="text/javascript">var _paq = window._paq = window._paq || []; |
| /* tracker methods like "setCustomDimension" should be called before "trackPageView" */ |
| _paq.push(["setDoNotTrack", true]); |
| _paq.push(["disableCookies"]); |
| _paq.push(['trackPageView']); |
| _paq.push(['enableLinkTracking']); |
| (function() { |
| var u="https://analytics.apache.org/"; |
| _paq.push(['setTrackerUrl', u+'matomo.php']); |
| _paq.push(['setSiteId', '56']); |
| var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; |
| g.async=true; g.src=u+'matomo.js'; s.parentNode.insertBefore(g,s); |
| })(); |
| </script><title>AINode | IoTDB Website</title><meta name="description" content="AINode AINode is a native IoTDB node that supports the registration, management, and invocation of time-series related models. It includes industry-leading self-developed time-s..."> |
| <link rel="preload" href="/assets/style-JraSdRX4.css" as="style"><link rel="stylesheet" href="/assets/style-JraSdRX4.css"> |
| <link rel="modulepreload" href="/assets/app-pmkzp6TS.js"><link rel="modulepreload" href="/assets/AINode_Upgrade_apache.html-CjNSUeWO.js"><link rel="modulepreload" href="/assets/AINode-0-en-CP_5ojW_.js"><link rel="modulepreload" href="/assets/AInode1-DHdC6PUe.js"><link rel="modulepreload" href="/assets/AInode2-CBFVblh0.js"><link rel="modulepreload" href="/assets/ainode-upgrade-state-apache-en-D7pHvmG1.js"> |
| |
| </head> |
| <body> |
| <div id="app"><!--[--><!--[--><!--[--><span tabindex="-1"></span><a href="#main-content" class="vp-skip-link sr-only">Skip to main content</a><!--]--><!--[--><div class="theme-container external-link-icon has-toc" vp-container><!--[--><header id="navbar" class="vp-navbar" vp-navbar><div class="vp-navbar-start"><button type="button" class="vp-toggle-sidebar-button" title="Toggle Sidebar"><span class="icon"></span></button><!--[--><a class="route-link vp-brand" href="/" aria-label="Take me home"><img class="vp-nav-logo" src="/logo.png" alt><!----><!----></a><!--]--></div><div class="vp-navbar-center"><!--[--><!--]--></div><div class="vp-navbar-end"><!--[--><!--[--><div id="docsearch-container" style="display:none;"></div><div class="docsearch-placeholder"><button type="button" aria-label="Search" aria-keyshortcuts="Control+k" class="DocSearch DocSearch-Button"><span class="DocSearch-Button-Container"><svg width="20" height="20" viewBox="0 0 24 24" aria-hidden="true" class="DocSearch-Search-Icon"><circle cx="11" cy="11" r="8" stroke="currentColor" fill="none" stroke-width="1.4"></circle><path d="m21 21-4.3-4.3" stroke="currentColor" fill="none" stroke-linecap="round" stroke-linejoin="round"></path></svg><span class="DocSearch-Button-Placeholder">Search</span></span><span class="DocSearch-Button-Keys"><kbd class="DocSearch-Button-Key DocSearch-Button-Key--ctrl">Ctrl</kbd><kbd class="DocSearch-Button-Key">K</kbd></span></button></div><!--]--><div><button id="custom-ask-ai-button"> ✨ Ask AI </button></div><nav class="vp-nav-links"><div class="vp-nav-item hide-in-mobile"><div class="vp-dropdown-wrapper"><button type="button" class="vp-dropdown-title" aria-label="Documentation"><!--[--><!---->Documentation<!--]--><span class="arrow"></span><ul class="vp-dropdown"><li class="vp-dropdown-item"><a class="route-link auto-link" href="/UserGuide/latest/QuickStart/QuickStart_apache.html" aria-label="v2.0.x"><!---->v2.0.x<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/UserGuide/V1.3.x/QuickStart/QuickStart_apache.html" aria-label="v1.3.x"><!---->v1.3.x<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/UserGuide/V1.2.x/QuickStart/QuickStart.html" aria-label="v1.2.x"><!---->v1.2.x<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/UserGuide/V0.13.x/QuickStart/QuickStart.html" aria-label="v0.13.x"><!---->v0.13.x<!----></a></li></ul></button></div></div><div class="vp-nav-item hide-in-mobile"><a class="auto-link external-link" href="https://cwiki.apache.org/confluence/display/IOTDB/System+Design" aria-label="Design" rel="noopener noreferrer" target="_blank"><!---->Design<!----></a></div><div class="vp-nav-item hide-in-mobile"><a class="route-link auto-link" href="/Download/" aria-label="Download"><!---->Download<!----></a></div><div class="vp-nav-item hide-in-mobile"><div class="vp-dropdown-wrapper"><button type="button" class="vp-dropdown-title" aria-label="Community"><!--[--><!---->Community<!--]--><span class="arrow"></span><ul class="vp-dropdown"><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/About-the-Community.html" aria-label="About the Community"><!---->About the Community<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/Development-Guide.html" aria-label="Development Guide"><!---->Development Guide<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/Community-Partners.html" aria-label="Community Partners"><!---->Community Partners<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/Communication-Channels.html" aria-label="Communication Channels"><!---->Communication Channels<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/Events-and-Reports.html" aria-label="Events and Reports"><!---->Events and Reports<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/Community/Committers.html" aria-label="Committers"><!---->Committers<!----></a></li></ul></button></div></div><div class="vp-nav-item hide-in-mobile"><div class="vp-dropdown-wrapper"><button type="button" class="vp-dropdown-title" aria-label="ASF"><!--[--><!---->ASF<!--]--><span class="arrow"></span><ul class="vp-dropdown"><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/" aria-label="Foundation" rel="noopener noreferrer" target="_blank"><!---->Foundation<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/licenses/" aria-label="License" rel="noopener noreferrer" target="_blank"><!---->License<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/security/" aria-label="Security" rel="noopener noreferrer" target="_blank"><!---->Security<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/foundation/sponsorship.html" aria-label="Sponsorship" rel="noopener noreferrer" target="_blank"><!---->Sponsorship<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/foundation/thanks.html" aria-label="Thanks" rel="noopener noreferrer" target="_blank"><!---->Thanks<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://www.apache.org/events/current-event" aria-label="Current Events" rel="noopener noreferrer" target="_blank"><!---->Current Events<!----></a></li><li class="vp-dropdown-item"><a class="auto-link external-link" href="https://privacy.apache.org/policies/privacy-policy-public.html" aria-label="Privacy" rel="noopener noreferrer" target="_blank"><!---->Privacy<!----></a></li></ul></button></div></div></nav><div class="vp-nav-item"><div class="vp-dropdown-wrapper"><button type="button" class="vp-dropdown-title" aria-label="Select language"><!--[--><svg xmlns="http://www.w3.org/2000/svg" class="icon i18n-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="i18n icon" name="i18n" style="width:1rem;height:1rem;vertical-align:middle;"><path d="M379.392 460.8 494.08 575.488l-42.496 102.4L307.2 532.48 138.24 701.44l-71.68-72.704L234.496 460.8l-45.056-45.056c-27.136-27.136-51.2-66.56-66.56-108.544h112.64c7.68 14.336 16.896 27.136 26.112 35.84l45.568 46.08 45.056-45.056C382.976 312.32 409.6 247.808 409.6 204.8H0V102.4h256V0h102.4v102.4h256v102.4H512c0 70.144-37.888 161.28-87.04 210.944L378.88 460.8zM576 870.4 512 1024H409.6l256-614.4H768l256 614.4H921.6l-64-153.6H576zM618.496 768h196.608L716.8 532.48 618.496 768z"></path></svg><!--]--><span class="arrow"></span><ul class="vp-dropdown"><li class="vp-dropdown-item"><a class="route-link route-link-active auto-link" href="/UserGuide/latest/AI-capability/AINode_Upgrade_apache.html" aria-label="English"><!---->English<!----></a></li><li class="vp-dropdown-item"><a class="route-link auto-link" href="/zh/UserGuide/latest/AI-capability/AINode_Upgrade_apache.html" aria-label="简体中文"><!---->简体中文<!----></a></li></ul></button></div></div><div class="vp-nav-item hide-in-mobile"><button type="button" class="vp-color-mode-switch" id="color-mode-switch"><svg xmlns="http://www.w3.org/2000/svg" class="icon auto-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="auto icon" name="auto" style="display:none;"><path d="M512 992C246.92 992 32 777.08 32 512S246.92 32 512 32s480 214.92 480 480-214.92 480-480 480zm0-840c-198.78 0-360 161.22-360 360 0 198.84 161.22 360 360 360s360-161.16 360-360c0-198.78-161.22-360-360-360zm0 660V212c165.72 0 300 134.34 300 300 0 165.72-134.28 300-300 300z"></path></svg><svg xmlns="http://www.w3.org/2000/svg" class="icon dark-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="dark icon" name="dark" style="display:none;"><path d="M524.8 938.667h-4.267a439.893 439.893 0 0 1-313.173-134.4 446.293 446.293 0 0 1-11.093-597.334A432.213 432.213 0 0 1 366.933 90.027a42.667 42.667 0 0 1 45.227 9.386 42.667 42.667 0 0 1 10.24 42.667 358.4 358.4 0 0 0 82.773 375.893 361.387 361.387 0 0 0 376.747 82.774 42.667 42.667 0 0 1 54.187 55.04 433.493 433.493 0 0 1-99.84 154.88 438.613 438.613 0 0 1-311.467 128z"></path></svg><svg xmlns="http://www.w3.org/2000/svg" class="icon light-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="light icon" name="light" style="display:block;"><path d="M952 552h-80a40 40 0 0 1 0-80h80a40 40 0 0 1 0 80zM801.88 280.08a41 41 0 0 1-57.96-57.96l57.96-58a41.04 41.04 0 0 1 58 58l-58 57.96zM512 752a240 240 0 1 1 0-480 240 240 0 0 1 0 480zm0-560a40 40 0 0 1-40-40V72a40 40 0 0 1 80 0v80a40 40 0 0 1-40 40zm-289.88 88.08-58-57.96a41.04 41.04 0 0 1 58-58l57.96 58a41 41 0 0 1-57.96 57.96zM192 512a40 40 0 0 1-40 40H72a40 40 0 0 1 0-80h80a40 40 0 0 1 40 40zm30.12 231.92a41 41 0 0 1 57.96 57.96l-57.96 58a41.04 41.04 0 0 1-58-58l58-57.96zM512 832a40 40 0 0 1 40 40v80a40 40 0 0 1-80 0v-80a40 40 0 0 1 40-40zm289.88-88.08 58 57.96a41.04 41.04 0 0 1-58 58l-57.96-58a41 41 0 0 1 57.96-57.96z"></path></svg></button></div><div class="vp-nav-item vp-action"><a class="vp-action-link" href="https://github.com/apache/iotdb" target="_blank" rel="noopener noreferrer" aria-label="GitHub"><svg xmlns="http://www.w3.org/2000/svg" class="icon github-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="github icon" name="github" style="width:1.25rem;height:1.25rem;vertical-align:middle;"><path d="M511.957 21.333C241.024 21.333 21.333 240.981 21.333 512c0 216.832 140.544 400.725 335.574 465.664 24.49 4.395 32.256-10.07 32.256-23.083 0-11.69.256-44.245 0-85.205-136.448 29.61-164.736-64.64-164.736-64.64-22.315-56.704-54.4-71.765-54.4-71.765-44.587-30.464 3.285-29.824 3.285-29.824 49.195 3.413 75.179 50.517 75.179 50.517 43.776 75.008 114.816 53.333 142.762 40.79 4.523-31.66 17.152-53.377 31.19-65.537-108.971-12.458-223.488-54.485-223.488-242.602 0-53.547 19.114-97.323 50.517-131.67-5.035-12.33-21.93-62.293 4.779-129.834 0 0 41.258-13.184 134.912 50.346a469.803 469.803 0 0 1 122.88-16.554c41.642.213 83.626 5.632 122.88 16.554 93.653-63.488 134.784-50.346 134.784-50.346 26.752 67.541 9.898 117.504 4.864 129.834 31.402 34.347 50.474 78.123 50.474 131.67 0 188.586-114.73 230.016-224.042 242.09 17.578 15.232 33.578 44.672 33.578 90.454v135.85c0 13.142 7.936 27.606 32.854 22.87C862.25 912.597 1002.667 728.747 1002.667 512c0-271.019-219.648-490.667-490.71-490.667z"></path></svg></a></div><!--]--><button type="button" class="vp-toggle-navbar-button" aria-label="Toggle Navbar" aria-expanded="false" aria-controls="nav-screen"><span><span class="vp-top"></span><span class="vp-middle"></span><span class="vp-bottom"></span></span></button></div></header><!----><!--]--><!----><div class="toggle-sidebar-wrapper"><span class="arrow start"></span></div><aside id="sidebar" class="vp-sidebar" vp-sidebar><!--[--><p class="vp-sidebar-header iotdb-sidebar-header"><span class="vp-sidebar-title">IoTDB User Guide (V2.0.x)</span></p><div class="sidebar-top-wrapper"><ul class="switch-list"><li class="switch-type switch-active">Tree</li><li class="switch-type">Table</li></ul><div class="help-icon-wrapper"><div class="help-icon"><span>Description</span></div></div></div><!--]--><ul class="vp-sidebar-links"><li><section class="vp-sidebar-group"><p class="vp-sidebar-header"><!----><span class="vp-sidebar-title">IoTDB User Guide (V2.0.x)</span><!----></p><ul class="vp-sidebar-links"></ul></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">About IoTDB</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Background knowledge</span><span class="vp-arrow end"></span></button><!----></section></li><li><a class="route-link auto-link vp-sidebar-link" href="/UserGuide/latest/QuickStart/QuickStart_apache.html" aria-label="Quick Start"><!---->Quick Start<!----></a></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Installation and Deployment</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Basic Functions</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Advanced Features</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable active" type="button"><!----><span class="vp-sidebar-title">AI capability</span><span class="vp-arrow down"></span></button><ul class="vp-sidebar-links"><li><a class="route-link auto-link vp-sidebar-link" href="/UserGuide/latest/AI-capability/AINode_apache.html" aria-label="AINode(V2.0.5/6)"><!---->AINode(V2.0.5/6)<!----></a></li><li><a class="route-link route-link-active auto-link vp-sidebar-link active" href="/UserGuide/latest/AI-capability/AINode_Upgrade_apache.html" aria-label="AINode(V2.0.8)"><!---->AINode(V2.0.8)<!----></a></li><li><a class="route-link auto-link vp-sidebar-link" href="/UserGuide/latest/AI-capability/TimeSeries-Large-Model.html" aria-label="TimeSeries Large Model(V2.0.5/6)"><!---->TimeSeries Large Model(V2.0.5/6)<!----></a></li><li><a class="route-link auto-link vp-sidebar-link" href="/UserGuide/latest/AI-capability/TimeSeries-Large-Model_Upgrade_apache.html" aria-label="TimeSeries Large Model(V2.0.8)"><!---->TimeSeries Large Model(V2.0.8)<!----></a></li></ul></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Tools System</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">API</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Ecosystem Integration</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">SQL Manual</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Technical Insider</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">Reference</span><span class="vp-arrow end"></span></button><!----></section></li><li><section class="vp-sidebar-group"><button class="vp-sidebar-header clickable" type="button"><!----><span class="vp-sidebar-title">FAQ</span><span class="vp-arrow end"></span></button><!----></section></li></ul><!----></aside><!--[--><main id="main-content" class="vp-page"><!--[--><!----><!----><nav class="vp-breadcrumb disable"></nav><div class="vp-page-title"><h1><!---->AINode</h1><div class="page-info"><!----><!----><span class="page-date-info" aria-label="Writing Date📅" data-balloon-pos="up"><svg xmlns="http://www.w3.org/2000/svg" class="icon calendar-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="calendar icon" name="calendar"><path d="M716.4 110.137c0-18.753-14.72-33.473-33.472-33.473-18.753 0-33.473 14.72-33.473 33.473v33.473h66.993v-33.473zm-334.87 0c0-18.753-14.72-33.473-33.473-33.473s-33.52 14.72-33.52 33.473v33.473h66.993v-33.473zm468.81 33.52H716.4v100.465c0 18.753-14.72 33.473-33.472 33.473a33.145 33.145 0 01-33.473-33.473V143.657H381.53v100.465c0 18.753-14.72 33.473-33.473 33.473a33.145 33.145 0 01-33.473-33.473V143.657H180.6A134.314 134.314 0 0046.66 277.595v535.756A134.314 134.314 0 00180.6 947.289h669.74a134.36 134.36 0 00133.94-133.938V277.595a134.314 134.314 0 00-133.94-133.938zm33.473 267.877H147.126a33.145 33.145 0 01-33.473-33.473c0-18.752 14.72-33.473 33.473-33.473h736.687c18.752 0 33.472 14.72 33.472 33.473a33.145 33.145 0 01-33.472 33.473z"></path></svg><span data-allow-mismatch="text">2/4/26</span><meta property="datePublished" content="2026-02-04T04:01:34.000Z"></span><!----><span class="page-reading-time-info" aria-label="Reading Time⌛" data-balloon-pos="up"><svg xmlns="http://www.w3.org/2000/svg" class="icon timer-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="timer icon" name="timer"><path d="M799.387 122.15c4.402-2.978 7.38-7.897 7.38-13.463v-1.165c0-8.933-7.38-16.312-16.312-16.312H256.33c-8.933 0-16.311 7.38-16.311 16.312v1.165c0 5.825 2.977 10.874 7.637 13.592 4.143 194.44 97.22 354.963 220.201 392.763-122.204 37.542-214.893 196.511-220.2 389.397-4.661 5.049-7.638 11.651-7.638 19.03v5.825h566.49v-5.825c0-7.379-2.849-13.981-7.509-18.9-5.049-193.016-97.867-351.985-220.2-389.527 123.24-37.67 216.446-198.453 220.588-392.892zM531.16 450.445v352.632c117.674 1.553 211.787 40.778 211.787 88.676H304.097c0-48.286 95.149-87.382 213.728-88.676V450.445c-93.077-3.107-167.901-81.297-167.901-177.093 0-8.803 6.99-15.793 15.793-15.793 8.803 0 15.794 6.99 15.794 15.793 0 80.261 63.69 145.635 142.01 145.635s142.011-65.374 142.011-145.635c0-8.803 6.99-15.793 15.794-15.793s15.793 6.99 15.793 15.793c0 95.019-73.789 172.82-165.96 177.093z"></path></svg><span>About 13 min</span><meta property="timeRequired" content="PT13M"></span><!----><!----></div><hr></div><!----><div class="" vp-content><!----><div id="markdown-content"><h1 id="ainode" tabindex="-1"><a class="header-anchor" href="#ainode"><span>AINode</span></a></h1><p>AINode is a native IoTDB node that supports the registration, management, and invocation of time-series related models. It includes industry-leading self-developed time-series large models, such as the Timer series models developed by Tsinghua University. These models can be called using standard SQL statements, enabling millisecond-level real-time inference for time series data, supporting applications such as time series trend prediction, missing value imputation, and anomaly detection.</p><p>The system architecture is shown in the following figure:</p><figure><img src="/img/AINode-0-en.png" alt="" tabindex="0" loading="lazy"><figcaption></figcaption></figure><p>The responsibilities of the three nodes are as follows:</p><ul><li><strong>ConfigNode</strong>: Responsible for distributed node management and load balancing.</li><li><strong>DataNode</strong>: Responsible for receiving and parsing users' SQL requests; responsible for storing time series data; responsible for data preprocessing calculations.</li><li><strong>AINode</strong>: Responsible for the management and use of time series models.</li></ul><h2 id="_1-advantages" tabindex="-1"><a class="header-anchor" href="#_1-advantages"><span>1. Advantages</span></a></h2><p>Compared to building a machine learning service separately, it has the following advantages:</p><ul><li><strong>Simple and Easy to Use</strong>: No need to use Python or Java programming. Machine learning model management and inference can be completed using SQL statements. For example, creating a model can use the CREATE MODEL statement, and using a model for inference can use the <code>CALL INFERENCE(...)</code> statement, making it simpler and more convenient.</li><li><strong>Avoid Data Migration</strong>: Using IoTDB's native machine learning can directly apply data stored in IoTDB to machine learning models for inference, without moving data to a separate machine learning service platform. This accelerates data processing, improves security, and reduces costs.</li></ul><figure><img src="/img/AInode1.png" alt="" tabindex="0" loading="lazy"><figcaption></figcaption></figure><ul><li><strong>Built-in Advanced Algorithms</strong>: Supports industry-leading machine learning analysis algorithms, covering typical time series analysis tasks, and empowering time series databases with native data analysis capabilities. For example: <ul><li><strong>Time Series Forecasting</strong>: Learn change patterns from past time series data; based on given past observations, output the most likely future sequence predictions.</li><li><strong>Anomaly Detection for Time Series</strong>: Detect and identify anomalies in given time series data, helping to discover abnormal behavior in time series.</li></ul></li></ul><h2 id="_2-basic-concepts" tabindex="-1"><a class="header-anchor" href="#_2-basic-concepts"><span>2. Basic Concepts</span></a></h2><ul><li><strong>Model</strong>: Machine learning model, which takes time series data as input and outputs the results or decisions of the analysis task. Model is the basic management unit of AINode, supporting model management including creation (registration), deletion, query, and usage (inference).</li><li><strong>Create</strong>: Load external designed or trained model files or algorithms into AINode, managed and used uniformly by IoTDB.</li><li><strong>Inference</strong>: Use the created model to complete the time series analysis task applicable to the model on the specified time series data.</li><li><strong>Built-in</strong>: AINode comes with common time series analysis scenario (e.g., prediction and anomaly detection) machine learning algorithms or self-developed models.</li></ul><figure><img src="/img/AInode2.png" alt="" tabindex="0" loading="lazy"><figcaption></figcaption></figure><h2 id="_3-installation-and-deployment" tabindex="-1"><a class="header-anchor" href="#_3-installation-and-deployment"><span>3. Installation and Deployment</span></a></h2><p>AINode deployment can be referenced in the document <a class="route-link" href="/UserGuide/latest/Deployment-and-Maintenance/AINode_Deployment_Upgrade_apache.html">AINode Deployment</a>.</p><h2 id="_4-usage-guide" tabindex="-1"><a class="header-anchor" href="#_4-usage-guide"><span>4. Usage Guide</span></a></h2><p>AINode supports two major functions: model inference and model management (registration, viewing, deletion, loading, unloading, etc.). The following sections will provide detailed instructions.</p><h3 id="_4-1-model-inference" tabindex="-1"><a class="header-anchor" href="#_4-1-model-inference"><span>4.1 Model Inference</span></a></h3><p>SQL syntax:</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">call</span><span style="color:#ABB2BF;"> inference(</span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">model_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">,inputSql,(</span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">parameterName</span><span style="color:#56B6C2;">>=<</span><span style="color:#ABB2BF;">parameterValue</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">)*)</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>After completing the model registration (built-in models do not require the registration process), the inference function of the model can be used by calling the inference function with the call keyword. The corresponding parameter descriptions are as follows:</p><ul><li><p><strong>model_id</strong>: Corresponds to an already registered model</p></li><li><p><strong>sql</strong>: SQL query statement, the result of the query is used as the input for model inference. The dimensions of the query results (rows and columns) need to match the size specified in the model's config. (It is recommended not to use the <code>SELECT *</code> clause in this sql, because in IoTDB, <code>*</code> does not sort the columns, so the column order is undefined. It is recommended to use <code>SELECT s0,s1</code> to ensure the column order matches the expected input of the model.)</p></li><li><p><strong>parameterName/parameterValue</strong>: currently supports:</p><table><thead><tr><th>Parameter Name</th><th>Parameter Type</th><th>Parameter Description</th><th>Default Value</th></tr></thead><tbody><tr><td><strong>generateTime</strong></td><td>boolean</td><td>Whether to include a timestamp column in the result</td><td>false</td></tr><tr><td><strong>outputLength</strong></td><td>int</td><td>Specifies the length of the output result</td><td>96</td></tr></tbody></table></li></ul><p>Notes:</p><ol><li>The prerequisite for using built-in time series large models for inference is that the corresponding model weights are locally stored, with the directory being <code>/IOTDB_AINODE_HOME/data/ainode/models/builtin/model_id/</code>. If the local model weights are not present, they will be automatically pulled from HuggingFace. Please ensure that the local environment can directly access HuggingFace.</li><li>In deep learning applications, time stamp derived features (the time column in the data) are often used as covariates together with the input to the model to improve model performance. However, the time column is generally not included in the model's output results. To ensure generality, the model inference results only correspond to the model's true output. If the model does not output a time column, the result will not include it.</li></ol><p><strong>Example</strong></p><p>Sample data <a href="/img/ETTh-tree.csv">ETTh-tree</a></p><p>Here is an example of using the sundial model for inference. The input is 96 rows, and the output is 48 rows. We use SQL to perform the inference.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">IoTDB</span><span style="color:#56B6C2;">></span><span style="color:#C678DD;"> select</span><span style="color:#ABB2BF;"> OT </span><span style="color:#C678DD;">from</span><span style="color:#D19A66;"> root</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">db</span><span style="color:#ABB2BF;">.**</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+---------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#C678DD;">Time</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">root</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">db</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">etth</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">OT</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+---------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-01T00:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">531</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-01T01:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">27</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">787</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-01T02:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">27</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">787</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-01T03:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">25</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">044</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-01T04:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">21</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">948</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| ...... | ...... |</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T19:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">29</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">546</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T20:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">29</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">475</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T21:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">29</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">264</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T22:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">953</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T23:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">31</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">726</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+---------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">Total </span><span style="color:#C678DD;">line</span><span style="color:#C678DD;"> number</span><span style="color:#56B6C2;"> =</span><span style="color:#D19A66;"> 96</span><span style="color:#ABB2BF;"> </span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#ABB2BF;">IoTDB</span><span style="color:#56B6C2;">></span><span style="color:#C678DD;"> call</span><span style="color:#ABB2BF;"> inference(sundial,</span><span style="color:#98C379;">"select OT from root.db.**"</span><span style="color:#ABB2BF;">, generateTime</span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;">True, outputLength</span><span style="color:#56B6C2;">=</span><span style="color:#D19A66;">48</span><span style="color:#ABB2BF;">)</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+------------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#C678DD;">Time</span><span style="color:#ABB2BF;">| </span><span style="color:#C678DD;">output</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+------------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T23:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">537494659423828</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-04T23:</span><span style="color:#D19A66;">59</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">22</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">500</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">29</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">619892120361328</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-05T00:</span><span style="color:#D19A66;">58</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">45</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">28</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">815832138061523</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-05T01:</span><span style="color:#D19A66;">58</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">500</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">27</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">91131019592285</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-05T02:</span><span style="color:#D19A66;">57</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">26</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">893848419189453</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| ...... | ...... |</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-06T17:</span><span style="color:#D19A66;">33</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">500</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">24</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">40607261657715</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-06T18:</span><span style="color:#D19A66;">32</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">25</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">00441551208496</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-06T19:</span><span style="color:#D19A66;">31</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">52</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">500</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">24</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">907312393188477</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-06T20:</span><span style="color:#D19A66;">31</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">15</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">000</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">25</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">156436920166016</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">2016</span><span style="color:#ABB2BF;">-</span><span style="color:#D19A66;">07</span><span style="color:#ABB2BF;">-06T21:</span><span style="color:#D19A66;">30</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">37</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">500</span><span style="color:#ABB2BF;">+</span><span style="color:#D19A66;">08</span><span style="color:#ABB2BF;">:</span><span style="color:#D19A66;">00</span><span style="color:#ABB2BF;">|</span><span style="color:#D19A66;">25</span><span style="color:#ABB2BF;">.</span><span style="color:#D19A66;">335433959960938</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-----------------------------+------------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">Total </span><span style="color:#C678DD;">line</span><span style="color:#C678DD;"> number</span><span style="color:#56B6C2;"> =</span><span style="color:#D19A66;"> 48</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><h3 id="_4-2-register-custom-models" tabindex="-1"><a class="header-anchor" href="#_4-2-register-custom-models"><span>4.2 Register Custom Models</span></a></h3><p><strong>The following requirements must be met for Transformers models to be registered in AINode:</strong></p><ol><li><p>AINode currently uses transformers v4.56.2, so when building the model, avoid inheriting from low versions (<4.50) interfaces;</p></li><li><p>The model must inherit from a type of AINode inference task pipeline (currently supports forecast pipeline):</p><ol><li>iotdb-core/ainode/iotdb/ainode/core/inference/pipeline/basic_pipeline.py</li></ol><div class="language-python line-numbers-mode" data-highlighter="shiki" data-ext="python" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-python"><span class="line"><span style="color:#C678DD;">class</span><span style="color:#E5C07B;"> BasicPipeline</span><span style="color:#ABB2BF;">(</span><span style="color:#D19A66;">ABC</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#56B6C2;"> __init__</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">model_id</span><span style="color:#ABB2BF;">, **</span><span style="color:#D19A66;font-style:italic;">model_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#E5C07B;"> self</span><span style="color:#ABB2BF;">.model_info </span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;"> model_info</span></span> |
| <span class="line"><span style="color:#E5C07B;"> self</span><span style="color:#ABB2BF;">.device </span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;"> model_kwargs.</span><span style="color:#61AFEF;">get</span><span style="color:#ABB2BF;">(</span><span style="color:#98C379;">"device"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"cpu"</span><span style="color:#ABB2BF;">)</span></span> |
| <span class="line"><span style="color:#E5C07B;"> self</span><span style="color:#ABB2BF;">.model </span><span style="color:#56B6C2;">=</span><span style="color:#61AFEF;"> load_model</span><span style="color:#ABB2BF;">(model_info, </span><span style="color:#E06C75;font-style:italic;">device_map</span><span style="color:#56B6C2;">=</span><span style="color:#E5C07B;">self</span><span style="color:#ABB2BF;">.device, **model_kwargs)</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#61AFEF;"> @abstractmethod</span></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#61AFEF;"> preprocess</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">inputs</span><span style="color:#ABB2BF;">, **</span><span style="color:#D19A66;font-style:italic;">infer_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#98C379;"> Preprocess the input data before the inference task, including shape verification and numerical conversion.</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#C678DD;"> pass</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#61AFEF;"> @abstractmethod</span></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#61AFEF;"> postprocess</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">output</span><span style="color:#ABB2BF;">, **</span><span style="color:#D19A66;font-style:italic;">infer_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#98C379;"> Postprocess the output results after the inference task.</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#C678DD;"> pass</span></span> |
| <span class="line"></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#C678DD;">class</span><span style="color:#E5C07B;"> ForecastPipeline</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;">BasicPipeline</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#56B6C2;"> __init__</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">model_info</span><span style="color:#ABB2BF;">, **</span><span style="color:#D19A66;font-style:italic;">model_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#56B6C2;"> super</span><span style="color:#ABB2BF;">().</span><span style="color:#56B6C2;">__init__</span><span style="color:#ABB2BF;">(model_info, </span><span style="color:#E06C75;font-style:italic;">model_kwargs</span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;">model_kwargs)</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#61AFEF;"> preprocess</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">inputs</span><span style="color:#ABB2BF;">: list[dict[</span><span style="color:#56B6C2;">str</span><span style="color:#ABB2BF;">, dict[</span><span style="color:#56B6C2;">str</span><span style="color:#ABB2BF;">, torch.Tensor] </span><span style="color:#56B6C2;">|</span><span style="color:#ABB2BF;"> torch.Tensor]], **</span><span style="color:#D19A66;font-style:italic;">infer_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#98C379;"> Preprocess the input data before passing it to the model for inference, verifying the shape and type of input data.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Args:</span></span> |
| <span class="line"><span style="color:#98C379;"> inputs (list[dict]):</span></span> |
| <span class="line"><span style="color:#98C379;"> Input data, a list of dictionaries, each dictionary contains:</span></span> |
| <span class="line"><span style="color:#98C379;"> - 'targets': A tensor of shape (input_length,) or (target_count, input_length).</span></span> |
| <span class="line"><span style="color:#98C379;"> - 'past_covariates': Optional, a dictionary of tensors, each tensor of shape (input_length,).</span></span> |
| <span class="line"><span style="color:#98C379;"> - 'future_covariates': Optional, a dictionary of tensors, each tensor of shape (input_length,).</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> infer_kwargs (dict, optional): Additional keyword parameters for inference, such as:</span></span> |
| <span class="line"><span style="color:#98C379;"> - `output_length`(int): Used to verify validity if 'future_covariates' is provided.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Raises:</span></span> |
| <span class="line"><span style="color:#98C379;"> ValueError: If the input format is incorrect (e.g., missing keys, invalid tensor shapes).</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Returns:</span></span> |
| <span class="line"><span style="color:#98C379;"> Preprocessed and verified input data, ready for model inference.</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#C678DD;"> pass</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#61AFEF;"> forecast</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">inputs</span><span style="color:#ABB2BF;">, **</span><span style="color:#D19A66;font-style:italic;">infer_kwargs</span><span style="color:#ABB2BF;">):</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#98C379;"> Perform prediction on the given input.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Parameters:</span></span> |
| <span class="line"><span style="color:#98C379;"> inputs: Input data for prediction. Type and structure depend on the specific model implementation.</span></span> |
| <span class="line"><span style="color:#98C379;"> **infer_kwargs: Additional inference parameters, for example:</span></span> |
| <span class="line"><span style="color:#98C379;"> - `output_length`(int): The number of time points the model should generate.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Returns:</span></span> |
| <span class="line"><span style="color:#98C379;"> Prediction output, specific form depends on the specific model implementation.</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#C678DD;"> pass</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#C678DD;"> def</span><span style="color:#61AFEF;"> postprocess</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;font-style:italic;">self</span><span style="color:#ABB2BF;">, </span><span style="color:#D19A66;font-style:italic;">outputs</span><span style="color:#ABB2BF;">: list[torch.Tensor], **</span><span style="color:#D19A66;font-style:italic;">infer_kwargs</span><span style="color:#ABB2BF;">) -> list[torch.Tensor]:</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#98C379;"> Postprocess the model output after inference, verifying the shape of the output tensor and ensuring it meets expected dimensions.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Args:</span></span> |
| <span class="line"><span style="color:#98C379;"> outputs:</span></span> |
| <span class="line"><span style="color:#98C379;"> Model output, a list of 2D tensors, each tensor shape is `[target_count, output_length]`.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Raises:</span></span> |
| <span class="line"><span style="color:#98C379;"> InferenceModelInternalException: If the output tensor shape is invalid (e.g., dimension error).</span></span> |
| <span class="line"><span style="color:#98C379;"> ValueError: If the output format is incorrect.</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#98C379;"> Returns:</span></span> |
| <span class="line"><span style="color:#98C379;"> list[torch.Tensor]:</span></span> |
| <span class="line"><span style="color:#98C379;"> Postprocessed output, a list of 2D tensors.</span></span> |
| <span class="line"><span style="color:#98C379;"> """</span></span> |
| <span class="line"><span style="color:#C678DD;"> pass</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li><li><p>Modify the model configuration file config.json to ensure it contains the following fields:</p><div class="language-json line-numbers-mode" data-highlighter="shiki" data-ext="json" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-json"><span class="line"><span style="color:#ABB2BF;">{</span></span> |
| <span class="line"><span style="color:#E06C75;"> "auto_map"</span><span style="color:#ABB2BF;">: {</span></span> |
| <span class="line"><span style="color:#E06C75;"> "AutoConfig"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"config.Chronos2CoreConfig"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;">// Specify the model Config class</span></span> |
| <span class="line"><span style="color:#E06C75;"> "AutoModelForCausalLM"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"model.Chronos2Model"</span><span style="color:#7F848E;font-style:italic;"> // Specify the model class</span></span> |
| <span class="line"><span style="color:#ABB2BF;"> },</span></span> |
| <span class="line"><span style="color:#E06C75;"> "pipeline_cls"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"pipeline_chronos2.Chronos2Pipeline"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;">// Specify the model's inference pipeline</span></span> |
| <span class="line"><span style="color:#E06C75;"> "model_type"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"custom_t5"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;">// Specify the model type</span></span> |
| <span class="line"><span style="color:#ABB2BF;">}</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><ol><li>Must specify the model's Config class and model class through auto_map;</li><li>Must integrate and specify the inference pipeline class;</li><li>For built-in (builtin) and user-defined (user_defined) models managed by AINode, the model category (model_type) also serves as a unique identifier. That is, the model category to be registered must not duplicate any existing model types.</li></ol></li><li><p>Ensure the model directory to be registered contains the following files, and the model configuration file name and weight file name are not customizable:</p><ol><li>Model configuration file: config.json;</li><li>Model weight file: model.safetensors;</li><li>Model code: other .py files.</li></ol></li></ol><p><strong>The SQL syntax for registering a custom model is as follows:</strong></p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">CREATE</span><span style="color:#ABB2BF;"> MODEL </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">model_id</span><span style="color:#56B6C2;">></span><span style="color:#C678DD;"> USING</span><span style="color:#ABB2BF;"> URI </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">uri</span><span style="color:#56B6C2;">></span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p><strong>Parameter descriptions:</strong></p><ul><li><strong>model_id</strong>: The unique identifier of the custom model; must be unique, with the following constraints: <ul><li>Allowed characters: [0-9 a-z A-Z _] (letters, numbers (not starting), underscore (not starting))</li><li>Length limit: 2-64 characters</li><li>Case-sensitive</li></ul></li><li><strong>uri</strong>: The local URI address containing the model code and weights.</li></ul><p><strong>Example of registration:</strong></p><p>Upload a custom Transformers model from a local path, AINode will copy the folder to the user_defined directory.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">CREATE</span><span style="color:#ABB2BF;"> MODEL chronos2 </span><span style="color:#C678DD;">USING</span><span style="color:#ABB2BF;"> URI </span><span style="color:#98C379;">'file:///path/to/chronos2'</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>After executing the SQL, the model registration process will be asynchronous. You can view the model registration status by viewing the model (see the View Model section). After the model registration is complete, you can call the specific function using normal query methods to perform model inference.</p><h3 id="_4-3-view-models" tabindex="-1"><a class="header-anchor" href="#_4-3-view-models"><span>4.3 View Models</span></a></h3><p>Registered models can be viewed using the view command.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">SHOW MODELS</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>In addition to directly displaying all model information, you can specify <code>model_id</code> to view specific model information.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">SHOW MODELS </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">model_id</span><span style="color:#56B6C2;">></span><span style="color:#7F848E;font-style:italic;"> -- Show specific model only</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>The model display result includes the following content:</p><table><thead><tr><th><strong>ModelId</strong></th><th><strong>ModelType</strong></th><th><strong>Category</strong></th><th><strong>State</strong></th></tr></thead><tbody><tr><td>Model ID</td><td>Model Type</td><td>Model Category</td><td>Model State</td></tr></tbody></table><p>Where, the State model state machine flowchart is as follows:</p><figure><img src="/img/ainode-upgrade-state-apache-en.png" alt="" tabindex="0" loading="lazy"><figcaption></figcaption></figure><p>State machine flow description:</p><ol><li>After starting AINode, executing <code>show models</code> command can only view the <strong>system built-in (BUILTIN)</strong> models.</li><li>Users can import their own models, which are identified as <strong>user-defined (USER_DEFINED)</strong>; AINode will try to parse the model type (ModelType) from the model configuration file. If parsing fails, this field will be displayed as empty.</li><li>Time series large models (built-in models) weight files are not packaged with AINode, and AINode automatically downloads them when starting. <ol><li>During download, the state is ACTIVATING; after successful download, it becomes ACTIVE; if download fails, it becomes INACTIVE.</li></ol></li></ol><p><strong>View Example</strong></p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">IoTDB</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;"> show models</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">---------------------+--------------+--------------+-------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| ModelId| ModelType| Category| </span><span style="color:#C678DD;">State</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">---------------------+--------------+--------------+-------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| arima| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| holtwinters| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">|exponential_smoothing| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| naive_forecaster| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| stl_forecaster| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| gaussian_hmm| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| gmm_hmm| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| stray| sktime| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| custom| | user_defined| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| timer_xl| timer| builtin| activating|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| sundial| sundial| builtin| active|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| chronos2| t5| builtin| inactive|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">---------------------+--------------+--------------+-------------+</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><p>Built-in traditional time series model introduction:</p><table><thead><tr><th>Model Name</th><th>Core Concept</th><th>Applicable Scenario</th><th>Main Features</th></tr></thead><tbody><tr><td><strong>ARIMA</strong> (AutoRegressive Integrated Moving Average)</td><td>Combines autoregression (AR), differencing (I), and moving average (MA), used for predicting stationary time series or data that can be made stationary through differencing.</td><td>Univariate time series prediction, such as stock prices, sales, economic indicators.</td><td>1. Suitable for linear trends and weak seasonality data. 2. Requires selecting parameters (p,d,q). 3. Sensitive to missing values.</td></tr><tr><td><strong>Holt-Winters</strong> (Three-parameter exponential smoothing)</td><td>Based on exponential smoothing, introduces level, trend, and seasonal components, suitable for data with trend and seasonality.</td><td>Time series with clear seasonality and trend, such as monthly sales, power demand.</td><td>1. Can handle additive or multiplicative seasonality. 2. Gives higher weight to recent data. 3. Simple to implement.</td></tr><tr><td><strong>Exponential Smoothing</strong> (Exponential smoothing)</td><td>Uses weighted average of historical data, with weights decreasing exponentially over time, emphasizing the importance of recent observations.</td><td>Data without significant seasonality but with trend, such as short-term demand prediction.</td><td>1. Few parameters, simple calculation. 2. Suitable for stationary or slowly changing sequences. 3. Can be extended to double or triple exponential smoothing.</td></tr><tr><td><strong>Naive Forecaster</strong> (Naive predictor)</td><td>Uses the observation of the most recent period as the prediction for the next period, the simplest baseline model.</td><td>As a benchmark for other models or simple prediction when data has no obvious pattern.</td><td>1. No training required. 2. Sensitive to sudden changes. 3. Seasonal naive variant can use the same period of the previous season to predict.</td></tr><tr><td><strong>STL Forecaster</strong> (Seasonal-Trend Decomposition Forecast)</td><td>Based on STL decomposition of time series, predicts trend, seasonal, and residual components separately, then combines them.</td><td>Data with complex seasonality, trend, and non-linear patterns, such as climate data, traffic flow.</td><td>1. Can handle non-fixed seasonality. 2. Robust to outliers. 3. After decomposition, other models can be combined to predict each component.</td></tr><tr><td><strong>Gaussian HMM</strong> (Gaussian Hidden Markov Model)</td><td>Assumes observed data is generated by hidden states, with each state's observation probability following a Gaussian distribution.</td><td>State sequence prediction or classification, such as speech recognition, financial state identification.</td><td>1. Suitable for time series state modeling. 2. Assumes observations are independent given the state. 3. Requires specifying the number of hidden states.</td></tr><tr><td><strong>GMM HMM</strong> (Gaussian Mixture Hidden Markov Model)</td><td>An extension of Gaussian HMM, where each state's observation probability is described by a Gaussian Mixture Model, capturing more complex observation distributions.</td><td>Scenarios requiring multi-modal observation distributions, such as complex action recognition, biosignal analysis.</td><td>1. More flexible than single Gaussian. 2. More parameters, higher computational complexity. 3. Requires training the number of GMM components.</td></tr><tr><td><strong>STRAY</strong> (Anomaly Detection based on Singular Value Decomposition)</td><td>Detects anomalies in high-dimensional data through Singular Value Decomposition (SVD), commonly used for time series anomaly detection.</td><td>High-dimensional time series anomaly detection, such as sensor networks, IT system monitoring.</td><td>1. No distribution assumption required. 2. Can handle high-dimensional data. 3. Sensitive to global anomalies, may miss local anomalies.</td></tr></tbody></table><p>Built-in time series large model introduction:</p><table><thead><tr><th>Model Name</th><th>Core Concept</th><th>Applicable Scenario</th><th>Main Features</th></tr></thead><tbody><tr><td><strong>Timer-XL</strong></td><td>Time series large model supporting ultra-long context, enhancing generalization capability through large-scale industrial data pre-training.</td><td>Complex industrial prediction requiring extremely long historical data, such as energy, aerospace, and transportation.</td><td>1. Ultra-long context support, can handle tens of thousands of time points as input. 2. Multi-scenario coverage, supports non-stationary, multi-variable, and covariate prediction. 3. Pre-trained on trillions of high-quality industrial time series data.</td></tr><tr><td><strong>Timer-Sundial</strong></td><td>A generative foundational model based on "Transformer + TimeFlow" architecture, focusing on probabilistic prediction.</td><td>Zero-shot prediction scenarios requiring quantification of uncertainty, such as finance, supply chain, and new energy power generation.</td><td>1. Strong zero-shot generalization capability, supports point prediction and probabilistic prediction. 2. Can flexibly analyze any statistical properties of the prediction distribution. 3. Innovative generative architecture, achieving efficient non-deterministic sample generation.</td></tr><tr><td><strong>Chronos-2</strong></td><td>A general time series foundational model based on discrete tokenization paradigm, converting prediction into language modeling tasks.</td><td>Rapid zero-shot univariate prediction, and scenarios that can leverage covariates (e.g., promotions, weather) to improve results.</td><td>1. Strong zero-shot probabilistic prediction capability. 2. Supports unified covariate modeling, but has strict input requirements: a. The set of names of future covariates must be a subset of the set of names of historical covariates; b. The length of each historical covariate must equal the length of the target variable; c. The length of each future covariate must equal the prediction length; 3. Uses an efficient encoder-style structure, balancing performance and inference speed.</td></tr></tbody></table><h3 id="_4-4-delete-models" tabindex="-1"><a class="header-anchor" href="#_4-4-delete-models"><span>4.4 Delete Models</span></a></h3><p>For registered models, users can delete them using SQL. AINode will delete the entire folder of the corresponding model in the user_defined directory. The SQL syntax is as follows:</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">DROP</span><span style="color:#ABB2BF;"> MODEL </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">model_id</span><span style="color:#56B6C2;">></span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>Specify the model_id that has been successfully registered to delete the corresponding model. Since model deletion involves model data cleanup, the operation will not be completed immediately, and the model state becomes DROPPING. Models in this state cannot be used for model inference. Note that this function does not support deleting built-in models.</p><h3 id="_4-5-load-unload-models" tabindex="-1"><a class="header-anchor" href="#_4-5-load-unload-models"><span>4.5 Load/Unload Models</span></a></h3><p>To adapt to different scenarios, AINode provides two model loading strategies:</p><ul><li><strong>On-demand loading</strong>: Load the model temporarily when inference is needed, and release resources after inference. Suitable for testing or low-load scenarios.</li><li><strong>Persistent loading</strong>: Permanently load the model into memory (CPU) or GPU memory to support high-concurrency inference. Users only need to specify the model to load or unload via SQL, and AINode will automatically manage the number of instances. The current state of persistent models can also be viewed at any time.</li></ul><p>The following details the loading/unloading model content:</p><ol><li>Configuration parameters</li></ol><p>Support editing the following configuration items to set persistent loading related parameters.</p><div class="language-properties line-numbers-mode" data-highlighter="shiki" data-ext="properties" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-properties"><span class="line"><span style="color:#7F848E;font-style:italic;"># The ratio of device memory/GPU memory available for AINode inference</span></span> |
| <span class="line"><span style="color:#7F848E;font-style:italic;"># Datatype: Float</span></span> |
| <span class="line"><span style="color:#C678DD;">ain_inference_memory_usage_ratio</span><span style="color:#ABB2BF;">=</span><span style="color:#98C379;">0.4</span></span> |
| <span class="line"></span> |
| <span class="line"><span style="color:#7F848E;font-style:italic;"># The memory ratio required by each instance of a loaded model, i.e., model memory * this value</span></span> |
| <span class="line"><span style="color:#7F848E;font-style:italic;"># Datatype: Float</span></span> |
| <span class="line"><span style="color:#C678DD;">ain_inference_extra_memory_ratio</span><span style="color:#ABB2BF;">=</span><span style="color:#98C379;">1.2</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><ol start="2"><li>View available devices</li></ol><p>Support using the following SQL command to view all available device IDs.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">SHOW AI_DEVICES</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>Example</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">IoTDB</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;"> show ai_devices</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| DeviceId|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| cpu|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">0</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">1</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><ol start="3"><li>Load models</li></ol><p>Support using the following SQL command to manually load models, and the system will <strong>automatically balance</strong> the number of model instances based on hardware resource usage.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">LOAD</span><span style="color:#ABB2BF;"> MODEL </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">existing_model_id</span><span style="color:#56B6C2;">></span><span style="color:#C678DD;"> TO</span><span style="color:#ABB2BF;"> DEVICES </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">(, </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">)*</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>Parameter requirements</p><ul><li><strong>existing_model_id:</strong> The specified model ID, currently only supports timer_xl and sundial.</li><li><strong>device_id:</strong> The location where the model is loaded. <ul><li><strong>cpu:</strong> Load to the memory of the AINode server.</li><li><strong>gpu_id:</strong> Load to the corresponding GPU of the AINode server, e.g., "0, 1" means loading to the two GPUs numbered 0 and 1.</li></ul></li></ul><p>Example</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">LOAD</span><span style="color:#ABB2BF;"> MODEL sundial </span><span style="color:#C678DD;">TO</span><span style="color:#ABB2BF;"> DEVICES </span><span style="color:#98C379;">'cpu,0,1'</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><ol start="4"><li>Unload models</li></ol><p>Support using the following SQL command to manually unload all instances of a specified model, and the system will <strong>redistribute</strong> the released resources to other models.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">UNLOAD</span><span style="color:#ABB2BF;"> MODEL </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">existing_model_id</span><span style="color:#56B6C2;">></span><span style="color:#C678DD;"> FROM</span><span style="color:#ABB2BF;"> DEVICES </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">(, </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">)*</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><p>Parameter requirements</p><ul><li><strong>existing_model_id:</strong> The specified model ID, currently only supports timer_xl and sundial.</li><li><strong>device_id:</strong> The location where the model is loaded. <ul><li><strong>cpu:</strong> Attempt to unload the specified model from the memory of the AINode server.</li><li><strong>gpu_id:</strong> Attempt to unload the specified model from the corresponding GPU of the AINode server, e.g., "0, 1" means attempting to unload from the two GPUs numbered 0 and 1.</li></ul></li></ul><p>Example</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#C678DD;">UNLOAD</span><span style="color:#ABB2BF;"> MODEL sundial </span><span style="color:#C678DD;">FROM</span><span style="color:#ABB2BF;"> DEVICES </span><span style="color:#98C379;">'cpu,0,1'</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div></div></div><ol start="5"><li>View loaded models</li></ol><p>Support using the following SQL command to view the models that have been manually loaded. You can specify the device with <code>device_id</code>.</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">SHOW LOADED MODELS</span></span> |
| <span class="line"><span style="color:#ABB2BF;">SHOW LOADED MODELS </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">(, </span><span style="color:#56B6C2;"><</span><span style="color:#ABB2BF;">device_id</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;">)* # View models </span><span style="color:#C678DD;">in</span><span style="color:#ABB2BF;"> specified devices</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div></div></div><p>Example: sundial model loaded in memory, gpu_0, and gpu_1</p><div class="language-sql line-numbers-mode" data-highlighter="shiki" data-ext="sql" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-sql"><span class="line"><span style="color:#ABB2BF;">IoTDB</span><span style="color:#56B6C2;">></span><span style="color:#ABB2BF;"> show loaded models</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+--------------+------------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| DeviceId| ModelId| </span><span style="color:#56B6C2;">Count</span><span style="color:#ABB2BF;">(instances)|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+--------------+------------------+</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| cpu| sundial| </span><span style="color:#D19A66;">4</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">0</span><span style="color:#ABB2BF;">| sundial| </span><span style="color:#D19A66;">6</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">| </span><span style="color:#D19A66;">1</span><span style="color:#ABB2BF;">| sundial| </span><span style="color:#D19A66;">6</span><span style="color:#ABB2BF;">|</span></span> |
| <span class="line"><span style="color:#ABB2BF;">+</span><span style="color:#7F848E;font-style:italic;">-------------+--------------+------------------+</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div><p>Notes:</p><ul><li>DeviceId: Device ID</li><li>ModelId: Loaded model ID</li><li>Count(instances): Number of model instances per device (automatically assigned by the system)</li></ul><h3 id="_4-6-introduction-to-time-series-large-models" tabindex="-1"><a class="header-anchor" href="#_4-6-introduction-to-time-series-large-models"><span>4.6 Introduction to Time Series Large Models</span></a></h3><p>AINode currently supports multiple time series large models. For related introductions and deployment usage, refer to <a class="route-link" href="/UserGuide/latest/AI-capability/TimeSeries-Large-Model_Upgrade_apache.html">Time Series Large Models</a></p><h2 id="_5-permission-management" tabindex="-1"><a class="header-anchor" href="#_5-permission-management"><span>5. Permission Management</span></a></h2><p>When using AINode-related features, permission management can be done using IoTDB's own authentication. Users can only use model management-related features if they have the USE_MODEL permission. When using inference features, users need to have access permissions to the source sequences corresponding to the SQL used for input.</p><table><thead><tr><th>Permission Name</th><th>Permission Scope</th><th>Administrator User (Default ROOT)</th><th>Normal User</th><th>Related to Path</th></tr></thead><tbody><tr><td>USE_MODEL</td><td>create model / show models / drop model</td><td>√</td><td>√</td><td>x</td></tr><tr><td>READ_DATA</td><td>call inference</td><td>√</td><td>√</td><td>√</td></tr></tbody></table><h2 id="_6-contributing-open-source-time-series-large-models-to-iotdb-ainode" tabindex="-1"><a class="header-anchor" href="#_6-contributing-open-source-time-series-large-models-to-iotdb-ainode"><span>6. Contributing Open Source Time Series Large Models to IoTDB-AINode</span></a></h2><p>Support adding new built-in models to AINode. The specific steps are as follows (using chronos2 as an example):</p><ul><li><strong>Send an email to the <a href="mailto:dev@iotdb.apache.org" target="_blank" rel="noopener noreferrer">dev@iotdb.apache.org</a> mailing list or submit an issue in the main repository to initiate preliminary discussion</strong></li><li><strong>Submit a Pull Request to the main branch</strong></li></ul><ol><li><p>Check the open source license used by the model and declare it in the IoTDB repository.</p></li><li><p>Create a new package for the new built-in model in iotdb-core/ainode/iotdb/ainode/core/model;</p><ol><li>Ensure it contains the model configuration class;</li><li>Ensure it contains the model class for executing inference tasks;</li><li>Ensure it contains the pipeline class inheriting from AINode inference task pipeline;</li></ol><div class="language-bash line-numbers-mode" data-highlighter="shiki" data-ext="bash" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-bash"><span class="line"><span style="color:#61AFEF;">root@rootMacBook-Pro</span><span style="color:#98C379;"> model</span><span style="color:#98C379;"> %</span><span style="color:#98C379;"> eza</span><span style="color:#D19A66;"> --tree</span><span style="color:#D19A66;"> --level=2</span><span style="color:#98C379;"> .</span></span> |
| <span class="line"><span style="color:#56B6C2;">.</span></span> |
| <span class="line"><span style="color:#61AFEF;">├──</span><span style="color:#98C379;"> chronos2</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> base.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> chronos_bolt.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> config.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> dataset.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> layers.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> model.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> pipeline_chronos2.py</span><span style="color:#7F848E;font-style:italic;"> # Inheriting AINode inference task pipeline</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> └──</span><span style="color:#98C379;"> utils.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">├──</span><span style="color:#98C379;"> sktime</span></span> |
| <span class="line"><span style="color:#61AFEF;">├──</span><span style="color:#98C379;"> sundial</span></span> |
| <span class="line"><span style="color:#61AFEF;">└──</span><span style="color:#98C379;"> timer_xl</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li><li><p>Add the metadata of the new model in iotdb-core/ainode/iotdb/ainode/core/model/model_info.py;</p><div class="language-python line-numbers-mode" data-highlighter="shiki" data-ext="python" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-python"><span class="line"><span style="color:#D19A66;">BUILTIN_HF_TRANSFORMERS_MODEL_MAP</span><span style="color:#56B6C2;"> =</span><span style="color:#ABB2BF;"> {</span></span> |
| <span class="line"><span style="color:#98C379;"> "chronos2"</span><span style="color:#ABB2BF;">: </span><span style="color:#61AFEF;">ModelInfo</span><span style="color:#ABB2BF;">(</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> model_id</span><span style="color:#56B6C2;">=</span><span style="color:#98C379;">"chronos2"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Unique identifier of the model</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> category</span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;">ModelCategory.</span><span style="color:#D19A66;">BUILTIN</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Model category, choose BUILTIN</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> state</span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;">ModelStates.</span><span style="color:#D19A66;">INACTIVE</span><span style="color:#ABB2BF;">,</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> model_type</span><span style="color:#56B6C2;">=</span><span style="color:#98C379;">"t5"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Model type, must not duplicate other builtin models</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> pipeline_cls</span><span style="color:#56B6C2;">=</span><span style="color:#98C379;">"pipeline_chronos2.Chronos2Pipeline"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Inheriting AINode inference pipeline</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> repo_id</span><span style="color:#56B6C2;">=</span><span style="color:#98C379;">"amazon/chronos-2"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># [Optional] Huggingface weights</span></span> |
| <span class="line"><span style="color:#E06C75;font-style:italic;"> auto_map</span><span style="color:#56B6C2;">=</span><span style="color:#ABB2BF;">{</span></span> |
| <span class="line"><span style="color:#98C379;"> "AutoConfig"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"config.Chronos2CoreConfig"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Ensure it points to the model's configuration class</span></span> |
| <span class="line"><span style="color:#98C379;"> "AutoModelForCausalLM"</span><span style="color:#ABB2BF;">: </span><span style="color:#98C379;">"model.Chronos2Model"</span><span style="color:#ABB2BF;">, </span><span style="color:#7F848E;font-style:italic;"># Ensure it points to the model class for inference</span></span> |
| <span class="line"><span style="color:#ABB2BF;"> },</span></span> |
| <span class="line"><span style="color:#ABB2BF;"> ),</span></span> |
| <span class="line"><span style="color:#ABB2BF;">}</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li><li><p>Add the corresponding model in integration-test/src/test/java/org/apache/iotdb/ainode/utils/AINodeTestUtils.java.</p><div class="language-java line-numbers-mode" data-highlighter="shiki" data-ext="java" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-java"><span class="line"><span style="color:#C678DD;">public</span><span style="color:#C678DD;"> static</span><span style="color:#C678DD;"> final</span><span style="color:#E5C07B;"> Map</span><span style="color:#ABB2BF;"><</span><span style="color:#E5C07B;">String</span><span style="color:#ABB2BF;">,</span><span style="color:#E5C07B;"> FakeModelInfo</span><span style="color:#ABB2BF;">></span><span style="color:#E06C75;"> BUILTIN_LTSM_MAP </span><span style="color:#56B6C2;">=</span></span> |
| <span class="line"><span style="color:#E5C07B;"> Stream</span><span style="color:#ABB2BF;">.</span><span style="color:#61AFEF;">of</span><span style="color:#ABB2BF;">(</span></span> |
| <span class="line"><span style="color:#C678DD;"> new</span><span style="color:#E5C07B;"> AbstractMap</span><span style="color:#ABB2BF;">.</span><span style="color:#E5C07B;">SimpleEntry</span><span style="color:#ABB2BF;"><>(</span></span> |
| <span class="line"><span style="color:#98C379;"> "sundial"</span><span style="color:#ABB2BF;">, </span><span style="color:#C678DD;">new</span><span style="color:#61AFEF;"> FakeModelInfo</span><span style="color:#ABB2BF;">(</span><span style="color:#98C379;">"sundial"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"sundial"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"builtin"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"active"</span><span style="color:#ABB2BF;">)),</span></span> |
| <span class="line"><span style="color:#C678DD;"> new</span><span style="color:#E5C07B;"> AbstractMap</span><span style="color:#ABB2BF;">.</span><span style="color:#E5C07B;">SimpleEntry</span><span style="color:#ABB2BF;"><>(</span></span> |
| <span class="line"><span style="color:#98C379;"> "timer_xl"</span><span style="color:#ABB2BF;">, </span><span style="color:#C678DD;">new</span><span style="color:#61AFEF;"> FakeModelInfo</span><span style="color:#ABB2BF;">(</span><span style="color:#98C379;">"timer_xl"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"timer"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"builtin"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"active"</span><span style="color:#ABB2BF;">)),</span></span> |
| <span class="line"><span style="color:#C678DD;"> new</span><span style="color:#E5C07B;"> AbstractMap</span><span style="color:#ABB2BF;">.</span><span style="color:#E5C07B;">SimpleEntry</span><span style="color:#ABB2BF;"><>(</span></span> |
| <span class="line"><span style="color:#98C379;"> "chronos2"</span><span style="color:#ABB2BF;">, </span><span style="color:#C678DD;">new</span><span style="color:#61AFEF;"> FakeModelInfo</span><span style="color:#ABB2BF;">(</span><span style="color:#98C379;">"chronos2"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"t5"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"builtin"</span><span style="color:#ABB2BF;">, </span><span style="color:#98C379;">"active"</span><span style="color:#ABB2BF;">)))</span></span> |
| <span class="line"><span style="color:#ABB2BF;"> .</span><span style="color:#61AFEF;">collect</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;">Collectors</span><span style="color:#ABB2BF;">.</span><span style="color:#61AFEF;">toMap</span><span style="color:#ABB2BF;">(</span><span style="color:#E5C07B;">Map</span><span style="color:#ABB2BF;">.</span><span style="color:#E5C07B;">Entry</span><span style="color:#C678DD;">::</span><span style="color:#ABB2BF;">getKey, </span><span style="color:#E5C07B;">Map</span><span style="color:#ABB2BF;">.</span><span style="color:#E5C07B;">Entry</span><span style="color:#C678DD;">::</span><span style="color:#ABB2BF;">getValue));</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li><li><p>Submit a pull request, and only merge after passing the pipeline tests.</p></li></ol><ul><li><strong>Packaging and Deployment</strong></li></ul><ol><li><p>Compile AINode to obtain the installation package.</p><div class="language-bash line-numbers-mode" data-highlighter="shiki" data-ext="bash" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-bash"><span class="line"><span style="color:#7F848E;font-style:italic;"># Command to simultaneously build IoTDB and AINode</span></span> |
| <span class="line"><span style="color:#61AFEF;">mvn</span><span style="color:#98C379;"> clean</span><span style="color:#98C379;"> package</span><span style="color:#D19A66;"> -pl</span><span style="color:#98C379;"> distribution</span><span style="color:#D19A66;"> -P</span><span style="color:#98C379;"> with-ainode</span><span style="color:#D19A66;"> -am</span><span style="color:#D19A66;"> -DskipTests</span></span> |
| <span class="line"><span style="color:#7F848E;font-style:italic;"># Command to build AINode only</span></span> |
| <span class="line"><span style="color:#61AFEF;">mvn</span><span style="color:#98C379;"> clean</span><span style="color:#98C379;"> package</span><span style="color:#D19A66;"> -pl</span><span style="color:#98C379;"> iotdb-core/ainode</span><span style="color:#D19A66;"> -P</span><span style="color:#98C379;"> with-ainode</span><span style="color:#D19A66;"> -am</span><span style="color:#D19A66;"> -DskipTests</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li><li><p>Ensure the model can obtain weights at runtime.</p><ol><li>If the model weights can be obtained online from Huggingface, ensure <code>repo_id</code> is filled in;</li><li>Otherwise, manually place the model weights in <code>data/ainode/models/builtin/<model_id></code> before starting AINode or during runtime.</li></ol><div class="language-bash line-numbers-mode" data-highlighter="shiki" data-ext="bash" style="background-color:#282c34;color:#abb2bf;"><pre class="shiki one-dark-pro vp-code"><code class="language-bash"><span class="line"><span style="color:#61AFEF;">root@rootMacBook-Pro</span><span style="color:#98C379;"> models</span><span style="color:#98C379;"> %</span><span style="color:#98C379;"> eza</span><span style="color:#D19A66;"> --tree</span><span style="color:#D19A66;"> --level=3</span><span style="color:#98C379;"> .</span></span> |
| <span class="line"><span style="color:#56B6C2;">.</span></span> |
| <span class="line"><span style="color:#61AFEF;">├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">├──</span><span style="color:#98C379;"> builtin</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> chronos2</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> config.json</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> └──</span><span style="color:#98C379;"> model.safetensors</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> sundial</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> config.json</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> │</span><span style="color:#98C379;"> └──</span><span style="color:#98C379;"> model.safetensors</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> └──</span><span style="color:#98C379;"> timer_xl</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> __init__.py</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> ├──</span><span style="color:#98C379;"> config.json</span></span> |
| <span class="line"><span style="color:#61AFEF;">│</span><span style="color:#98C379;"> └──</span><span style="color:#98C379;"> model.safetensors</span></span> |
| <span class="line"><span style="color:#61AFEF;">└──</span><span style="color:#98C379;"> user_defined</span></span></code></pre><div class="line-numbers" aria-hidden="true" style="counter-reset:line-number 0;"><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div><div class="line-number"></div></div></div></li></ol><ul><li><strong>Full Example</strong></li></ul><p>Full example can be referenced at <a href="https://github.com/apache/iotdb/pull/16903" target="_blank" rel="noopener noreferrer">https://github.com/apache/iotdb/pull/16903</a></p></div><!----><!----><!----></div><footer class="vp-page-meta"><div class="vp-meta-item edit-link"><a class="auto-link external-link vp-meta-label" href="https://github.com/apache/iotdb-docs/edit/main/src/UserGuide/latest/AI-capability/AINode_Upgrade_apache.md" aria-label="Found Error? Edit this page on GitHub" rel="noopener noreferrer" target="_blank"><!--[--><svg xmlns="http://www.w3.org/2000/svg" class="icon edit-icon" viewBox="0 0 1024 1024" fill="currentColor" aria-label="edit icon" name="edit"><path d="M430.818 653.65a60.46 60.46 0 0 1-50.96-93.281l71.69-114.012 7.773-10.365L816.038 80.138A60.46 60.46 0 0 1 859.225 62a60.46 60.46 0 0 1 43.186 18.138l43.186 43.186a60.46 60.46 0 0 1 0 86.373L588.879 565.55l-8.637 8.637-117.466 68.234a60.46 60.46 0 0 1-31.958 11.229z"></path><path d="M728.802 962H252.891A190.883 190.883 0 0 1 62.008 771.98V296.934a190.883 190.883 0 0 1 190.883-192.61h267.754a60.46 60.46 0 0 1 0 120.92H252.891a69.962 69.962 0 0 0-69.098 69.099V771.98a69.962 69.962 0 0 0 69.098 69.098h475.911A69.962 69.962 0 0 0 797.9 771.98V503.363a60.46 60.46 0 1 1 120.922 0V771.98A190.883 190.883 0 0 1 728.802 962z"></path></svg><!--]-->Found Error? Edit this page on GitHub<!----></a></div><div class="vp-meta-item git-info"><div class="update-time"><span class="vp-meta-label">Last Updated: </span><time class="vp-meta-info" datetime="2026-02-04T04:01:34.000Z" data-allow-mismatch>2/4/26, 4:01 AM</time></div><!----></div></footer><nav class="vp-page-nav"><a class="route-link auto-link prev" href="/UserGuide/latest/AI-capability/AINode_apache.html" aria-label="AINode(V2.0.5/6)"><div class="hint"><span class="arrow start"></span>Prev</div><div class="link"><!---->AINode(V2.0.5/6)</div></a><a class="route-link auto-link next" href="/UserGuide/latest/AI-capability/TimeSeries-Large-Model.html" aria-label="TimeSeries Large Model(V2.0.5/6)"><div class="hint">Next<span class="arrow end"></span></div><div class="link">TimeSeries Large Model(V2.0.5/6)<!----></div></a></nav><!----><!----><!--]--></main><!--]--><footer class="site-footer"><span id="doc-version" style="display:none;">latest</span><p class="copyright-text">Copyright © 2026 The Apache Software Foundation. |
| Apache IoTDB, IoTDB, Apache, the Apache feather logo, and the Apache IoTDB project logo are either registered trademarks or trademarks of The Apache Software Foundation in all countries</p><p style="text-align:center;margin-top:10px;color:#909399;font-size:12px;margin:0 30px;"><strong>Having questions?</strong> Connect with us on QQ, WeChat, or Slack. <a href="https://github.com/apache/iotdb/issues/1995">Join the community</a></p></footer></div><!--]--><!--]--><!--[--><!----><!--]--><!--]--></div> |
| <script type="module" src="/assets/app-pmkzp6TS.js" defer></script> |
| </body> |
| </html> |
