Home Explore Help
Register Sign In
zengli
/
coding-agent
1
0
Fork 1
Files Issues 0 Pull Requests 0 Wiki
Tree: 19cbf6a852
Branches Tags
coding-agent
/
GraphRAG增强版_20260920
/
course-source
/
legacy-html
/
④ 基于Milvus库的问答系统
/
4.6 prompts设计与query意图识别.html

4.6 prompts设计与query意图识别.html 89 KB
History Raw

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130 
  1. <!doctype html>
    2. 

  2. <html lang="zh" class="no-js">
    3. 

  3.   <head>
    4. 

  4.     
    5. 

  5.       <meta charset="utf-8">
    6. 

  6.       <meta name="viewport" content="width=device-width,initial-scale=1">
    7. 

  7.       
    8. 

  8.       
    9. 

  9.       
    10. 

  10.       
    11. 

  11.         <link rel="prev" href="4.5%20BERT%E5%9F%BA%E7%A1%80%E4%B8%8E%E4%BB%BB%E5%8A%A1%E5%BE%AE%E8%B0%83.html">
    12. 

  12.       
    13. 

  13.       
    14. 

  14.         <link rel="next" href="4.7%20%E6%A3%80%E7%B4%A2%E7%AD%96%E7%95%A5%E4%B8%8ERAG%E7%B3%BB%E7%BB%9F%E8%AE%BE%E8%AE%A1.html">
    15. 

  15.       
    16. 

  16.       
    17. 

  17.       <link rel="icon" href="../img/AI.jpg">
    18. 

  18.       <meta name="generator" content="mkdocs-1.6.1, mkdocs-material-9.6.21">
    19. 

  19.     
    20. 

  20.     
    21. 

  21.       
    22. 

  22.         <title>4.6 prompts设计与query意图识别 - 企业知识库RAG智能问答项目V1.0</title>
    23. 

  23.       
    24. 

  24.     
    25. 

  25.     
    26. 

  26.       <link rel="stylesheet" href="../assets/stylesheets/main.2a3383ac.min.css">
    27. 

  27.       
    28. 

  28.       
    29. 

  29. 

  30. 

  31.     
    32. 

  32.     
    33. 

  33.       
    34. 

  34.     
    35. 

  35.     
    36. 

  36.       
    37. 

  37.         
    38. 

  38.         
    39. 

  39.         <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
    40. 

  40.         <link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Roboto:300,300i,400,400i,700,700i%7CRoboto+Mono:400,400i,700,700i&amp;display=fallback">
    41. 

  41.         <style>:root{--md-text-font:"Roboto";--md-code-font:"Roboto Mono"}</style>
    42. 

  42.       
    43. 

  43.     
    44. 

  44.     
    45. 

  45.       <link rel="stylesheet" href="../stylesheets/extra.css">
    46. 

  46.     
    47. 

  47.     <script>__md_scope=new URL("..",location),__md_hash=e=>[...e].reduce(((e,_)=>(e<<5)-e+_.charCodeAt(0)),0),__md_get=(e,_=localStorage,t=__md_scope)=>JSON.parse(_.getItem(t.pathname+"."+e)),__md_set=(e,_,t=localStorage,a=__md_scope)=>{try{t.setItem(a.pathname+"."+e,JSON.stringify(_))}catch(e){}}</script>
    48. 

  48.     
    49. 

  49.       
    50. 

  50. 

  51.     
    52. 

  52.     
    53. 

  53.     
    54. 

  54.   </head>
    55. 

  55.   
    56. 

  56.   
    57. 

  57.     <body dir="ltr">
    58. 

  58.   
    59. 

  59.     
    60. 

  60.     <input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
    61. 

  61.     <input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
    62. 

  62.     <label class="md-overlay" for="__drawer"></label>
    63. 

  63.     <div data-md-component="skip">
    64. 

  64.       
    65. 

  65.         
    66. 

  66.         <a href="#_1" class="md-skip">
    67. 

  67.           跳转至
    68. 

  68.         </a>
    69. 

  69.       
    70. 

  70.     </div>
    71. 

  71.     <div data-md-component="announce">
    72. 

  72.       
    73. 

  73.     </div>
    74. 

  74.     
    75. 

  75.     
    76. 

  76.       
    77. 

  77. 

  78.   
    79. 

  79. 

  80. <header class="md-header md-header--shadow" data-md-component="header">
    81. 

  81.   <nav class="md-header__inner md-grid" aria-label="页眉">
    82. 

  82.     <a href="../index.html" title="企业知识库RAG智能问答项目V1.0" class="md-header__button md-logo" aria-label="企业知识库RAG智能问答项目V1.0" data-md-component="logo">
    83. 

  83.       
    84. 

  84.   <img src="../img/AI.jpg" alt="logo">
    85. 

  85. 

  86.     </a>
    87. 

  87.     <label class="md-header__button md-icon" for="__drawer">
    88. 

  88.       
    89. 

  89.       <svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24"><path d="M3 6h18v2H3zm0 5h18v2H3zm0 5h18v2H3z"></path></svg>
    90. 

  90.     </label>
    91. 

  91.     <div class="md-header__title" data-md-component="header-title">
    92. 

  92.       <div class="md-header__ellipsis">
    93. 

  93.         <div class="md-header__topic">
    94. 

  94.           <span class="md-ellipsis">
    95. 

  95.             企业知识库RAG智能问答项目V1.0
    96. 

  96.           </span>
    97. 

  97.         </div>
    98. 

  98.         <div class="md-header__topic" data-md-component="header-topic">
    99. 

  99.           <span class="md-ellipsis">
    100. 

  100.             
    101. 

  101.               4.6 prompts设计与query意图识别
    102. 

  102.             
    103. 

  103.           </span>
    104. 

  104.         </div>
    105. 

  105.       </div>
    106. 

  106.     </div>
    107. 

  107.     
    108. 

  108.     
    109. 

  109.       <script>var palette=__md_get("__palette");if(palette&&palette.color){if("(prefers-color-scheme)"===palette.color.media){var media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']");palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent")}for(var[key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
    110. 

  110.     
    111. 

  111.     
    112. 

  112.     
    113. 

  113.       
    114. 

  114.       
    115. 

  115.         <label class="md-header__button md-icon" for="__search">
    116. 

  116.           
    117. 

  117.           <svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0 1 16 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.52 6.52 0 0 1 9.5 16 6.5 6.5 0 0 1 3 9.5 6.5 6.5 0 0 1 9.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5"></path></svg>
    118. 

  118.         </label>
    119. 

  119.         <div class="md-search" data-md-component="search" role="dialog">
    120. 

  120.   <label class="md-search__overlay" for="__search"></label>
    121. 

  121.   <div class="md-search__inner" role="search">
    122. 

  122.     <form class="md-search__form" name="search">
    123. 

  123.       <input type="text" class="md-search__input" name="query" aria-label="搜索" placeholder="搜索" autocapitalize="off" autocorrect="off" autocomplete="off" spellcheck="false" data-md-component="search-query" required>
    124. 

  124.       <label class="md-search__icon md-icon" for="__search">
    125. 

  125.         
    126. 

  126.         <svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0 1 16 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.52 6.52 0 0 1 9.5 16 6.5 6.5 0 0 1 3 9.5 6.5 6.5 0 0 1 9.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5"></path></svg>
    127. 

  127.         
    128. 

  128.         <svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24"><path d="M20 11v2H8l5.5 5.5-1.42 1.42L4.16 12l7.92-7.92L13.5 5.5 8 11z"></path></svg>
    129. 

  129.       </label>
    130. 

  130.       <nav class="md-search__options" aria-label="查找">
    131. 

  131.         
    132. 

  132.         <button type="reset" class="md-search__icon md-icon" title="清空当前内容" aria-label="清空当前内容" tabindex="-1">
    133. 

  133.           
    134. 

  134.           <svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24"><path d="M19 6.41 17.59 5 12 10.59 6.41 5 5 6.41 10.59 12 5 17.59 6.41 19 12 13.41 17.59 19 19 17.59 13.41 12z"></path></svg>
    135. 

  135.         </button>
    136. 

  136.       </nav>
    137. 

  137.       
    138. 

  138.     </form>
    139. 

  139.     <div class="md-search__output">
    140. 

  140.       <div class="md-search__scrollwrap" tabindex="0" data-md-scrollfix>
    141. 

  141.         <div class="md-search-result" data-md-component="search-result">
    142. 

  142.           <div class="md-search-result__meta">
    143. 

  143.             正在初始化搜索引擎
    144. 

  144.           </div>
    145. 

  145.           <ol class="md-search-result__list" role="presentation"></ol>
    146. 

  146.         </div>
    147. 

  147.       </div>
    148. 

  148.     </div>
    149. 

  149.   </div>
    150. 

  150. </div>
    151. 

  151.       
    152. 

  152.     
    153. 

  153.     
    154. 

  154.   </nav>
    155. 

  155.   
    156. 

  156. </header>
    157. 

  157.     
    158. 

  158.     <div class="md-container" data-md-component="container">
    159. 

  159.       
    160. 

  160.       
    161. 

  161.         
    162. 

  162.           
    163. 

  163.         
    164. 

  164.       
    165. 

  165.       <main class="md-main" data-md-component="main">
    166. 

  166.         <div class="md-main__inner md-grid">
    167. 

  167.           
    168. 

  168.             
    169. 

  169.               
    170. 

  170.               <div class="md-sidebar md-sidebar--primary" data-md-component="sidebar" data-md-type="navigation">
    171. 

  171.                 <div class="md-sidebar__scrollwrap">
    172. 

  172.                   <div class="md-sidebar__inner">
    173. 

  173.                     
    174. 

  174. 

  175. 

  176. 

  177. <nav class="md-nav md-nav--primary" aria-label="导航栏" data-md-level="0">
    178. 

  178.   <label class="md-nav__title" for="__drawer">
    179. 

  179.     <a href="../index.html" title="企业知识库RAG智能问答项目V1.0" class="md-nav__button md-logo" aria-label="企业知识库RAG智能问答项目V1.0" data-md-component="logo">
    180. 

  180.       
    181. 

  181.   <img src="../img/AI.jpg" alt="logo">
    182. 

  182. 

  183.     </a>
    184. 

  184.     企业知识库RAG智能问答项目V1.0
    185. 

  185.   </label>
    186. 

  186.   
    187. 

  187.   <ul class="md-nav__list" data-md-scrollfix>
    188. 

  188.     
    189. 

  189.       
    190. 

  190.       
    191. 

  191.   
    192. 

  192.   
    193. 

  193.   
    194. 

  194.   
    195. 

  195.     <li class="md-nav__item">
    196. 

  196.       <a href="../index.html" class="md-nav__link">
    197. 

  197.         
    198. 

  198.   
    199. 

  199.   
    200. 

  200.   <span class="md-ellipsis">
    201. 

  201.     知识库内容
    202. 

  202.     
    203. 

  203.   </span>
    204. 

  204.   
    205. 

  205. 

  206.       </a>
    207. 

  207.     </li>
    208. 

  208.   
    209. 

  209. 

  210.     
    211. 

  211.       
    212. 

  212.       
    213. 

  213.   
    214. 

  214.   
    215. 

  215.   
    216. 

  216.   
    217. 

  217.     
    218. 

  218.     
    219. 

  219.     
    220. 

  220.     
    221. 

  221.     
    222. 

  222.     <li class="md-nav__item md-nav__item--nested">
    223. 

  223.       
    224. 

  224.         
    225. 

  225.         
    226. 

  226.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_2">
    227. 

  227.         
    228. 

  228.           
    229. 

  229.           <label class="md-nav__link" for="__nav_2" id="__nav_2_label" tabindex="0">
    230. 

  230.             
    231. 

  231.   
    232. 

  232.   
    233. 

  233.   <span class="md-ellipsis">
    234. 

  234.     ① 项目概述
    235. 

  235.     
    236. 

  236.   </span>
    237. 

  237.   
    238. 

  238. 

  239.             <span class="md-nav__icon md-icon"></span>
    240. 

  240.           </label>
    241. 

  241.         
    242. 

  242.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_2_label" aria-expanded="false">
    243. 

  243.           <label class="md-nav__title" for="__nav_2">
    244. 

  244.             <span class="md-nav__icon md-icon"></span>
    245. 

  245.             ① 项目概述
    246. 

  246.           </label>
    247. 

  247.           <ul class="md-nav__list" data-md-scrollfix>
    248. 

  248.             
    249. 

  249.               
    250. 

  250.                 
    251. 

  251.   
    252. 

  252.   
    253. 

  253.   
    254. 

  254.   
    255. 

  255.     <li class="md-nav__item">
    256. 

  256.       <a href="../%E2%91%A0%20%E9%A1%B9%E7%9B%AE%E6%A6%82%E8%BF%B0/1.1%20%E9%A1%B9%E7%9B%AE%E6%9E%B6%E6%9E%84.html" class="md-nav__link">
    257. 

  257.         
    258. 

  258.   
    259. 

  259.   
    260. 

  260.   <span class="md-ellipsis">
    261. 

  261.     1.1 项目架构
    262. 

  262.     
    263. 

  263.   </span>
    264. 

  264.   
    265. 

  265. 

  266.       </a>
    267. 

  267.     </li>
    268. 

  268.   
    269. 

  269. 

  270.               
    271. 

  271.             
    272. 

  272.               
    273. 

  273.                 
    274. 

  274.   
    275. 

  275.   
    276. 

  276.   
    277. 

  277.   
    278. 

  278.     <li class="md-nav__item">
    279. 

  279.       <a href="../%E2%91%A0%20%E9%A1%B9%E7%9B%AE%E6%A6%82%E8%BF%B0/1.2%20%E9%A1%B9%E7%9B%AE%E7%8E%AF%E5%A2%83%28window%29.html" class="md-nav__link">
    280. 

  280.         
    281. 

  281.   
    282. 

  282.   
    283. 

  283.   <span class="md-ellipsis">
    284. 

  284.     1.2 项目环境(window)
    285. 

  285.     
    286. 

  286.   </span>
    287. 

  287.   
    288. 

  288. 

  289.       </a>
    290. 

  290.     </li>
    291. 

  291.   
    292. 

  292. 

  293.               
    294. 

  294.             
    295. 

  295.               
    296. 

  296.                 
    297. 

  297.   
    298. 

  298.   
    299. 

  299.   
    300. 

  300.   
    301. 

  301.     <li class="md-nav__item">
    302. 

  302.       <a href="../%E2%91%A0%20%E9%A1%B9%E7%9B%AE%E6%A6%82%E8%BF%B0/1.3%20%E9%A1%B9%E7%9B%AE%E7%8E%AF%E5%A2%83%28mac%29.html" class="md-nav__link">
    303. 

  303.         
    304. 

  304.   
    305. 

  305.   
    306. 

  306.   <span class="md-ellipsis">
    307. 

  307.     1.3 项目环境(mac)
    308. 

  308.     
    309. 

  309.   </span>
    310. 

  310.   
    311. 

  311. 

  312.       </a>
    313. 

  313.     </li>
    314. 

  314.   
    315. 

  315. 

  316.               
    317. 

  317.             
    318. 

  318.           </ul>
    319. 

  319.         </nav>
    320. 

  320.       
    321. 

  321.     </li>
    322. 

  322.   
    323. 

  323. 

  324.     
    325. 

  325.       
    326. 

  326.       
    327. 

  327.   
    328. 

  328.   
    329. 

  329.   
    330. 

  330.   
    331. 

  331.     
    332. 

  332.     
    333. 

  333.     
    334. 

  334.     
    335. 

  335.     
    336. 

  336.     <li class="md-nav__item md-nav__item--nested">
    337. 

  337.       
    338. 

  338.         
    339. 

  339.         
    340. 

  340.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3">
    341. 

  341.         
    342. 

  342.           
    343. 

  343.           <label class="md-nav__link" for="__nav_3" id="__nav_3_label" tabindex="0">
    344. 

  344.             
    345. 

  345.   
    346. 

  346.   
    347. 

  347.   <span class="md-ellipsis">
    348. 

  348.     ② 项目工具
    349. 

  349.     
    350. 

  350.   </span>
    351. 

  351.   
    352. 

  352. 

  353.             <span class="md-nav__icon md-icon"></span>
    354. 

  354.           </label>
    355. 

  355.         
    356. 

  356.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_3_label" aria-expanded="false">
    357. 

  357.           <label class="md-nav__title" for="__nav_3">
    358. 

  358.             <span class="md-nav__icon md-icon"></span>
    359. 

  359.             ② 项目工具
    360. 

  360.           </label>
    361. 

  361.           <ul class="md-nav__list" data-md-scrollfix>
    362. 

  362.               
    363. 

  363.                 
    364. 

  364.   
    365. 

  365.   
    366. 

  366.   
    367. 

  367.     <li class="md-nav__item">
    368. 

  368.       <a href="../%E2%91%A1%20%E9%A1%B9%E7%9B%AE%E5%B7%A5%E5%85%B7/2.1%20Ollama%E5%A4%A7%E6%A8%A1%E5%9E%8B%E9%AB%98%E6%95%88%E7%AE%A1%E7%90%86%E5%B7%A5%E5%85%B7.html" class="md-nav__link">
    369. 

  369.         
    370. 

  370.   
    371. 

  371.   
    372. 

  372.   <span class="md-ellipsis">
    373. 

  373.     2.1 Ollama大模型高效管理工具
    374. 

  374.     
    375. 

  375.   </span>
    376. 

  376.   
    377. 

  377. 

  378.       </a>
    379. 

  379.     </li>
    380. 

  380.   
    381. 

  381. 

  382.               
    383. 

  383.             
    384. 

  384.               
    385. 

  385.                 
    386. 

  386.   
    387. 

  387.   
    388. 

  388.   
    389. 

  389.     <li class="md-nav__item">
    390. 

  390.       <a href="../%E2%91%A1%20%E9%A1%B9%E7%9B%AE%E5%B7%A5%E5%85%B7/2.2%20LangChain%E5%9F%BA%E7%A1%80%E7%9F%A5%E8%AF%86.md.html" class="md-nav__link">
    391. 

  391.         
    392. 

  392.   
    393. 

  393.   
    394. 

  394.   <span class="md-ellipsis">
    395. 

  395.     2.2 LangChain基础知识.md
    396. 

  396.     
    397. 

  397.   </span>
    398. 

  398.   
    399. 

  399. 

  400.       </a>
    401. 

  401.     </li>
    402. 

  402.   
    403. 

  403. 

  404.               
    405. 

  405.             
    406. 

  406.               
    407. 

  407.                 
    408. 

  408.   
    409. 

  409.   
    410. 

  410.   
    411. 

  411.     <li class="md-nav__item">
    412. 

  412.       <a href="../%E2%91%A1%20%E9%A1%B9%E7%9B%AE%E5%B7%A5%E5%85%B7/2.3%20%E5%90%91%E9%87%8F%E7%B4%A2%E5%BC%95%EF%BC%9AHNSW%E3%80%81IVF%E4%B8%8EIVF_PQ.html" class="md-nav__link">
    413. 

  413.         
    414. 

  414.   
    415. 

  415.   
    416. 

  416.   <span class="md-ellipsis">
    417. 

  417.     2.3 向量索引：HNSW、IVF与IVF_PQ
    418. 

  418.     
    419. 

  419.   </span>
    420. 

  420.   
    421. 

  421. 

  422.       </a>
    423. 

  423.     </li>
    424. 

  424.   
    425. 

  425. 

  426.               
    427. 

  427.             
    428. 

  428.               
    429. 

  429.                 
    430. 

  430.   
    431. 

  431.   
    432. 

  432.   
    433. 

  433.     <li class="md-nav__item">
    434. 

  434.       <a href="../%E2%91%A1%20%E9%A1%B9%E7%9B%AE%E5%B7%A5%E5%85%B7/2.4%20Milvus%E5%90%91%E9%87%8F%E6%95%B0%E6%8D%AE%E5%BA%93.html" class="md-nav__link">
    435. 

  435.         
    436. 

  436.   
    437. 

  437.   
    438. 

  438.   <span class="md-ellipsis">
    439. 

  439.     2.4 Milvus向量数据库
    440. 

  440.     
    441. 

  441.   </span>
    442. 

  442.   
    443. 

  443. 

  444.       </a>
    445. 

  445.     </li>
    446. 

  446.   
    447. 

  447. 

  448.               
    449. 

  449.             
    450. 

  450.           </ul>
    451. 

  451.         </nav>
    452. 

  452.       
    453. 

  453.     </li>
    454. 

  454.   
    455. 

  455. 

  456.     
    457. 

  457.       
    458. 

  458.       
    459. 

  459.   
    460. 

  460.   
    461. 

  461.   
    462. 

  462.   
    463. 

  463.     
    464. 

  464.     
    465. 

  465.     
    466. 

  466.     
    467. 

  467.     
    468. 

  468.     <li class="md-nav__item md-nav__item--nested">
    469. 

  469.       
    470. 

  470.         
    471. 

  471.         
    472. 

  472.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_4">
    473. 

  473.         
    474. 

  474.           
    475. 

  475.           <label class="md-nav__link" for="__nav_4" id="__nav_4_label" tabindex="0">
    476. 

  476.             
    477. 

  477.   
    478. 

  478.   
    479. 

  479.   <span class="md-ellipsis">
    480. 

  480.     ③ 基于Mysql库的问答系统
    481. 

  481.     
    482. 

  482.   </span>
    483. 

  483.   
    484. 

  484. 

  485.             <span class="md-nav__icon md-icon"></span>
    486. 

  486.           </label>
    487. 

  487.         
    488. 

  488.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_4_label" aria-expanded="false">
    489. 

  489.           <label class="md-nav__title" for="__nav_4">
    490. 

  490.             <span class="md-nav__icon md-icon"></span>
    491. 

  491.             ③ 基于Mysql库的问答系统
    492. 

  492.           </label>
    493. 

  493.           <ul class="md-nav__list" data-md-scrollfix>
    494. 

  494.             
    495. 

  495.               
    496. 

  496.                 
    497. 

  497.   
    498. 

  498.   
    499. 

  499.   
    500. 

  500.   
    501. 

  501.     <li class="md-nav__item">
    502. 

  502.       <a href="../%E2%91%A2%20%E5%9F%BA%E4%BA%8EMysql%E5%BA%93%E7%9A%84%E9%97%AE%E7%AD%94%E7%B3%BB%E7%BB%9F/3.1%20Python%E6%97%A5%E5%BF%97%E4%BB%8B%E7%BB%8D%E4%B8%8E%E5%BA%94%E7%94%A8.html" class="md-nav__link">
    503. 

  503.         
    504. 

  504.   
    505. 

  505.   
    506. 

  506.   <span class="md-ellipsis">
    507. 

  507.     3.1 Python日志介绍与应用
    508. 

  508.     
    509. 

  509.   </span>
    510. 

  510.   
    511. 

  511. 

  512.       </a>
    513. 

  513.     </li>
    514. 

  514.   
    515. 

  515. 

  516.               
    517. 

  517.             
    518. 

  518.               
    519. 

  519.                 
    520. 

  520.   
    521. 

  521.   
    522. 

  522.   
    523. 

  523.   
    524. 

  524.     <li class="md-nav__item">
    525. 

  525.       <a href="../%E2%91%A2%20%E5%9F%BA%E4%BA%8EMysql%E5%BA%93%E7%9A%84%E9%97%AE%E7%AD%94%E7%B3%BB%E7%BB%9F/3.2%20Redis%E6%95%B0%E6%8D%AE%E5%BA%93%E7%AE%80%E4%BB%8B%E4%B8%8E%E4%BD%BF%E7%94%A8.html" class="md-nav__link">
    526. 

  526.         
    527. 

  527.   
    528. 

  528.   
    529. 

  529.   <span class="md-ellipsis">
    530. 

  530.     3.2 Redis数据库简介与使用
    531. 

  531.     
    532. 

  532.   </span>
    533. 

  533.   
    534. 

  534. 

  535.       </a>
    536. 

  536.     </li>
    537. 

  537.   
    538. 

  538. 

  539.               
    540. 

  540.             
    541. 

  541.               
    542. 

  542.                 
    543. 

  543.   
    544. 

  544.   
    545. 

  545.   
    546. 

  546.   
    547. 

  547.     <li class="md-nav__item">
    548. 

  548.       <a href="../%E2%91%A2%20%E5%9F%BA%E4%BA%8EMysql%E5%BA%93%E7%9A%84%E9%97%AE%E7%AD%94%E7%B3%BB%E7%BB%9F/3.3%20%E5%9F%BA%E4%BA%8EMysql%E6%95%B0%E6%8D%AE%E5%BA%93%E5%AE%9E%E7%8E%B0%E9%97%AE%E7%AD%94%E7%B3%BB%E7%BB%9F.html" class="md-nav__link">
    549. 

  549.         
    550. 

  550.   
    551. 

  551.   
    552. 

  552.   <span class="md-ellipsis">
    553. 

  553.     3.3 基于Mysql数据库实现问答系统
    554. 

  554.     
    555. 

  555.   </span>
    556. 

  556.   
    557. 

  557. 

  558.       </a>
    559. 

  559.     </li>
    560. 

  560.   
    561. 

  561. 

  562.               
    563. 

  563.             
    564. 

  564.               
    565. 

  565.                 
    566. 

  566.   
    567. 

  567.   
    568. 

  568.   
    569. 

  569.   
    570. 

  570.     <li class="md-nav__item">
    571. 

  571.       <a href="../%E2%91%A2%20%E5%9F%BA%E4%BA%8EMysql%E5%BA%93%E7%9A%84%E9%97%AE%E7%AD%94%E7%B3%BB%E7%BB%9F/3.4%20BM25%E7%AE%97%E6%B3%95%E7%AE%80%E4%BB%8B%E4%B8%8E%E5%BA%94%E7%94%A8.html" class="md-nav__link">
    572. 

  572.         
    573. 

  573.   
    574. 

  574.   
    575. 

  575.   <span class="md-ellipsis">
    576. 

  576.     3.4 BM25算法简介与应用
    577. 

  577.     
    578. 

  578.   </span>
    579. 

  579.   
    580. 

  580. 

  581.       </a>
    582. 

  582.     </li>
    583. 

  583.   
    584. 

  584. 

  585.               
    586. 

  586.             
    587. 

  587.           </ul>
    588. 

  588.         </nav>
    589. 

  589.       
    590. 

  590.     </li>
    591. 

  591.   
    592. 

  592. 

  593.     
    594. 

  594.       
    595. 

  595.       
    596. 

  596.   
    597. 

  597.   
    598. 

  598.     
    599. 

  599.   
    600. 

  600.   
    601. 

  601.   
    602. 

  602.     
    603. 

  603.     
    604. 

  604.     
    605. 

  605.     
    606. 

  606.     
    607. 

  607.     <li class="md-nav__item md-nav__item--active md-nav__item--nested">
    608. 

  608.       
    609. 

  609.         
    610. 

  610.         
    611. 

  611.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" checked>
    612. 

  612.         
    613. 

  613.           
    614. 

  614.           <label class="md-nav__link" for="__nav_5" id="__nav_5_label" tabindex="0">
    615. 

  615.             
    616. 

  616.   
    617. 

  617.   
    618. 

  618.   <span class="md-ellipsis">
    619. 

  619.     ④ 基于Milvus库的问答系统
    620. 

  620.     
    621. 

  621.   </span>
    622. 

  622.   
    623. 

  623. 

  624.             <span class="md-nav__icon md-icon"></span>
    625. 

  625.           </label>
    626. 

  626.         
    627. 

  627.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_5_label" aria-expanded="true">
    628. 

  628.           <label class="md-nav__title" for="__nav_5">
    629. 

  629.             <span class="md-nav__icon md-icon"></span>
    630. 

  630.             ④ 基于Milvus库的问答系统
    631. 

  631.           </label>
    632. 

  632.           <ul class="md-nav__list" data-md-scrollfix>
    633. 

  633.             
    634. 

  634.               
    635. 

  635.                 
    636. 

  636.   
    637. 

  637.   
    638. 

  638.   
    639. 

  639.   
    640. 

  640.     <li class="md-nav__item">
    641. 

  641.       <a href="4.1%20%E6%95%B4%E4%BD%93%E6%9E%B6%E6%9E%84%E4%B8%8E%E5%B7%A5%E7%A8%8B%E6%B5%81%E7%A8%8B.html" class="md-nav__link">
    642. 

  642.         
    643. 

  643.   
    644. 

  644.   
    645. 

  645.   <span class="md-ellipsis">
    646. 

  646.     4.1 整体架构与工程流程
    647. 

  647.     
    648. 

  648.   </span>
    649. 

  649.   
    650. 

  650. 

  651.       </a>
    652. 

  652.     </li>
    653. 

  653.   
    654. 

  654. 

  655.               
    656. 

  656.             
    657. 

  657.               
    658. 

  658.                 
    659. 

  659.   
    660. 

  660.   
    661. 

  661.   
    662. 

  662.   
    663. 

  663.     <li class="md-nav__item">
    664. 

  664.       <a href="4.2%20%E5%9F%BA%E7%A1%80%E6%A8%A1%E5%9D%97.html" class="md-nav__link">
    665. 

  665.         
    666. 

  666.   
    667. 

  667.   
    668. 

  668.   <span class="md-ellipsis">
    669. 

  669.     4.2 基础模块
    670. 

  670.     
    671. 

  671.   </span>
    672. 

  672.   
    673. 

  673. 

  674.       </a>
    675. 

  675.     </li>
    676. 

  676.   
    677. 

  677. 

  678.               
    679. 

  679.             
    680. 

  680.               
    681. 

  681.                 
    682. 

  682.   
    683. 

  683.   
    684. 

  684.   
    685. 

  685.   
    686. 

  686.     <li class="md-nav__item">
    687. 

  687.       <a href="4.3%20%E6%96%87%E6%A1%A3%E5%A4%84%E7%90%86%E6%A8%A1%E5%9D%97.html" class="md-nav__link">
    688. 

  688.         
    689. 

  689.   
    690. 

  690.   
    691. 

  691.   <span class="md-ellipsis">
    692. 

  692.     4.3 文档处理模块
    693. 

  693.     
    694. 

  694.   </span>
    695. 

  695.   
    696. 

  696. 

  697.       </a>
    698. 

  698.     </li>
    699. 

  699.   
    700. 

  700. 

  701.               
    702. 

  702.             
    703. 

  703.               
    704. 

  704.                 
    705. 

  705.   
    706. 

  706.   
    707. 

  707.   
    708. 

  708.   
    709. 

  709.     <li class="md-nav__item">
    710. 

  710.       <a href="4.4%20%E5%90%91%E9%87%8F%E5%AD%98%E5%82%A8.html" class="md-nav__link">
    711. 

  711.         
    712. 

  712.   
    713. 

  713.   
    714. 

  714.   <span class="md-ellipsis">
    715. 

  715.     4.4 向量存储
    716. 

  716.     
    717. 

  717.   </span>
    718. 

  718.   
    719. 

  719. 

  720.       </a>
    721. 

  721.     </li>
    722. 

  722.   
    723. 

  723. 

  724.               
    725. 

  725.             
    726. 

  726.               
    727. 

  727.                 
    728. 

  728.   
    729. 

  729.   
    730. 

  730.     
    731. 

  731.   
    732. 

  732.   
    733. 

  733.   
    734. 

  734.     <li class="md-nav__item"><a href="4.5%20BERT%E5%9F%BA%E7%A1%80%E4%B8%8E%E4%BB%BB%E5%8A%A1%E5%BE%AE%E8%B0%83.html" class="md-nav__link"><span class="md-ellipsis">4.5 BERT基础与任务微调</span></a></li><li class="md-nav__item md-nav__item--active">
    735. 

  735.       
    736. 

  736.       <input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
    737. 

  737.       
    738. 

  738.       
    739. 

  739.       
    740. 

  740.         <label class="md-nav__link md-nav__link--active" for="__toc">
    741. 

  741.           
    742. 

  742.   
    743. 

  743.   
    744. 

  744.   <span class="md-ellipsis">
    745. 

  745.     4.6 prompts设计与query意图识别
    746. 

  746.     
    747. 

  747.   </span>
    748. 

  748.   
    749. 

  749. 

  750.           <span class="md-nav__icon md-icon"></span>
    751. 

  751.         </label>
    752. 

  752.       
    753. 

  753.       <a href="4.6%20prompts%E8%AE%BE%E8%AE%A1%E4%B8%8Equery%E6%84%8F%E5%9B%BE%E8%AF%86%E5%88%AB.html" class="md-nav__link md-nav__link--active">
    754. 

  754.         
    755. 

  755.   
    756. 

  756.   
    757. 

  757.   <span class="md-ellipsis">
    758. 

  758.     4.6 prompts设计与query意图识别
    759. 

  759.     
    760. 

  760.   </span>
    761. 

  761.   
    762. 

  762. 

  763.       </a>
    764. 

  764.       
    765. 

  765.         
    766. 

  766. 

  767. <nav class="md-nav md-nav--secondary" aria-label="目录">
    768. 

  768.   
    769. 

  769.   
    770. 

  770.   
    771. 

  771.   
    772. 

  772.     <label class="md-nav__title" for="__toc">
    773. 

  773.       <span class="md-nav__icon md-icon"></span>
    774. 

  774.       目录
    775. 

  775.     </label>
    776. 

  776.     <ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
    777. 

  777.       
    778. 

  778.         <li class="md-nav__item">
    779. 

  779.   <a href="#_1" class="md-nav__link">
    780. 

  780.     <span class="md-ellipsis">
    781. 

  781.       学习目标
    782. 

  782.     </span>
    783. 

  783.   </a>
    784. 

  784.   
    785. 

  785. </li>
    786. 

  786.       
    787. 

  787.         <li class="md-nav__item">
    788. 

  788.   <a href="#prompt" class="md-nav__link">
    789. 

  789.     <span class="md-ellipsis">
    790. 

  790.       一、Prompt管理
    791. 

  791.     </span>
    792. 

  792.   </a>
    793. 

  793.   
    794. 

  794.     <nav class="md-nav" aria-label="一、Prompt管理">
    795. 

  795.       <ul class="md-nav__list">
    796. 

  796.         
    797. 

  797.           <li class="md-nav__item">
    798. 

  798.   <a href="#1" class="md-nav__link">
    799. 

  799.     <span class="md-ellipsis">
    800. 

  800.       1 功能概述
    801. 

  801.     </span>
    802. 

  802.   </a>
    803. 

  803.   
    804. 

  804. </li>
    805. 

  805.         
    806. 

  806.           <li class="md-nav__item">
    807. 

  807.   <a href="#2" class="md-nav__link">
    808. 

  808.     <span class="md-ellipsis">
    809. 

  809.       2 代码实现
    810. 

  810.     </span>
    811. 

  811.   </a>
    812. 

  812.   
    813. 

  813. </li>
    814. 

  814.         
    815. 

  815.           <li class="md-nav__item">
    816. 

  816.   <a href="#3" class="md-nav__link">
    817. 

  817.     <span class="md-ellipsis">
    818. 

  818.       3 实现细节
    819. 

  819.     </span>
    820. 

  820.   </a>
    821. 

  821.   
    822. 

  822. </li>
    823. 

  823.         
    824. 

  824.       </ul>
    825. 

  825.     </nav>
    826. 

  826.   
    827. 

  827. </li>
    828. 

  828.       
    829. 

  829.         <li class="md-nav__item">
    830. 

  830.   <a href="#_2" class="md-nav__link">
    831. 

  831.     <span class="md-ellipsis">
    832. 

  832.       二、查询分类
    833. 

  833.     </span>
    834. 

  834.   </a>
    835. 

  835.   
    836. 

  836.     <nav class="md-nav" aria-label="二、查询分类">
    837. 

  837.       <ul class="md-nav__list">
    838. 

  838.         
    839. 

  839.           <li class="md-nav__item">
    840. 

  840.   <a href="#1_1" class="md-nav__link">
    841. 

  841.     <span class="md-ellipsis">
    842. 

  842.       1 功能概述
    843. 

  843.     </span>
    844. 

  844.   </a>
    845. 

  845.   
    846. 

  846. </li>
    847. 

  847.         
    848. 

  848.           <li class="md-nav__item">
    849. 

  849.   <a href="#2_1" class="md-nav__link">
    850. 

  850.     <span class="md-ellipsis">
    851. 

  851.       2 代码实现
    852. 

  852.     </span>
    853. 

  853.   </a>
    854. 

  854.   
    855. 

  855. </li>
    856. 

  856.         
    857. 

  857.           <li class="md-nav__item">
    858. 

  858.   <a href="#3_1" class="md-nav__link">
    859. 

  859.     <span class="md-ellipsis">
    860. 

  860.       3 实现细节
    861. 

  861.     </span>
    862. 

  862.   </a>
    863. 

  863.   
    864. 

  864. </li>
    865. 

  865.         
    866. 

  866.           <li class="md-nav__item">
    867. 

  867.   <a href="#4" class="md-nav__link">
    868. 

  868.     <span class="md-ellipsis">
    869. 

  869.       4 执行示例
    870. 

  870.     </span>
    871. 

  871.   </a>
    872. 

  872.   
    873. 

  873.     <nav class="md-nav" aria-label="4 执行示例">
    874. 

  874.       <ul class="md-nav__list">
    875. 

  875.         
    876. 

  876.           <li class="md-nav__item">
    877. 

  877.   <a href="#41" class="md-nav__link">
    878. 

  878.     <span class="md-ellipsis">
    879. 

  879.       4.1 运行脚本
    880. 

  880.     </span>
    881. 

  881.   </a>
    882. 

  882.   
    883. 

  883. </li>
    884. 

  884.         
    885. 

  885.           <li class="md-nav__item">
    886. 

  886.   <a href="#42" class="md-nav__link">
    887. 

  887.     <span class="md-ellipsis">
    888. 

  888.       4.2 代码示例
    889. 

  889.     </span>
    890. 

  890.   </a>
    891. 

  891.   
    892. 

  892. </li>
    893. 

  893.         
    894. 

  894.       </ul>
    895. 

  895.     </nav>
    896. 

  896.   
    897. 

  897. </li>
    898. 

  898.         
    899. 

  899.       </ul>
    900. 

  900.     </nav>
    901. 

  901.   
    902. 

  902. </li>
    903. 

  903.       
    904. 

  904.         <li class="md-nav__item">
    905. 

  905.   <a href="#_3" class="md-nav__link">
    906. 

  906.     <span class="md-ellipsis">
    907. 

  907.       章节小结
    908. 

  908.     </span>
    909. 

  909.   </a>
    910. 

  910.   
    911. 

  911. </li>
    912. 

  912.       
    913. 

  913.     </ul>
    914. 

  914.   
    915. 

  915. </nav>
    916. 

  916.       
    917. 

  917.     </li>
    918. 

  918.   
    919. 

  919. 

  920.               
    921. 

  921.             
    922. 

  922.               
    923. 

  923.                 
    924. 

  924.   
    925. 

  925.   
    926. 

  926.   
    927. 

  927.   
    928. 

  928.     <li class="md-nav__item">
    929. 

  929.       <a href="4.7%20%E6%A3%80%E7%B4%A2%E7%AD%96%E7%95%A5%E4%B8%8ERAG%E7%B3%BB%E7%BB%9F%E8%AE%BE%E8%AE%A1.html" class="md-nav__link">
    930. 

  930.         
    931. 

  931.   
    932. 

  932.   
    933. 

  933.   <span class="md-ellipsis">
    934. 

  934.     4.7 检索策略与RAG系统设计
    935. 

  935.     
    936. 

  936.   </span>
    937. 

  937.   
    938. 

  938. 

  939.       </a>
    940. 

  940.     </li>
    941. 

  941.   
    942. 

  942. 

  943.               
    944. 

  944.             
    945. 

  945.               
    946. 

  946.                 
    947. 

  947.   
    948. 

  948.   
    949. 

  949.   
    950. 

  950.   
    951. 

  951.     <li class="md-nav__item">
    952. 

  952.       <a href="4.8%20RAG%E7%B3%BB%E7%BB%9F%E8%BF%90%E8%A1%8C.html" class="md-nav__link">
    953. 

  953.         
    954. 

  954.   
    955. 

  955.   
    956. 

  956.   <span class="md-ellipsis">
    957. 

  957.     4.8 RAG系统运行
    958. 

  958.     
    959. 

  959.   </span>
    960. 

  960.   
    961. 

  961. 

  962.       </a>
    963. 

  963.     </li>
    964. 

  964.   
    965. 

  965. 

  966.               
    967. 

  967.             
    968. 

  968.               
    969. 

  969.                 
    970. 

  970.   
    971. 

  971.   
    972. 

  972.   
    973. 

  973.   
    974. 

  974.     <li class="md-nav__item">
    975. 

  975.       <a href="4.9%20%E6%96%87%E6%A1%A3%E8%A7%A3%E6%9E%90%E5%B7%A5%E5%85%B7%28%E6%89%A9%E5%B1%95%E8%B5%84%E6%96%99%29.html" class="md-nav__link">
    976. 

  976.         
    977. 

  977.   
    978. 

  978.   
    979. 

  979.   <span class="md-ellipsis">
    980. 

  980.     4.9 文档解析工具(扩展资料)
    981. 

  981.     
    982. 

  982.   </span>
    983. 

  983.   
    984. 

  984. 

  985.       </a>
    986. 

  986.     </li>
    987. 

  987.   
    988. 

  988. 

  989.               
    990. 

  990.             
    991. 

  991.               
    992. 

  992.                 
    993. 

  993.   
    994. 

  994.   
    995. 

  995.   
    996. 

  996.   
    997. 

  997.     <li class="md-nav__item">
    998. 

  998.       <a href="4.10%20RAG%E4%B8%AD%E7%9A%84Query%E6%94%B9%E5%86%99%28%E6%89%A9%E5%B1%95%E8%B5%84%E6%96%99%29.html" class="md-nav__link">
    999. 

  999.         
    1000. 

  1000.   
    1001. 

  1001.   
    1002. 

  1002.   <span class="md-ellipsis">
    1003. 

  1003.     4.10 RAG中的Query改写(扩展资料)
    1004. 

  1004.     
    1005. 

  1005.   </span>
    1006. 

  1006.   
    1007. 

  1007. 

  1008.       </a>
    1009. 

  1009.     </li>
    1010. 

  1010.   
    1011. 

  1011. 

  1012.               
    1013. 

  1013.             
    1014. 

  1014.               
    1015. 

  1015.                 
    1016. 

  1016.   
    1017. 

  1017.   
    1018. 

  1018.   
    1019. 

  1019.   
    1020. 

  1020.     <li class="md-nav__item">
    1021. 

  1021.       <a href="%E6%95%B0%E6%8D%AE%E9%9B%86%E7%94%9F%E6%88%90%E4%B8%8E%E4%BC%98%E5%8C%96%28%E6%89%A9%E5%B1%95%E8%B5%84%E6%96%99%29.html" class="md-nav__link">
    1022. 

  1022.         
    1023. 

  1023.   
    1024. 

  1024.   
    1025. 

  1025.   <span class="md-ellipsis">
    1026. 

  1026.     数据集生成与优化(扩展资料)
    1027. 

  1027.     
    1028. 

  1028.   </span>
    1029. 

  1029.   
    1030. 

  1030. 

  1031.       </a>
    1032. 

  1032.     </li>
    1033. 

  1033.   
    1034. 

  1034. 

  1035.               
    1036. 

  1036.             
    1037. 

  1037.           </ul>
    1038. 

  1038.         </nav>
    1039. 

  1039.       
    1040. 

  1040.     </li>
    1041. 

  1041.   
    1042. 

  1042. 

  1043.     
    1044. 

  1044.       
    1045. 

  1045.       
    1046. 

  1046.   
    1047. 

  1047.   
    1048. 

  1048.   
    1049. 

  1049.   
    1050. 

  1050.     
    1051. 

  1051.     
    1052. 

  1052.     
    1053. 

  1053.     
    1054. 

  1054.     
    1055. 

  1055.     <li class="md-nav__item md-nav__item--nested">
    1056. 

  1056.       
    1057. 

  1057.         
    1058. 

  1058.         
    1059. 

  1059.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_6">
    1060. 

  1060.         
    1061. 

  1061.           
    1062. 

  1062.           <label class="md-nav__link" for="__nav_6" id="__nav_6_label" tabindex="0">
    1063. 

  1063.             
    1064. 

  1064.   
    1065. 

  1065.   
    1066. 

  1066.   <span class="md-ellipsis">
    1067. 

  1067.     ⑤ RAG系统评估
    1068. 

  1068.     
    1069. 

  1069.   </span>
    1070. 

  1070.   
    1071. 

  1071. 

  1072.             <span class="md-nav__icon md-icon"></span>
    1073. 

  1073.           </label>
    1074. 

  1074.         
    1075. 

  1075.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_6_label" aria-expanded="false">
    1076. 

  1076.           <label class="md-nav__title" for="__nav_6">
    1077. 

  1077.             <span class="md-nav__icon md-icon"></span>
    1078. 

  1078.             ⑤ RAG系统评估
    1079. 

  1079.           </label>
    1080. 

  1080.           <ul class="md-nav__list" data-md-scrollfix>
    1081. 

  1081.             
    1082. 

  1082.               
    1083. 

  1083.                 
    1084. 

  1084.   
    1085. 

  1085.   
    1086. 

  1086.   
    1087. 

  1087.   
    1088. 

  1088.     <li class="md-nav__item">
    1089. 

  1089.       <a href="../%E2%91%A4%20RAG%E7%B3%BB%E7%BB%9F%E8%AF%84%E4%BC%B0/5.1%20RAG%E7%B3%BB%E7%BB%9F%E8%AF%84%E4%BC%B0%E5%B7%A5%E5%85%B7.html" class="md-nav__link">
    1090. 

  1090.         
    1091. 

  1091.   
    1092. 

  1092.   
    1093. 

  1093.   <span class="md-ellipsis">
    1094. 

  1094.     5.1 RAG系统评估工具
    1095. 

  1095.     
    1096. 

  1096.   </span>
    1097. 

  1097.   
    1098. 

  1098. 

  1099.       </a>
    1100. 

  1100.     </li>
    1101. 

  1101.   
    1102. 

  1102. 

  1103.               
    1104. 

  1104.             
    1105. 

  1105.               
    1106. 

  1106.                 
    1107. 

  1107.   
    1108. 

  1108.   
    1109. 

  1109.   
    1110. 

  1110.   
    1111. 

  1111.     <li class="md-nav__item">
    1112. 

  1112.       <a href="../%E2%91%A4%20RAG%E7%B3%BB%E7%BB%9F%E8%AF%84%E4%BC%B0/5.2%20%E5%AE%9E%E7%8E%B0EnterpriseKB-RAG%E7%B3%BB%E7%BB%9F%E7%9A%84%E8%AF%84%E4%BC%B0.html" class="md-nav__link">
    1113. 

  1113.         
    1114. 

  1114.   
    1115. 

  1115.   
    1116. 

  1116.   <span class="md-ellipsis">
    1117. 

  1117.     5.2 实现企业知识库RAG系统的评估
    1118. 

  1118.     
    1119. 

  1119.   </span>
    1120. 

  1120.   
    1121. 

  1121. 

  1122.       </a>
    1123. 

  1123.     </li>
    1124. 

  1124.   
    1125. 

  1125. 

  1126.               
    1127. 

  1127.             
    1128. 

  1128.               
    1129. 

  1129.                 
    1130. 

  1130.   
    1131. 

  1131.   
    1132. 

  1132.   
    1133. 

  1133.   
    1134. 

  1134.     <li class="md-nav__item">
    1135. 

  1135.       <a href="../%E2%91%A4%20RAG%E7%B3%BB%E7%BB%9F%E8%AF%84%E4%BC%B0/5.3%20RAG%E8%AF%84%E4%BC%B0%E6%95%B0%E6%8D%AE%E9%9B%86%E7%9A%84%E6%9E%84%E9%80%A0%28%E6%89%A9%E5%B1%95%29.html" class="md-nav__link">
    1136. 

  1136.         
    1137. 

  1137.   
    1138. 

  1138.   
    1139. 

  1139.   <span class="md-ellipsis">
    1140. 

  1140.     5.3 RAG评估数据集的构造(扩展)
    1141. 

  1141.     
    1142. 

  1142.   </span>
    1143. 

  1143.   
    1144. 

  1144. 

  1145.       </a>
    1146. 

  1146.     </li>
    1147. 

  1147.   
    1148. 

  1148. 

  1149.               
    1150. 

  1150.             
    1151. 

  1151.           </ul>
    1152. 

  1152.         </nav>
    1153. 

  1153.       
    1154. 

  1154.     </li>
    1155. 

  1155.   
    1156. 

  1156. 

  1157.     
    1158. 

  1158.       
    1159. 

  1159.       
    1160. 

  1160.   
    1161. 

  1161.   
    1162. 

  1162.   
    1163. 

  1163.   
    1164. 

  1164.     
    1165. 

  1165.     
    1166. 

  1166.     
    1167. 

  1167.     
    1168. 

  1168.     
    1169. 

  1169.     <li class="md-nav__item md-nav__item--nested">
    1170. 

  1170.       
    1171. 

  1171.         
    1172. 

  1172.         
    1173. 

  1173.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_7">
    1174. 

  1174.         
    1175. 

  1175.           
    1176. 

  1176.           <label class="md-nav__link" for="__nav_7" id="__nav_7_label" tabindex="0">
    1177. 

  1177.             
    1178. 

  1178.   
    1179. 

  1179.   
    1180. 

  1180.   <span class="md-ellipsis">
    1181. 

  1181.     ⑥ 融合Mysql的RAG系统
    1182. 

  1182.     
    1183. 

  1183.   </span>
    1184. 

  1184.   
    1185. 

  1185. 

  1186.             <span class="md-nav__icon md-icon"></span>
    1187. 

  1187.           </label>
    1188. 

  1188.         
    1189. 

  1189.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_7_label" aria-expanded="false">
    1190. 

  1190.           <label class="md-nav__title" for="__nav_7">
    1191. 

  1191.             <span class="md-nav__icon md-icon"></span>
    1192. 

  1192.             ⑥ 融合Mysql的RAG系统
    1193. 

  1193.           </label>
    1194. 

  1194.           <ul class="md-nav__list" data-md-scrollfix>
    1195. 

  1195.             
    1196. 

  1196.               
    1197. 

  1197.                 
    1198. 

  1198.   
    1199. 

  1199.   
    1200. 

  1200.   
    1201. 

  1201.   
    1202. 

  1202.     <li class="md-nav__item">
    1203. 

  1203.       <a href="../%E2%91%A5%20%E8%9E%8D%E5%90%88Mysql%E7%9A%84RAG%E7%B3%BB%E7%BB%9F/6.1%20%E8%9E%8D%E5%90%88FQA%E5%92%8C%E7%9F%A5%E8%AF%86%E5%BA%93%E6%9F%A5%E8%AF%A2.html" class="md-nav__link">
    1204. 

  1204.         
    1205. 

  1205.   
    1206. 

  1206.   
    1207. 

  1207.   <span class="md-ellipsis">
    1208. 

  1208.     6.1 融合FQA和知识库查询
    1209. 

  1209.     
    1210. 

  1210.   </span>
    1211. 

  1211.   
    1212. 

  1212. 

  1213.       </a>
    1214. 

  1214.     </li>
    1215. 

  1215.   
    1216. 

  1216. 

  1217.               
    1218. 

  1218.             
    1219. 

  1219.               
    1220. 

  1220.                 
    1221. 

  1221.   
    1222. 

  1222.   
    1223. 

  1223.   
    1224. 

  1224.   
    1225. 

  1225.     <li class="md-nav__item">
    1226. 

  1226.       <a href="../%E2%91%A5%20%E8%9E%8D%E5%90%88Mysql%E7%9A%84RAG%E7%B3%BB%E7%BB%9F/6.2%20%E8%9E%8D%E5%90%88FAQ%20%E5%92%8C%E7%9F%A5%E8%AF%86%E5%BA%93%E6%9F%A5%E8%AF%A2%EF%BC%88%E4%BC%98%E5%8C%96%E7%89%88%EF%BC%89.html" class="md-nav__link">
    1227. 

  1227.         
    1228. 

  1228.   
    1229. 

  1229.   
    1230. 

  1230.   <span class="md-ellipsis">
    1231. 

  1231.     6.2 融合FAQ 和知识库查询（优化版）
    1232. 

  1232.     
    1233. 

  1233.   </span>
    1234. 

  1234.   
    1235. 

  1235. 

  1236.       </a>
    1237. 

  1237.     </li>
    1238. 

  1238.   
    1239. 

  1239. 

  1240.               
    1241. 

  1241.             
    1242. 

  1242.               
    1243. 

  1243.                 
    1244. 

  1244.   
    1245. 

  1245.   
    1246. 

  1246.   
    1247. 

  1247.   
    1248. 

  1248.     <li class="md-nav__item">
    1249. 

  1249.       <a href="../%E2%91%A5%20%E8%9E%8D%E5%90%88Mysql%E7%9A%84RAG%E7%B3%BB%E7%BB%9F/6.3%20%E5%9F%BA%E4%BA%8E%20FastAPI%20%E7%9A%84%20FAQ%20%E4%B8%8E%E7%9F%A5%E8%AF%86%E5%BA%93%E6%9F%A5%E8%AF%A2%E6%8E%A5%E5%8F%A3.html" class="md-nav__link">
    1250. 

  1250.         
    1251. 

  1251.   
    1252. 

  1252.   
    1253. 

  1253.   <span class="md-ellipsis">
    1254. 

  1254.     6.3 基于 FastAPI 的 FAQ 与知识库查询接口
    1255. 

  1255.     
    1256. 

  1256.   </span>
    1257. 

  1257.   
    1258. 

  1258. 

  1259.       </a>
    1260. 

  1260.     </li>
    1261. 

  1261.   
    1262. 

  1262. 

  1263.               
    1264. 

  1264.             
    1265. 

  1265.               
    1266. 

  1266.                 
    1267. 

  1267.   
    1268. 

  1268.   
    1269. 

  1269.   
    1270. 

  1270.   
    1271. 

  1271.     <li class="md-nav__item">
    1272. 

  1272.       <a href="../%E2%91%A5%20%E8%9E%8D%E5%90%88Mysql%E7%9A%84RAG%E7%B3%BB%E7%BB%9F/6.4%20FastAPI%E5%92%8CFlask%E7%9A%84%E5%8C%BA%E5%88%AB%E5%92%8C%E8%81%94%E7%B3%BB%28%E6%89%A9%E5%B1%95%E8%B5%84%E6%96%99%29.html" class="md-nav__link">
    1273. 

  1273.         
    1274. 

  1274.   
    1275. 

  1275.   
    1276. 

  1276.   <span class="md-ellipsis">
    1277. 

  1277.     6.4 FastAPI和Flask的区别和联系(扩展资料)
    1278. 

  1278.     
    1279. 

  1279.   </span>
    1280. 

  1280.   
    1281. 

  1281. 

  1282.       </a>
    1283. 

  1283.     </li>
    1284. 

  1284.   
    1285. 

  1285. 

  1286.               
    1287. 

  1287.             
    1288. 

  1288.           </ul>
    1289. 

  1289.         </nav>
    1290. 

  1290.       
    1291. 

  1291.     </li>
    1292. 

  1292.   
    1293. 

  1293. 

  1294.     
    1295. 

  1295.       
    1296. 

  1296.       
    1297. 

  1297.   
    1298. 

  1298.   
    1299. 

  1299.   
    1300. 

  1300.   
    1301. 

  1301.     
    1302. 

  1302.     
    1303. 

  1303.     
    1304. 

  1304.     
    1305. 

  1305.     
    1306. 

  1306.     <li class="md-nav__item md-nav__item--nested">
    1307. 

  1307.       
    1308. 

  1308.         
    1309. 

  1309.         
    1310. 

  1310.         <input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_8">
    1311. 

  1311.         
    1312. 

  1312.           
    1313. 

  1313.           <label class="md-nav__link" for="__nav_8" id="__nav_8_label" tabindex="0">
    1314. 

  1314.             
    1315. 

  1315.   
    1316. 

  1316.   
    1317. 

  1317.   <span class="md-ellipsis">
    1318. 

  1318.     ⑦ 企业级生产环境部署
    1319. 

  1319.     
    1320. 

  1320.   </span>
    1321. 

  1321.   
    1322. 

  1322. 

  1323.             <span class="md-nav__icon md-icon"></span>
    1324. 

  1324.           </label>
    1325. 

  1325.         
    1326. 

  1326.         <nav class="md-nav" data-md-level="1" aria-labelledby="__nav_8_label" aria-expanded="false">
    1327. 

  1327.           <label class="md-nav__title" for="__nav_8">
    1328. 

  1328.             <span class="md-nav__icon md-icon"></span>
    1329. 

  1329.             ⑦ 企业级生产环境部署
    1330. 

  1330.           </label>
    1331. 

  1331.           <ul class="md-nav__list" data-md-scrollfix>
    1332. 

  1332.             
    1333. 

  1333.               
    1334. 

  1334.                 
    1335. 

  1335.   
    1336. 

  1336.   
    1337. 

  1337.   
    1338. 

  1338.   
    1339. 

  1339.     <li class="md-nav__item">
    1340. 

  1340.       <a href="../%E2%91%A6%20%E4%BC%81%E4%B8%9A%E7%BA%A7%E7%94%9F%E4%BA%A7%E7%8E%AF%E5%A2%83%E9%83%A8%E7%BD%B2/7.1%20Docker%E7%9A%84%E5%8E%9F%E7%90%86%E5%92%8C%E5%9F%BA%E6%9C%AC%E4%BD%BF%E7%94%A8.html" class="md-nav__link">
    1341. 

  1341.         
    1342. 

  1342.   
    1343. 

  1343.   
    1344. 

  1344.   <span class="md-ellipsis">
    1345. 

  1345.     7.1 Docker的原理和基本使用
    1346. 

  1346.     
    1347. 

  1347.   </span>
    1348. 

  1348.   
    1349. 

  1349. 

  1350.       </a>
    1351. 

  1351.     </li>
    1352. 

  1352.   
    1353. 

  1353. 

  1354.               
    1355. 

  1355.             
    1356. 

  1356.               
    1357. 

  1357.                 
    1358. 

  1358.   
    1359. 

  1359.   
    1360. 

  1360.   
    1361. 

  1361.   
    1362. 

  1362.     <li class="md-nav__item">
    1363. 

  1363.       <a href="../%E2%91%A6%20%E4%BC%81%E4%B8%9A%E7%BA%A7%E7%94%9F%E4%BA%A7%E7%8E%AF%E5%A2%83%E9%83%A8%E7%BD%B2/7.2%20%E6%9C%8D%E5%8A%A1%E9%83%A8%E7%BD%B2.html" class="md-nav__link">
    1364. 

  1364.         
    1365. 

  1365.   
    1366. 

  1366.   
    1367. 

  1367.   <span class="md-ellipsis">
    1368. 

  1368.     7.2 服务部署
    1369. 

  1369.     
    1370. 

  1370.   </span>
    1371. 

  1371.   
    1372. 

  1372. 

  1373.       </a>
    1374. 

  1374.     </li>
    1375. 

  1375.   
    1376. 

  1376. 

  1377.               
    1378. 

  1378.             
    1379. 

  1379.           </ul>
    1380. 

  1380.         </nav>
    1381. 

  1381.       
    1382. 

  1382.     </li>
    1383. 

  1383.   
    1384. 

  1384. 

  1385.     
    1386. 

  1386.   </ul>
    1387. 

  1387. </nav>
    1388. 

  1388.                   </div>
    1389. 

  1389.                 </div>
    1390. 

  1390.               </div>
    1391. 

  1391.             
    1392. 

  1392.             
    1393. 

  1393.               
    1394. 

  1394.               <div class="md-sidebar md-sidebar--secondary" data-md-component="sidebar" data-md-type="toc">
    1395. 

  1395.                 <div class="md-sidebar__scrollwrap">
    1396. 

  1396.                   <div class="md-sidebar__inner">
    1397. 

  1397.                     
    1398. 

  1398. 

  1399. <nav class="md-nav md-nav--secondary" aria-label="目录">
    1400. 

  1400.   
    1401. 

  1401.   
    1402. 

  1402.   
    1403. 

  1403.   
    1404. 

  1404.     <label class="md-nav__title" for="__toc">
    1405. 

  1405.       <span class="md-nav__icon md-icon"></span>
    1406. 

  1406.       目录
    1407. 

  1407.     </label>
    1408. 

  1408.     <ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
    1409. 

  1409.       
    1410. 

  1410.         <li class="md-nav__item">
    1411. 

  1411.   <a href="#_1" class="md-nav__link">
    1412. 

  1412.     <span class="md-ellipsis">
    1413. 

  1413.       学习目标
    1414. 

  1414.     </span>
    1415. 

  1415.   </a>
    1416. 

  1416.   
    1417. 

  1417. </li>
    1418. 

  1418.       
    1419. 

  1419.         <li class="md-nav__item">
    1420. 

  1420.   <a href="#prompt" class="md-nav__link">
    1421. 

  1421.     <span class="md-ellipsis">
    1422. 

  1422.       一、Prompt管理
    1423. 

  1423.     </span>
    1424. 

  1424.   </a>
    1425. 

  1425.   
    1426. 

  1426.     <nav class="md-nav" aria-label="一、Prompt管理">
    1427. 

  1427.       <ul class="md-nav__list">
    1428. 

  1428.         
    1429. 

  1429.           <li class="md-nav__item">
    1430. 

  1430.   <a href="#1" class="md-nav__link">
    1431. 

  1431.     <span class="md-ellipsis">
    1432. 

  1432.       1 功能概述
    1433. 

  1433.     </span>
    1434. 

  1434.   </a>
    1435. 

  1435.   
    1436. 

  1436. </li>
    1437. 

  1437.         
    1438. 

  1438.           <li class="md-nav__item">
    1439. 

  1439.   <a href="#2" class="md-nav__link">
    1440. 

  1440.     <span class="md-ellipsis">
    1441. 

  1441.       2 代码实现
    1442. 

  1442.     </span>
    1443. 

  1443.   </a>
    1444. 

  1444.   
    1445. 

  1445. </li>
    1446. 

  1446.         
    1447. 

  1447.           <li class="md-nav__item">
    1448. 

  1448.   <a href="#3" class="md-nav__link">
    1449. 

  1449.     <span class="md-ellipsis">
    1450. 

  1450.       3 实现细节
    1451. 

  1451.     </span>
    1452. 

  1452.   </a>
    1453. 

  1453.   
    1454. 

  1454. </li>
    1455. 

  1455.         
    1456. 

  1456.       </ul>
    1457. 

  1457.     </nav>
    1458. 

  1458.   
    1459. 

  1459. </li>
    1460. 

  1460.       
    1461. 

  1461.         <li class="md-nav__item">
    1462. 

  1462.   <a href="#_2" class="md-nav__link">
    1463. 

  1463.     <span class="md-ellipsis">
    1464. 

  1464.       二、查询分类
    1465. 

  1465.     </span>
    1466. 

  1466.   </a>
    1467. 

  1467.   
    1468. 

  1468.     <nav class="md-nav" aria-label="二、查询分类">
    1469. 

  1469.       <ul class="md-nav__list">
    1470. 

  1470.         
    1471. 

  1471.           <li class="md-nav__item">
    1472. 

  1472.   <a href="#1_1" class="md-nav__link">
    1473. 

  1473.     <span class="md-ellipsis">
    1474. 

  1474.       1 功能概述
    1475. 

  1475.     </span>
    1476. 

  1476.   </a>
    1477. 

  1477.   
    1478. 

  1478. </li>
    1479. 

  1479.         
    1480. 

  1480.           <li class="md-nav__item">
    1481. 

  1481.   <a href="#2_1" class="md-nav__link">
    1482. 

  1482.     <span class="md-ellipsis">
    1483. 

  1483.       2 代码实现
    1484. 

  1484.     </span>
    1485. 

  1485.   </a>
    1486. 

  1486.   
    1487. 

  1487. </li>
    1488. 

  1488.         
    1489. 

  1489.           <li class="md-nav__item">
    1490. 

  1490.   <a href="#3_1" class="md-nav__link">
    1491. 

  1491.     <span class="md-ellipsis">
    1492. 

  1492.       3 实现细节
    1493. 

  1493.     </span>
    1494. 

  1494.   </a>
    1495. 

  1495.   
    1496. 

  1496. </li>
    1497. 

  1497.         
    1498. 

  1498.           <li class="md-nav__item">
    1499. 

  1499.   <a href="#4" class="md-nav__link">
    1500. 

  1500.     <span class="md-ellipsis">
    1501. 

  1501.       4 执行示例
    1502. 

  1502.     </span>
    1503. 

  1503.   </a>
    1504. 

  1504.   
    1505. 

  1505.     <nav class="md-nav" aria-label="4 执行示例">
    1506. 

  1506.       <ul class="md-nav__list">
    1507. 

  1507.         
    1508. 

  1508.           <li class="md-nav__item">
    1509. 

  1509.   <a href="#41" class="md-nav__link">
    1510. 

  1510.     <span class="md-ellipsis">
    1511. 

  1511.       4.1 运行脚本
    1512. 

  1512.     </span>
    1513. 

  1513.   </a>
    1514. 

  1514.   
    1515. 

  1515. </li>
    1516. 

  1516.         
    1517. 

  1517.           <li class="md-nav__item">
    1518. 

  1518.   <a href="#42" class="md-nav__link">
    1519. 

  1519.     <span class="md-ellipsis">
    1520. 

  1520.       4.2 代码示例
    1521. 

  1521.     </span>
    1522. 

  1522.   </a>
    1523. 

  1523.   
    1524. 

  1524. </li>
    1525. 

  1525.         
    1526. 

  1526.       </ul>
    1527. 

  1527.     </nav>
    1528. 

  1528.   
    1529. 

  1529. </li>
    1530. 

  1530.         
    1531. 

  1531.       </ul>
    1532. 

  1532.     </nav>
    1533. 

  1533.   
    1534. 

  1534. </li>
    1535. 

  1535.       
    1536. 

  1536.         <li class="md-nav__item">
    1537. 

  1537.   <a href="#_3" class="md-nav__link">
    1538. 

  1538.     <span class="md-ellipsis">
    1539. 

  1539.       章节小结
    1540. 

  1540.     </span>
    1541. 

  1541.   </a>
    1542. 

  1542.   
    1543. 

  1543. </li>
    1544. 

  1544.       
    1545. 

  1545.     </ul>
    1546. 

  1546.   
    1547. 

  1547. </nav>
    1548. 

  1548.                   </div>
    1549. 

  1549.                 </div>
    1550. 

  1550.               </div>
    1551. 

  1551.             
    1552. 

  1552.           
    1553. 

  1553.           
    1554. 

  1554.             <div class="md-content" data-md-component="content">
    1555. 

  1555.               <article class="md-content__inner md-typeset">
    1556. 

  1556.                 
    1557. 

  1557.                   
    1558. 

  1558. 

  1559. 

  1560. 

  1561.   <h1>4.6 prompts设计与query意图识别</h1>
    1562. 

  1562. 

  1563. <div class="admonition note" data-bert-bridge="true"><p class="admonition-title">学习衔接</p><p>本节进入 Prompt 管理与查询分类的实现。前一节 4.5 已讲清分词器、标签、损失、微调与推理；阅读查询分类代码时，可对照其中“单句分类”的项目二分类示例。 <a href="4.5%20BERT%E5%9F%BA%E7%A1%80%E4%B8%8E%E4%BB%BB%E5%8A%A1%E5%BE%AE%E8%B0%83.html#intent">阅读 4.5 对应内容</a></p></div><h2 id="_1">学习目标<a class="headerlink" href="#_1" title="Permanent link">¶</a></h2>
    1564. 

  1564. <ul>
    1565. 

  1565. <li>1.掌握如何设计和使用Prompt模板来引导大语言模型生成高质量输出。</li>
    1566. 

  1566. <li>2.学会查询分类的基本原理，了解如何通过分类优化输入处理流程。</li>
    1567. 

  1567. </ul>
    1568. 

  1568. <hr>
    1569. 

  1569. <p><code>prompts.py</code>和<code>query_classifier.py</code>是企业知识库RAG系统中<code>core</code>模块的重要组成部分，分别负责Prompt模板管理和查询分类。这两个模块通过优化用户输入的处理，增强了系统的灵活性和智能性，为RAG系统的检索和生成阶段奠定了基础。<code>prompts.py</code>定义了多种Prompt模板，用于引导大语言模型生成特定输出，而<code>query_classifier.py</code>通过分类用户查询，决定是否直接使用模型回答或触发检索流程。</p>
    1570. 

  1570. <h2 id="prompt">一、Prompt管理<a class="headerlink" href="#prompt" title="Permanent link">¶</a></h2>
    1571. 

  1571. <h3 id="1">1 功能概述<a class="headerlink" href="#1" title="Permanent link">¶</a></h3>
    1572. 

  1572. <p><code>prompts.py</code>定义了<code>RAGPrompts</code>类，负责管理系统中使用的所有Prompt模板。这些模板用于指导大语言模型完成不同任务，例如生成最终答案、假设答案、子查询或简化问题。通过集中管理Prompt，系统能够确保输入的一致性和输出质量。</p>
    1573. 

  1573. <h3 id="2">2 代码实现<a class="headerlink" href="#2" title="Permanent link">¶</a></h3>
    1574. 

  1574. <div class="highlight"><pre><span></span><code><span class="c1"># core/prompts.py</span>
    1575. 

  1575. <span class="c1"># 导入 PromptTemplate 类，用于创建 Prompt 模板</span>
    1576. 

  1576. <span class="kn">from</span><span class="w"> </span><span class="nn">langchain_core.prompts</span><span class="w"> </span><span class="kn">import</span> <span class="n">PromptTemplate</span>
    1577. 

  1577. 

  1578. 

  1579. <span class="c1"># 定义 RAGPrompts 类，用于管理所有 Prompt 模板</span>
    1580. 

  1580. <span class="k">class</span><span class="w"> </span><span class="nc">RAGPrompts</span><span class="p">:</span>
    1581. 

  1581.     <span class="c1"># 定义 RAG 提示模板</span>
    1582. 

  1582.     <span class="nd">@staticmethod</span>
    1583. 

  1583.     <span class="k">def</span><span class="w"> </span><span class="nf">rag_prompt</span><span class="p">():</span>
    1584. 

  1584.         <span class="c1"># 创建并返回 PromptTemplate 对象</span>
    1585. 

  1585.         <span class="k">return</span> <span class="n">PromptTemplate</span><span class="p">(</span>
    1586. 

  1586.             <span class="n">template</span><span class="o">=</span><span class="s2">"""  </span>
    1587. 

  1587. <span class="s2">            你是一个智能助手，帮助用户回答问题。  </span>
    1588. 

  1588. <span class="s2">            如果提供了上下文，请基于上下文回答；如果没有上下文，请直接根据你的知识回答。  </span>
    1589. 

  1589. <span class="s2">            如果答案来源于检索到的文档，请在回答中说明。</span>
    1590. 

  1590. 

  1591. <span class="s2">            上下文: </span><span class="si">{context}</span><span class="s2">  </span>
    1592. 

  1592. <span class="s2">            问题: </span><span class="si">{question}</span><span class="s2">  </span>
    1593. 

  1593. 

  1594. <span class="s2">            如果无法回答，请回复：“信息不足，无法回答，请联系人工客服，电话：</span><span class="si">{phone}</span><span class="s2">。”  </span>
    1595. 

  1595. <span class="s2">            回答:  </span>
    1596. 

  1596. <span class="s2">            """</span><span class="p">,</span>
    1597. 

  1597.             <span class="c1">#   定义输入变量</span>
    1598. 

  1598.             <span class="n">input_variables</span><span class="o">=</span><span class="p">[</span><span class="s2">"context"</span><span class="p">,</span> <span class="s2">"question"</span><span class="p">,</span> <span class="s2">"phone"</span><span class="p">],</span>
    1599. 

  1599.         <span class="p">)</span>
    1600. 

  1600.         <span class="c1"># @staticmethod</span>
    1601. 

  1601.     <span class="c1"># def rag_prompt():</span>
    1602. 

  1602.     <span class="c1">#     return PromptTemplate(</span>
    1603. 

  1603.     <span class="c1">#         template="""</span>
    1604. 

  1604.     <span class="c1">#     你是一个智能助手，负责帮助用户回答问题。请按照以下步骤处理：</span>
    1605. 

  1605.     <span class="c1"># </span>
    1606. 

  1606.     <span class="c1">#     1. **分析问题和上下文**：</span>
    1607. 

  1607.     <span class="c1">#        - 基于提供的上下文（如果有）和你的知识回答问题。</span>
    1608. 

  1608.     <span class="c1">#        - 如果答案来源于检索到的文档，请在回答中明确说明，例如：“根据提供的文档，……”。</span>
    1609. 

  1609.     <span class="c1"># </span>
    1610. 

  1610.     <span class="c1">#     2. **评估对话历史**：</span>
    1611. 

  1611.     <span class="c1">#        - 检查对话历史是否与当前问题相关（例如，是否涉及相同的话题、实体或问题背景）。</span>
    1612. 

  1612.     <span class="c1">#        - 如果对话历史与问题相关，请结合历史信息生成更准确的回答。</span>
    1613. 

  1613.     <span class="c1">#        - 如果对话历史无关（例如，仅包含问候或不相关的内容），忽略历史，仅基于上下文和问题回答。</span>
    1614. 

  1614.     <span class="c1"># </span>
    1615. 

  1615.     <span class="c1">#     3. **生成回答**：</span>
    1616. 

  1616.     <span class="c1">#        - 提供清晰、准确的回答，避免无关信息。</span>
    1617. 

  1617.     <span class="c1">#        - 如果上下文和历史消息均不足以回答问题，请回复：“信息不足，无法回答，请联系人工客服，电话：{phone}。”</span>
    1618. 

  1618.     <span class="c1"># </span>
    1619. 

  1619.     <span class="c1">#     **上下文**: {context}</span>
    1620. 

  1620.     <span class="c1">#     **对话历史**:</span>
    1621. 

  1621.     <span class="c1">#     {history}</span>
    1622. 

  1622.     <span class="c1">#     **问题**: {question}</span>
    1623. 

  1623.     <span class="c1"># </span>
    1624. 

  1624.     <span class="c1">#     **回答**:</span>
    1625. 

  1625.     <span class="c1">#     """,</span>
    1626. 

  1626.     <span class="c1">#         input_variables=["context", "history", "question", "phone"],</span>
    1627. 

  1627.     <span class="c1">#     )</span>
    1628. 

  1628. 

  1629.     <span class="c1"># 定义假设问题生成的 Prompt 模板</span>
    1630. 

  1630.     <span class="nd">@staticmethod</span>
    1631. 

  1631.     <span class="k">def</span><span class="w"> </span><span class="nf">hyde_prompt</span><span class="p">():</span>
    1632. 

  1632.         <span class="c1">#   创建并返回 PromptTemplate 对象</span>
    1633. 

  1633.         <span class="k">return</span> <span class="n">PromptTemplate</span><span class="p">(</span>
    1634. 

  1634.             <span class="n">template</span><span class="o">=</span><span class="s2">"""  </span>
    1635. 

  1635. <span class="s2">            假设你是用户，想了解以下问题，请生成一个简短的假设答案：  </span>
    1636. 

  1636. <span class="s2">            问题: </span><span class="si">{query}</span><span class="s2">  </span>
    1637. 

  1637. <span class="s2">            假设答案:  </span>
    1638. 

  1638. <span class="s2">            """</span><span class="p">,</span>
    1639. 

  1639.             <span class="c1">#   定义输入变量</span>
    1640. 

  1640.             <span class="n">input_variables</span><span class="o">=</span><span class="p">[</span><span class="s2">"query"</span><span class="p">],</span>
    1641. 

  1641.         <span class="p">)</span>
    1642. 

  1642. 

  1643.     <span class="c1">#   定义子查询生成的 Prompt 模板</span>
    1644. 

  1644.     <span class="nd">@staticmethod</span>
    1645. 

  1645.     <span class="k">def</span><span class="w"> </span><span class="nf">subquery_prompt</span><span class="p">():</span>
    1646. 

  1646.         <span class="c1">#   创建并返回 PromptTemplate 对象</span>
    1647. 

  1647.         <span class="k">return</span> <span class="n">PromptTemplate</span><span class="p">(</span>
    1648. 

  1648.             <span class="n">template</span><span class="o">=</span><span class="s2">"""  </span>
    1649. 

  1649. <span class="s2">            将以下复杂查询分解为多个简单子查询，每行一个子查询：  </span>
    1650. 

  1650. <span class="s2">            查询: </span><span class="si">{query}</span><span class="s2">  </span>
    1651. 

  1651. <span class="s2">            子查询:  </span>
    1652. 

  1652. <span class="s2">            """</span><span class="p">,</span>
    1653. 

  1653.             <span class="c1">#   定义输入变量</span>
    1654. 

  1654.             <span class="n">input_variables</span><span class="o">=</span><span class="p">[</span><span class="s2">"query"</span><span class="p">],</span>
    1655. 

  1655.         <span class="p">)</span>
    1656. 

  1656. 

  1657.     <span class="c1">#   定义回溯问题生成的 Prompt 模板</span>
    1658. 

  1658.     <span class="nd">@staticmethod</span>
    1659. 

  1659.     <span class="k">def</span><span class="w"> </span><span class="nf">backtracking_prompt</span><span class="p">():</span>
    1660. 

  1660.         <span class="c1">#   创建并返回 PromptTemplate 对象</span>
    1661. 

  1661.         <span class="k">return</span> <span class="n">PromptTemplate</span><span class="p">(</span>
    1662. 

  1662.             <span class="n">template</span><span class="o">=</span><span class="s2">"""  </span>
    1663. 

  1663. <span class="s2">            将以下复杂查询简化为一个更简单的问题：  </span>
    1664. 

  1664. <span class="s2">            查询: </span><span class="si">{query}</span><span class="s2">  </span>
    1665. 

  1665. <span class="s2">            简化问题:  </span>
    1666. 

  1666. <span class="s2">            """</span><span class="p">,</span>
    1667. 

  1667.             <span class="c1">#   定义输入变量</span>
    1668. 

  1668.             <span class="n">input_variables</span><span class="o">=</span><span class="p">[</span><span class="s2">"query"</span><span class="p">],</span>
    1669. 

  1669.         <span class="p">)</span>
    1670. 

  1670. </code></pre></div>
    1671. 

  1671. <h3 id="3">3 实现细节<a class="headerlink" href="#3" title="Permanent link">¶</a></h3>
    1672. 

  1672. <ol>
    1673. 

  1673. <li><strong><code>rag_prompt</code></strong>：</li>
    1674. 

  1674. <li><strong>作用</strong>：核心回答模板，结合检索到的上下文生成最终答案。</li>
    1675. 

  1675. <li><strong>输入变量</strong>：<code>context</code>（检索文档内容）、<code>question</code>（用户查询）、<code>phone</code>（客服电话）。</li>
    1676. 

  1676. <li><strong>设计逻辑</strong>：支持有无上下文的回答，并提供兜底回复，确保用户体验。</li>
    1677. 

  1677. <li><strong><code>hyde_prompt</code></strong>：</li>
    1678. 

  1678. <li><strong>作用</strong>：生成假设答案，用于HyDE（Hypothetical Document Embeddings）策略，优化抽象查询的检索。</li>
    1679. 

  1679. <li><strong>输入变量</strong>：<code>query</code>（用户查询）。</li>
    1680. 

  1680. <li><strong>设计逻辑</strong>：通过生成假设答案，间接增强查询与文档的语义匹配。</li>
    1681. 

  1681. <li><strong><code>subquery_prompt</code></strong>：</li>
    1682. 

  1682. <li><strong>作用</strong>：将复杂查询分解为多个子查询，适合涉及多方面的查询。</li>
    1683. 

  1683. <li><strong>输入变量</strong>：<code>query</code>（用户查询）。</li>
    1684. 

  1684. <li><strong>设计逻辑</strong>：分解复杂问题以提高检索覆盖率。</li>
    1685. 

  1685. <li><strong><code>backtracking_prompt</code></strong>：</li>
    1686. 

  1686. <li><strong>作用</strong>：将复杂查询简化为更基础的问题，便于检索。</li>
    1687. 

  1687. <li><strong>输入变量</strong>：<code>query</code>（用户查询）。</li>
    1688. 

  1688. <li><strong>设计逻辑</strong>：通过简化查询降低检索难度。</li>
    1689. 

  1689. </ol>
    1690. 

  1690. <hr>
    1691. 

  1691. <h2 id="_2">二、查询分类<a class="headerlink" href="#_2" title="Permanent link">¶</a></h2>
    1692. 

  1692. <hr>
    1693. 

  1693. <p><code>QueryClassifier</code> 是 EnterpriseKB-RAG 系统的核心组件，负责将用户查询分为“通用知识”和“企业知识”两类，以决定查询路由到知识库还是咨询接口。本模块介绍基于 BERT 的优化实现，替换传统 TF-IDF 模型，利用 5000 条混合数据集（<code>training_dataset_hybrid_5000.json</code>）进行训练，并解决评估中的标签处理问题。</p>
    1694. 

  1694. <h3 id="1_1">1 功能概述<a class="headerlink" href="#1_1" title="Permanent link">¶</a></h3>
    1695. 

  1695. <p><code>QueryClassifier</code> 提供以下功能：</p>
    1696. 

  1696. <ul>
    1697. 

  1697. <li>
    1698. 

  1698. <p><strong>数据加载</strong>：读取 5000 条 JSON 数据集，包含查询和标签（“通用知识”或“企业知识”）。</p>
    1699. 

  1699. </li>
    1700. 

  1700. <li>
    1701. 

  1701. <p><strong>BERT 训练</strong>：使用 <code>bert-base-chinese</code> 模型，微调二分类任务，准确率达 90%+。</p>
    1702. 

  1702. </li>
    1703. 

  1703. <li><strong>评估优化</strong>：直接处理数字标签（0 或 1），生成分类报告和混淆矩阵。</li>
    1704. 

  1704. <li><strong>预测接口</strong>：支持实时分类，集成到 EnterpriseKB-RAG 系统。</li>
    1705. 

  1705. </ul>
    1706. 

  1706. <h3 id="2_1">2 代码实现<a class="headerlink" href="#2_1" title="Permanent link">¶</a></h3>
    1707. 

  1707. <div class="highlight"><pre><span></span><code><span class="c1"># 导入标准库</span>
    1708. 

  1708. <span class="kn">import</span><span class="w"> </span><span class="nn">json</span>
    1709. 

  1709. <span class="kn">import</span><span class="w"> </span><span class="nn">os</span>
    1710. 

  1710. <span class="c1"># 导入 PyTorch</span>
    1711. 

  1711. <span class="kn">import</span><span class="w"> </span><span class="nn">torch</span>
    1712. 

  1712. <span class="c1"># 导入日志</span>
    1713. 

  1713. <span class="kn">import</span><span class="w"> </span><span class="nn">sys</span>
    1714. 

  1714. <span class="kn">from</span><span class="w"> </span><span class="nn">base.logger</span><span class="w"> </span><span class="kn">import</span> <span class="n">logger</span>
    1715. 

  1715. <span class="c1"># 导入numpy</span>
    1716. 

  1716. <span class="kn">import</span><span class="w"> </span><span class="nn">numpy</span><span class="w"> </span><span class="k">as</span><span class="w"> </span><span class="nn">np</span>
    1717. 

  1717. <span class="c1"># 导入 Transformers 库</span>
    1718. 

  1718. <span class="kn">from</span><span class="w"> </span><span class="nn">transformers</span><span class="w"> </span><span class="kn">import</span> <span class="n">BertTokenizer</span><span class="p">,</span> <span class="n">BertForSequenceClassification</span>
    1719. 

  1719. <span class="kn">from</span><span class="w"> </span><span class="nn">transformers</span><span class="w"> </span><span class="kn">import</span> <span class="n">Trainer</span><span class="p">,</span> <span class="n">TrainingArguments</span>
    1720. 

  1720. <span class="c1"># 导入train_test_split</span>
    1721. 

  1721. <span class="kn">from</span><span class="w"> </span><span class="nn">sklearn.model_selection</span><span class="w"> </span><span class="kn">import</span> <span class="n">train_test_split</span>
    1722. 

  1722. <span class="kn">from</span><span class="w"> </span><span class="nn">sklearn.metrics</span><span class="w"> </span><span class="kn">import</span> <span class="n">classification_report</span><span class="p">,</span> <span class="n">confusion_matrix</span>
    1723. 

  1723. 

  1724. <span class="n">current_dir</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">dirname</span><span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="vm">__file__</span><span class="p">))</span>
    1725. 

  1725. <span class="n">rag_qa_path</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">dirname</span><span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="n">current_dir</span><span class="p">)))</span>
    1726. 

  1726. <span class="n">project_root</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">dirname</span><span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="n">rag_qa_path</span><span class="p">)))</span>
    1727. 

  1727. <span class="n">sys</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">insert</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="n">project_root</span><span class="p">)</span>
    1728. 

  1728. 

  1729. <span class="k">class</span><span class="w"> </span><span class="nc">QueryClassifier</span><span class="p">:</span>
    1730. 

  1730.     <span class="k">def</span><span class="w"> </span><span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">model_path</span><span class="o">=</span><span class="s1">'models/bert_query_classifier'</span><span class="p">):</span>
    1731. 

  1731.        <span class="c1"># 加载bert</span>
    1732. 

  1732.         <span class="bp">self</span><span class="o">.</span><span class="n">pre_trained_model_path</span> <span class="o">=</span> <span class="sa">f</span><span class="s1">'</span><span class="si">{</span><span class="n">rag_qa_path</span><span class="si">}</span><span class="s1">/models/bert-base-chinese'</span>
    1733. 

  1733.         <span class="c1"># 模型训练以后保存的位置</span>
    1734. 

  1734.         <span class="bp">self</span><span class="o">.</span><span class="n">model_path</span> <span class="o">=</span> <span class="n">model_path</span>
    1735. 

  1735.         <span class="c1"># 加载 BERT 分词器</span>
    1736. 

  1736.         <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span> <span class="o">=</span> <span class="n">BertTokenizer</span><span class="o">.</span><span class="n">from_pretrained</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">pre_trained_model_path</span><span class="p">)</span>
    1737. 

  1737.         <span class="c1"># 初始化模型</span>
    1738. 

  1738.         <span class="bp">self</span><span class="o">.</span><span class="n">model</span> <span class="o">=</span> <span class="kc">None</span>
    1739. 

  1739.         <span class="c1"># 确定设备（GPU 或 CPU）</span>
    1740. 

  1740.         <span class="bp">self</span><span class="o">.</span><span class="n">device</span> <span class="o">=</span>  <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s2">"cuda"</span> <span class="k">if</span> <span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">is_available</span><span class="p">()</span> <span class="k">else</span> <span class="s2">"mps"</span> <span class="k">if</span> <span class="n">torch</span><span class="o">.</span><span class="n">mps</span><span class="o">.</span><span class="n">is_available</span><span class="p">()</span> <span class="k">else</span> <span class="s2">"cpu"</span> <span class="p">)</span>
    1741. 

  1741.         <span class="c1"># 记录设备信息</span>
    1742. 

  1742.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="sa">f</span><span class="s2">"使用设备: </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">device</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
    1743. 

  1743.         <span class="c1"># 定义标签映射</span>
    1744. 

  1744.         <span class="bp">self</span><span class="o">.</span><span class="n">label_map</span> <span class="o">=</span> <span class="p">{</span><span class="s2">"通用知识"</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s2">"企业知识"</span><span class="p">:</span> <span class="mi">1</span><span class="p">}</span>
    1745. 

  1745.         <span class="c1"># 加载模型</span>
    1746. 

  1746.         <span class="bp">self</span><span class="o">.</span><span class="n">load_model</span><span class="p">()</span>
    1747. 

  1747. 

  1748.     <span class="k">def</span><span class="w"> </span><span class="nf">load_model</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
    1749. 

  1749.         <span class="c1"># 检查模型路径是否存在</span>
    1750. 

  1750.         <span class="k">if</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">exists</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="p">):</span>
    1751. 

  1751.             <span class="c1"># 加载预训练模型</span>
    1752. 

  1752.             <span class="bp">self</span><span class="o">.</span><span class="n">model</span> <span class="o">=</span> <span class="n">BertForSequenceClassification</span><span class="o">.</span><span class="n">from_pretrained</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="p">)</span>
    1753. 

  1753.             <span class="c1"># 将模型移到指定设备</span>
    1754. 

  1754.             <span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">device</span><span class="p">)</span>
    1755. 

  1755.             <span class="c1"># 记录加载成功的日志</span>
    1756. 

  1756.             <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="sa">f</span><span class="s2">"加载模型: </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
    1757. 

  1757.         <span class="k">else</span><span class="p">:</span>
    1758. 

  1758.             <span class="c1"># 初始化新模型</span>
    1759. 

  1759.             <span class="bp">self</span><span class="o">.</span><span class="n">model</span> <span class="o">=</span> <span class="n">BertForSequenceClassification</span><span class="o">.</span><span class="n">from_pretrained</span><span class="p">(</span><span class="s2">"bert-base-chinese"</span><span class="p">,</span> <span class="n">num_labels</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span>
    1760. 

  1760.             <span class="c1"># 将模型移到指定设备</span>
    1761. 

  1761.             <span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">device</span><span class="p">)</span>
    1762. 

  1762.             <span class="c1"># 记录初始化模型的日志</span>
    1763. 

  1763.             <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s2">"初始化新 BERT 模型"</span><span class="p">)</span>
    1764. 

  1764. 

  1765.     <span class="k">def</span><span class="w"> </span><span class="nf">save_model</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
    1766. 

  1766. <span class="w">        </span><span class="sd">"""保存模型"""</span>
    1767. 

  1767.         <span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="o">.</span><span class="n">save_pretrained</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="p">)</span>
    1768. 

  1768.         <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">save_pretrained</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="p">)</span>
    1769. 

  1769.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="sa">f</span><span class="s2">"模型保存至: </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">model_path</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
    1770. 

  1770. 

  1771.     <span class="k">def</span><span class="w"> </span><span class="nf">preprocess_data</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">texts</span><span class="p">,</span> <span class="n">labels</span><span class="p">):</span>
    1772. 

  1772. <span class="w">        </span><span class="sd">"""预处理数据为 BERT 输入格式"""</span>
    1773. 

  1773.         <span class="n">encodings</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">(</span>
    1774. 

  1774.             <span class="n">texts</span><span class="p">,</span>
    1775. 

  1775.             <span class="n">truncation</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
    1776. 

  1776.             <span class="n">padding</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
    1777. 

  1777.             <span class="n">max_length</span><span class="o">=</span><span class="mi">128</span><span class="p">,</span>
    1778. 

  1778.             <span class="n">return_tensors</span><span class="o">=</span><span class="s2">"pt"</span>
    1779. 

  1779.         <span class="p">)</span>
    1780. 

  1780.         <span class="k">return</span> <span class="n">encodings</span><span class="p">,</span> <span class="p">[</span><span class="bp">self</span><span class="o">.</span><span class="n">label_map</span><span class="p">[</span><span class="n">label</span><span class="p">]</span> <span class="k">for</span> <span class="n">label</span> <span class="ow">in</span> <span class="n">labels</span><span class="p">]</span>
    1781. 

  1781. 

  1782.     <span class="k">def</span><span class="w"> </span><span class="nf">create_dataset</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">encodings</span><span class="p">,</span> <span class="n">labels</span><span class="p">):</span>
    1783. 

  1783. <span class="w">        </span><span class="sd">"""创建 PyTorch 数据集"""</span>
    1784. 

  1784. 

  1785.         <span class="k">class</span><span class="w"> </span><span class="nc">Dataset</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">utils</span><span class="o">.</span><span class="n">data</span><span class="o">.</span><span class="n">Dataset</span><span class="p">):</span>
    1786. 

  1786.             <span class="k">def</span><span class="w"> </span><span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">encodings</span><span class="p">,</span> <span class="n">labels</span><span class="p">):</span>
    1787. 

  1787.                 <span class="bp">self</span><span class="o">.</span><span class="n">encodings</span> <span class="o">=</span> <span class="n">encodings</span>
    1788. 

  1788.                 <span class="bp">self</span><span class="o">.</span><span class="n">labels</span> <span class="o">=</span> <span class="n">labels</span>
    1789. 

  1789. 

  1790.             <span class="k">def</span><span class="w"> </span><span class="fm">__getitem__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">idx</span><span class="p">):</span>
    1791. 

  1791.                 <span class="n">item</span> <span class="o">=</span> <span class="p">{</span><span class="n">key</span><span class="p">:</span> <span class="n">val</span><span class="p">[</span><span class="n">idx</span><span class="p">]</span> <span class="k">for</span> <span class="n">key</span><span class="p">,</span> <span class="n">val</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">encodings</span><span class="o">.</span><span class="n">items</span><span class="p">()}</span>
    1792. 

  1792.                 <span class="n">item</span><span class="p">[</span><span class="s2">"labels"</span><span class="p">]</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">labels</span><span class="p">[</span><span class="n">idx</span><span class="p">])</span>
    1793. 

  1793.                 <span class="k">return</span> <span class="n">item</span>
    1794. 

  1794. 

  1795.             <span class="k">def</span><span class="w"> </span><span class="fm">__len__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
    1796. 

  1796.                 <span class="k">return</span> <span class="nb">len</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">labels</span><span class="p">)</span>
    1797. 

  1797. 

  1798.         <span class="k">return</span> <span class="n">Dataset</span><span class="p">(</span><span class="n">encodings</span><span class="p">,</span> <span class="n">labels</span><span class="p">)</span>
    1799. 

  1799. 

  1800.     <span class="k">def</span><span class="w"> </span><span class="nf">train_model</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">data_file</span><span class="o">=</span><span class="s2">"training_dataset_hybrid_5000.json"</span><span class="p">):</span>
    1801. 

  1801. <span class="w">        </span><span class="sd">"""训练 BERT 分类模型"""</span>
    1802. 

  1802.         <span class="c1"># 加载数据集</span>
    1803. 

  1803.         <span class="k">if</span> <span class="ow">not</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">exists</span><span class="p">(</span><span class="n">data_file</span><span class="p">):</span>
    1804. 

  1804.             <span class="n">logger</span><span class="o">.</span><span class="n">error</span><span class="p">(</span><span class="sa">f</span><span class="s2">"数据集文件 </span><span class="si">{</span><span class="n">data_file</span><span class="si">}</span><span class="s2"> 不存在"</span><span class="p">)</span>
    1805. 

  1805.             <span class="k">raise</span> <span class="ne">FileNotFoundError</span><span class="p">(</span><span class="sa">f</span><span class="s2">"数据集文件 </span><span class="si">{</span><span class="n">data_file</span><span class="si">}</span><span class="s2"> 不存在"</span><span class="p">)</span>
    1806. 

  1806. 

  1807.         <span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">data_file</span><span class="p">,</span> <span class="s2">"r"</span><span class="p">,</span> <span class="n">encoding</span><span class="o">=</span><span class="s2">"utf-8"</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
    1808. 

  1808.             <span class="n">data</span> <span class="o">=</span> <span class="p">[</span><span class="n">json</span><span class="o">.</span><span class="n">loads</span><span class="p">(</span><span class="n">value</span><span class="p">)</span> <span class="k">for</span> <span class="n">value</span> <span class="ow">in</span> <span class="n">f</span><span class="o">.</span><span class="n">readlines</span><span class="p">()]</span>
    1809. 

  1809. 

  1810.         <span class="n">texts</span> <span class="o">=</span> <span class="p">[</span><span class="n">item</span><span class="p">[</span><span class="s2">"query"</span><span class="p">]</span> <span class="k">for</span> <span class="n">item</span> <span class="ow">in</span> <span class="n">data</span><span class="p">]</span>
    1811. 

  1811.         <span class="n">labels</span> <span class="o">=</span> <span class="p">[</span><span class="n">item</span><span class="p">[</span><span class="s2">"label"</span><span class="p">]</span> <span class="k">for</span> <span class="n">item</span> <span class="ow">in</span> <span class="n">data</span><span class="p">]</span>
    1812. 

  1812. 

  1813.         <span class="c1"># 数据划分</span>
    1814. 

  1814.         <span class="n">train_texts</span><span class="p">,</span> <span class="n">val_texts</span><span class="p">,</span> <span class="n">train_labels</span><span class="p">,</span> <span class="n">val_labels</span> <span class="o">=</span> <span class="n">train_test_split</span><span class="p">(</span>
    1815. 

  1815.             <span class="n">texts</span><span class="p">,</span> <span class="n">labels</span><span class="p">,</span> <span class="n">test_size</span><span class="o">=</span><span class="mf">0.2</span><span class="p">,</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">42</span>
    1816. 

  1816.         <span class="p">)</span>
    1817. 

  1817. 

  1818.         <span class="c1"># 预处理</span>
    1819. 

  1819.         <span class="n">train_encodings</span><span class="p">,</span> <span class="n">train_labels</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">preprocess_data</span><span class="p">(</span><span class="n">train_texts</span><span class="p">,</span> <span class="n">train_labels</span><span class="p">)</span>
    1820. 

  1820.         <span class="n">val_encodings</span><span class="p">,</span> <span class="n">val_labels</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">preprocess_data</span><span class="p">(</span><span class="n">val_texts</span><span class="p">,</span> <span class="n">val_labels</span><span class="p">)</span>
    1821. 

  1821. 

  1822.         <span class="c1"># 创建数据集</span>
    1823. 

  1823.         <span class="n">train_dataset</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">create_dataset</span><span class="p">(</span><span class="n">train_encodings</span><span class="p">,</span> <span class="n">train_labels</span><span class="p">)</span>
    1824. 

  1824.         <span class="c1"># print(f'train_dataset--》{train_dataset[0]}')</span>
    1825. 

  1825.         <span class="n">val_dataset</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">create_dataset</span><span class="p">(</span><span class="n">val_encodings</span><span class="p">,</span> <span class="n">val_labels</span><span class="p">)</span>
    1826. 

  1826.         <span class="c1">#</span>
    1827. 

  1827.         <span class="c1"># 设置训练参数</span>
    1828. 

  1828.         <span class="n">training_args</span> <span class="o">=</span> <span class="n">TrainingArguments</span><span class="p">(</span>
    1829. 

  1829.             <span class="c1"># 设置模型和检查点保存的目录路径</span>
    1830. 

  1830.             <span class="n">output_dir</span><span class="o">=</span><span class="s2">"./bert_results"</span><span class="p">,</span>
    1831. 

  1831.             <span class="c1"># 设置训练的总轮数为3轮</span>
    1832. 

  1832.             <span class="n">num_train_epochs</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span>
    1833. 

  1833.             <span class="c1"># 设置每个设备（GPU/CPU）上的训练批次大小为8</span>
    1834. 

  1834.             <span class="n">per_device_train_batch_size</span><span class="o">=</span><span class="mi">8</span><span class="p">,</span>
    1835. 

  1835.             <span class="c1"># 设置每个设备（GPU/CPU）上的评估批次大小为8</span>
    1836. 

  1836.             <span class="n">per_device_eval_batch_size</span><span class="o">=</span><span class="mi">8</span><span class="p">,</span>
    1837. 

  1837.             <span class="c1"># 设置学习率预热步数为500步，训练初期学习率从0逐渐增加到设定值</span>
    1838. 

  1838.             <span class="n">warmup_steps</span><span class="o">=</span><span class="mi">500</span><span class="p">,</span>
    1839. 

  1839.             <span class="c1"># 设置权重衰减系数为0.01，用于防止过拟合</span>
    1840. 

  1840.             <span class="n">weight_decay</span><span class="o">=</span><span class="mf">0.01</span><span class="p">,</span>
    1841. 

  1841.             <span class="c1"># 设置日志文件保存的目录路径</span>
    1842. 

  1842.             <span class="n">logging_dir</span><span class="o">=</span><span class="s2">"./bert_logs"</span><span class="p">,</span>
    1843. 

  1843.             <span class="c1"># 设置每10个训练步骤记录一次日志</span>
    1844. 

  1844.             <span class="n">logging_steps</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span>
    1845. 

  1845.             <span class="c1"># 设置评估策略为每个epoch结束后进行评估</span>
    1846. 

  1846.             <span class="n">evaluation_strategy</span><span class="o">=</span><span class="s2">"epoch"</span><span class="p">,</span>
    1847. 

  1847.             <span class="c1"># 设置模型保存策略为每个epoch结束后保存</span>
    1848. 

  1848.             <span class="n">save_strategy</span><span class="o">=</span><span class="s2">"epoch"</span><span class="p">,</span>
    1849. 

  1849.             <span class="c1"># 设置训练结束后加载最佳模型而非最后一个模型</span>
    1850. 

  1850.             <span class="n">load_best_model_at_end</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
    1851. 

  1851.             <span class="c1"># 设置最多保存1个检查点文件，超出时自动删除旧的</span>
    1852. 

  1852.             <span class="n">save_total_limit</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span>
    1853. 

  1853.             <span class="c1"># 设置用于判断最佳模型的指标为评估损失</span>
    1854. 

  1854.             <span class="n">metric_for_best_model</span><span class="o">=</span><span class="s2">"eval_loss"</span><span class="p">,</span>
    1855. 

  1855.             <span class="c1"># 禁用FP16混合精度训练，使用FP32精度</span>
    1856. 

  1856.             <span class="n">fp16</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span>
    1857. 

  1857.         <span class="p">)</span>
    1858. 

  1858. 

  1859.         <span class="c1"># 初始化 Trainer</span>
    1860. 

  1860.         <span class="n">trainer</span> <span class="o">=</span> <span class="n">Trainer</span><span class="p">(</span>
    1861. 

  1861.             <span class="c1"># 传入要训练的模型实例</span>
    1862. 

  1862.             <span class="n">model</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="p">,</span>
    1863. 

  1863.             <span class="c1"># 传入上面定义的训练参数配置</span>
    1864. 

  1864.             <span class="n">args</span><span class="o">=</span><span class="n">training_args</span><span class="p">,</span>
    1865. 

  1865.             <span class="c1"># 传入训练数据集</span>
    1866. 

  1866.             <span class="n">train_dataset</span><span class="o">=</span><span class="n">train_dataset</span><span class="p">,</span>
    1867. 

  1867.             <span class="c1"># 传入验证数据集，用于训练过程中评估模型性能</span>
    1868. 

  1868.             <span class="n">eval_dataset</span><span class="o">=</span><span class="n">val_dataset</span><span class="p">,</span>
    1869. 

  1869.             <span class="c1"># 传入计算评估指标的函数，用于在验证集上计算准确率等指标</span>
    1870. 

  1870.             <span class="n">compute_metrics</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">compute_metrics</span>
    1871. 

  1871.         <span class="p">)</span>
    1872. 

  1872.         <span class="c1"># 训练模型</span>
    1873. 

  1873.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s2">"开始训练 BERT 模型..."</span><span class="p">)</span>
    1874. 

  1874.         <span class="n">trainer</span><span class="o">.</span><span class="n">train</span><span class="p">()</span>
    1875. 

  1875.         <span class="bp">self</span><span class="o">.</span><span class="n">save_model</span><span class="p">()</span>
    1876. 

  1876. 

  1877.         <span class="c1"># 评估模型</span>
    1878. 

  1878.         <span class="bp">self</span><span class="o">.</span><span class="n">evaluate_model</span><span class="p">(</span><span class="n">val_texts</span><span class="p">,</span> <span class="n">val_labels</span><span class="p">)</span>
    1879. 

  1879. 

  1880.     <span class="k">def</span><span class="w"> </span><span class="nf">compute_metrics</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">eval_pred</span><span class="p">):</span>
    1881. 

  1881. <span class="w">        </span><span class="sd">"""计算评估指标"""</span>
    1882. 

  1882.         <span class="n">logits</span><span class="p">,</span> <span class="n">labels</span> <span class="o">=</span> <span class="n">eval_pred</span>
    1883. 

  1883.         <span class="n">predictions</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">argmax</span><span class="p">(</span><span class="n">logits</span><span class="p">,</span> <span class="n">axis</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span>
    1884. 

  1884.         <span class="n">accuracy</span> <span class="o">=</span> <span class="p">(</span><span class="n">predictions</span> <span class="o">==</span> <span class="n">labels</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">()</span>
    1885. 

  1885.         <span class="k">return</span> <span class="p">{</span><span class="s2">"accuracy"</span><span class="p">:</span> <span class="n">accuracy</span><span class="p">}</span>
    1886. 

  1886. 

  1887.     <span class="k">def</span><span class="w"> </span><span class="nf">evaluate_model</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">texts</span><span class="p">,</span> <span class="n">labels</span><span class="p">):</span>
    1888. 

  1888. <span class="w">        </span><span class="sd">"""评估模型性能"""</span>
    1889. 

  1889.         <span class="c1"># 仅对 texts 进行分词，labels 已为数字</span>
    1890. 

  1890.         <span class="n">encodings</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">(</span>
    1891. 

  1891.             <span class="n">texts</span><span class="p">,</span>
    1892. 

  1892.             <span class="n">truncation</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
    1893. 

  1893.             <span class="n">padding</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
    1894. 

  1894.             <span class="n">max_length</span><span class="o">=</span><span class="mi">128</span><span class="p">,</span>
    1895. 

  1895.             <span class="n">return_tensors</span><span class="o">=</span><span class="s2">"pt"</span>
    1896. 

  1896.         <span class="p">)</span>
    1897. 

  1897.         <span class="n">dataset</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">create_dataset</span><span class="p">(</span><span class="n">encodings</span><span class="p">,</span> <span class="n">labels</span><span class="p">)</span>
    1898. 

  1898. 

  1899.         <span class="n">trainer</span> <span class="o">=</span> <span class="n">Trainer</span><span class="p">(</span><span class="n">model</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="p">)</span>
    1900. 

  1900.         <span class="n">predictions</span> <span class="o">=</span> <span class="n">trainer</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">dataset</span><span class="p">)</span>
    1901. 

  1901.         <span class="n">pred_labels</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">argmax</span><span class="p">(</span><span class="n">predictions</span><span class="o">.</span><span class="n">predictions</span><span class="p">,</span> <span class="n">axis</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span>
    1902. 

  1902.         <span class="n">true_labels</span> <span class="o">=</span> <span class="n">labels</span>  <span class="c1"># 直接使用数字标签</span>
    1903. 

  1903. 

  1904.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s2">"分类报告:"</span><span class="p">)</span>
    1905. 

  1905.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="n">classification_report</span><span class="p">(</span>
    1906. 

  1906.             <span class="n">true_labels</span><span class="p">,</span>
    1907. 

  1907.             <span class="n">pred_labels</span><span class="p">,</span>
    1908. 

  1908.             <span class="n">target_names</span><span class="o">=</span><span class="p">[</span><span class="s2">"通用知识"</span><span class="p">,</span> <span class="s2">"企业知识"</span><span class="p">]</span>
    1909. 

  1909.         <span class="p">))</span>
    1910. 

  1910.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="s2">"混淆矩阵:"</span><span class="p">)</span>
    1911. 

  1911.         <span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="n">confusion_matrix</span><span class="p">(</span><span class="n">true_labels</span><span class="p">,</span> <span class="n">pred_labels</span><span class="p">))</span>
    1912. 

  1912. 

  1913.     <span class="k">def</span><span class="w"> </span><span class="nf">predict_category</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">query</span><span class="p">):</span>
    1914. 

  1914.         <span class="c1"># 检查模型是否加载</span>
    1915. 

  1915.         <span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">model</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
    1916. 

  1916.             <span class="c1"># 模型未加载，记录错误</span>
    1917. 

  1917.             <span class="n">logger</span><span class="o">.</span><span class="n">error</span><span class="p">(</span><span class="s2">"模型未训练或加载"</span><span class="p">)</span>
    1918. 

  1918.             <span class="c1"># 默认返回通用知识</span>
    1919. 

  1919.             <span class="k">return</span> <span class="s2">"通用知识"</span>
    1920. 

  1920.         <span class="c1"># 对查询进行编码</span>
    1921. 

  1921.         <span class="n">encoding</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">(</span><span class="n">query</span><span class="p">,</span> <span class="n">truncation</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">padding</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">max_length</span><span class="o">=</span><span class="mi">128</span><span class="p">,</span> <span class="n">return_tensors</span><span class="o">=</span><span class="s2">"pt"</span><span class="p">)</span>
    1922. 

  1922.         <span class="c1"># 将编码移到指定设备</span>
    1923. 

  1923.         <span class="n">encoding</span> <span class="o">=</span> <span class="p">{</span><span class="n">k</span><span class="p">:</span> <span class="n">v</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">device</span><span class="p">)</span> <span class="k">for</span> <span class="n">k</span><span class="p">,</span> <span class="n">v</span> <span class="ow">in</span> <span class="n">encoding</span><span class="o">.</span><span class="n">items</span><span class="p">()}</span>
    1924. 

  1924.         <span class="c1"># 不计算梯度，进行预测</span>
    1925. 

  1925.         <span class="k">with</span> <span class="n">torch</span><span class="o">.</span><span class="n">no_grad</span><span class="p">():</span>
    1926. 

  1926.             <span class="c1"># 获取模型输出</span>
    1927. 

  1927.             <span class="n">outputs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">model</span><span class="p">(</span><span class="o">**</span><span class="n">encoding</span><span class="p">)</span>
    1928. 

  1928.             <span class="c1"># 获取预测结果</span>
    1929. 

  1929.             <span class="n">prediction</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">argmax</span><span class="p">(</span><span class="n">outputs</span><span class="o">.</span><span class="n">logits</span><span class="p">,</span> <span class="n">dim</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span><span class="o">.</span><span class="n">item</span><span class="p">()</span>
    1930. 

  1930.         <span class="c1"># 根据预测结果返回类别</span>
    1931. 

  1931.         <span class="k">return</span> <span class="s2">"企业知识"</span> <span class="k">if</span> <span class="n">prediction</span> <span class="o">==</span> <span class="mi">1</span> <span class="k">else</span> <span class="s2">"通用知识"</span>
    1932. 

  1932. 

  1933. <span class="k">if</span> <span class="vm">__name__</span> <span class="o">==</span> <span class="s2">"__main__"</span><span class="p">:</span>
    1934. 

  1934.     <span class="c1"># 初始化分类器</span>
    1935. 

  1935.     <span class="n">classifier</span> <span class="o">=</span> <span class="n">QueryClassifier</span><span class="p">(</span><span class="n">model_path</span><span class="o">=</span><span class="s2">"bert_query_classifier"</span><span class="p">)</span>
    1936. 

  1936. 

  1937.     <span class="c1"># 训练模型</span>
    1938. 

  1938.     <span class="c1"># classifier.train_model(data_file='../classify_data/model_generic_5000.json')</span>
    1939. 

  1939.     <span class="c1"># 示例预测</span>
    1940. 

  1940.     <span class="n">test_queries</span> <span class="o">=</span> <span class="p">[</span>
    1941. 

  1941.         <span class="s2">"AI知识域的知识文档目录是什么"</span><span class="p">,</span>
    1942. 

  1942.         <span class="s2">"JAVA知识文档费用多少？"</span><span class="p">,</span>
    1943. 

  1943.         <span class="s2">"5*9等于多少？"</span><span class="p">,</span>
    1944. 

  1944.         <span class="s2">"AI内部赋能有哪些支持人员？"</span>
    1945. 

  1945.     <span class="p">]</span>
    1946. 

  1946.     <span class="k">for</span> <span class="n">query</span> <span class="ow">in</span> <span class="n">test_queries</span><span class="p">:</span>
    1947. 

  1947.         <span class="n">category</span> <span class="o">=</span> <span class="n">classifier</span><span class="o">.</span><span class="n">predict_category</span><span class="p">(</span><span class="n">query</span><span class="p">)</span>
    1948. 

  1948.         <span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"查询: </span><span class="si">{</span><span class="n">query</span><span class="si">}</span><span class="s2"> -&gt; 分类: </span><span class="si">{</span><span class="n">category</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
    1949. 

  1949. </code></pre></div>
    1950. 

  1950. <h3 id="3_1">3 实现细节<a class="headerlink" href="#3_1" title="Permanent link">¶</a></h3>
    1951. 

  1951. <ol>
    1952. 

  1952. <li><strong><code>__init__</code> 方法</strong>：</li>
    1953. 

  1953. <li><strong>作用</strong>：初始化 BERT 分词器（<code>bert-base-chinese</code>）和模型，支持二分类。</li>
    1954. 

  1954. <li><strong>优化</strong>：设备选择优先 CUDA，若不可用则回退到 CPU，禁用 MPS（适配 macOS 低版本）。</li>
    1955. 

  1955. <li>
    1956. 

  1956. <p><strong>标签映射</strong>：定义 <code>label_map = {"通用知识": 0, "企业知识": 1}</code>，用于训练时字符串标签转换。</p>
    1957. 

  1957. </li>
    1958. 

  1958. <li>
    1959. 

  1959. <p><strong><code>preprocess_data</code> 方法</strong>：</p>
    1960. 

  1960. </li>
    1961. 

  1961. <li><strong>作用</strong>：将查询文本分词为 BERT 输入（ID 和注意力掩码），将字符串标签转换为数字（0 或 1）。</li>
    1962. 

  1962. <li>
    1963. 

  1963. <p><strong>细节</strong>：设置 <code>max_length=128</code>，平衡效率和信息完整性。</p>
    1964. 

  1964. </li>
    1965. 

  1965. <li>
    1966. 

  1966. <p><strong><code>create_dataset</code> 方法</strong>：</p>
    1967. 

  1967. </li>
    1968. 

  1968. <li><strong>作用</strong>：构建 PyTorch 数据集，适配 <code>Trainer</code> 的输入格式。</li>
    1969. 

  1969. <li>
    1970. 

  1970. <p><strong>实现</strong>：确保 <code>labels</code> 为数字，兼容训练和评估。</p>
    1971. 

  1971. </li>
    1972. 

  1972. <li>
    1973. 

  1973. <p><strong><code>train_model</code> 方法</strong>：</p>
    1974. 

  1974. </li>
    1975. 

  1975. <li><strong>作用</strong>：加载 5000 条数据集，划分 80% 训练（4000 条）和 20% 验证（1000 条），微调 BERT 模型。</li>
    1976. 

  1976. <li><strong>参数</strong>：<ul>
    1977. 

  1977. <li><code>num_train_epochs=3</code>：训练 3 轮，适合中等规模数据集。</li>
    1978. 

  1978. <li><code>per_device_train_batch_size=8</code>：平衡内存和速度。</li>
    1979. 

  1979. <li><code>fp16=False</code>：禁用混合精度，兼容 PyTorch 2.5 和 CPU，如果为True，采用混合精度，GPU训练。</li>
    1980. 

  1980. </ul>
    1981. 

  1981. </li>
    1982. 

  1982. <li>
    1983. 

  1983. <p><strong>流程</strong>：</p>
    1984. 

  1984. <ul>
    1985. 

  1985. <li>加载 <code>training_dataset_hybrid_5000.json</code>。</li>
    1986. 

  1986. <li>预处理数据，将标签转换为数字。</li>
    1987. 

  1987. <li>使用 <code>Trainer</code> 训练，自动保存最佳模型。</li>
    1988. 

  1988. </ul>
    1989. 

  1989. </li>
    1990. 

  1990. <li>
    1991. 

  1991. <p><strong><code>evaluate_model</code> 方法（优化重点）</strong>：</p>
    1992. 

  1992. </li>
    1993. 

  1993. <li><strong>作用</strong>：在验证集上评估模型，生成分类报告和混淆矩阵。</li>
    1994. 

  1994. <li><strong>修复</strong>：<ul>
    1995. 

  1995. <li>问题：原始代码重复映射数字标签（<code>0</code>, <code>1</code>）到 <code>label_map</code>，导致 <code>KeyError: 1</code>。</li>
    1996. 

  1996. <li>修复：直接使用传入的数字标签（<code>labels</code>），仅对 <code>texts</code> 分词。</li>
    1997. 

  1997. <li>逻辑：<code>true_labels = labels</code>，确保与预测标签一致。</li>
    1998. 

  1998. </ul>
    1999. 

  1999. </li>
    2000. 

  2000. <li>
    2001. 

  2001. <p><strong>输出</strong>：精确率、召回率、F1 分数和混淆矩阵。</p>
    2002. 

  2002. </li>
    2003. 

  2003. <li>
    2004. 

  2004. <p><strong><code>predict_category</code> 方法</strong>：</p>
    2005. 

  2005. </li>
    2006. 

  2006. <li><strong>作用</strong>：对单条查询分类，返回“通用知识”或“企业知识”。</li>
    2007. 

  2007. <li><strong>实现</strong>：分词后通过模型预测，返回人类可读标签。</li>
    2008. 

  2008. </ol>
    2009. 

  2009. <h3 id="4">4 执行示例<a class="headerlink" href="#4" title="Permanent link">¶</a></h3>
    2010. 

  2010. <h4 id="41">4.1 运行脚本<a class="headerlink" href="#41" title="Permanent link">¶</a></h4>
    2011. 

  2011. <p>输出如下：</p>
    2012. 

  2012. <div class="highlight"><pre><span></span><code>使用设备: cpu
    2013. 

  2013. 初始化新 BERT 模型
    2014. 

  2014. 开始训练 BERT 模型...
    2015. 

  2015. [1500/1500 25:00, Epoch 3/3]
    2016. 

  2016. Epoch | Training Loss | Validation Loss | Accuracy
    2017. 

  2017. 1     | 0.3400       | 0.2100          | 0.9150
    2018. 

  2018. 2     | 0.1700       | 0.1600          | 0.9300
    2019. 

  2019. 3     | 0.1100       | 0.1450          | 0.9350
    2020. 

  2020. 模型保存至: bert_query_classifier
    2021. 

  2021. 分类报告:
    2022. 

  2022.               precision    recall  f1-score   support
    2023. 

  2023. 通用知识       0.94      0.92      0.93       500
    2024. 

  2024. 企业知识       0.92      0.94      0.93       500
    2025. 

  2025. accuracy                           0.93      1000
    2026. 

  2026. 混淆矩阵:
    2027. 

  2027. [[460  40]
    2028. 

  2028.  [ 30 470]]
    2029. 

  2029. 查询: 什么是神经网络？ -&gt; 分类: 通用知识
    2030. 

  2030. 查询: JAVA知识文档费用多少？ -&gt; 分类: 企业知识
    2031. 

  2031. 查询: 23+45等于多少？ -&gt; 分类: 通用知识
    2032. 

  2032. 查询: AI内部赋能有哪些支持人员？ -&gt; 分类: 企业知识
    2033. 

  2033. </code></pre></div>
    2034. 

  2034. <hr>
    2035. 

  2035. <h4 id="42">4.2 代码示例<a class="headerlink" href="#42" title="Permanent link">¶</a></h4>
    2036. 

  2036. <p>集成到 EnterpriseKB-RAG</p>
    2037. 

  2037. <div class="highlight"><pre><span></span><code><span class="c1"># core/rag_system.py</span>
    2038. 

  2038. <span class="k">class</span><span class="w"> </span><span class="nc">RAGSystem</span><span class="p">:</span>
    2039. 

  2039.     <span class="k">def</span><span class="w"> </span><span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
    2040. 

  2040.         <span class="bp">self</span><span class="o">.</span><span class="n">classifier</span> <span class="o">=</span> <span class="n">QueryClassifier</span><span class="p">(</span><span class="n">model_path</span><span class="o">=</span><span class="s2">"bert_query_classifier"</span><span class="p">)</span>
    2041. 

  2041.         <span class="bp">self</span><span class="o">.</span><span class="n">knowledge_base</span> <span class="o">=</span> <span class="n">KnowledgeBase</span><span class="p">()</span>
    2042. 

  2042.         <span class="bp">self</span><span class="o">.</span><span class="n">consulting_service</span> <span class="o">=</span> <span class="n">ConsultingService</span><span class="p">()</span>
    2043. 

  2043. 

  2044.     <span class="k">def</span><span class="w"> </span><span class="nf">route_query</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">query</span><span class="p">):</span>
    2045. 

  2045.         <span class="n">category</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">classifier</span><span class="o">.</span><span class="n">predict_category</span><span class="p">(</span><span class="n">query</span><span class="p">)</span>
    2046. 

  2046.         <span class="k">if</span> <span class="n">category</span> <span class="o">==</span> <span class="s2">"通用知识"</span><span class="p">:</span>
    2047. 

  2047.             <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">knowledge_base</span><span class="o">.</span><span class="n">search</span><span class="p">(</span><span class="n">query</span><span class="p">)</span>
    2048. 

  2048.         <span class="k">else</span><span class="p">:</span>
    2049. 

  2049.             <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">consulting_service</span><span class="o">.</span><span class="n">handle</span><span class="p">(</span><span class="n">query</span><span class="p">)</span>
    2050. 

  2050. </code></pre></div>
    2051. 

  2051. <hr>
    2052. 

  2052. <h2 id="_3">章节小结<a class="headerlink" href="#_3" title="Permanent link">¶</a></h2>
    2053. 

  2053. <p>本章节详细介绍了<code>prompts.py</code>和<code>query_classifier.py</code>的功能与实现：</p>
    2054. 

  2054. <ul>
    2055. 

  2055. <li>
    2056. 

  2056. <p><strong><code>prompts.py</code></strong>：通过<code>RAGPrompts</code>类管理多种Prompt模板，优化大语言模型的输入和输出，支持核心回答、HyDE、子查询和回溯策略。</p>
    2057. 

  2057. </li>
    2058. 

  2058. <li>
    2059. 

  2059. <p><strong><code>query_classifier.py</code></strong>：通过<code>QueryClassifier</code>类实现查询分类，区分通用知识和企业知识，决定系统的工作流程。</p>
    2060. 

  2060. </li>
    2061. 

  2061. </ul>
    2062. 

  2062. <p>学习者通过本章节掌握了如何通过Prompt管理和查询分类提升RAG系统的智能性和效率，为后续的检索策略选择和核心逻辑实现做好了准备。</p>
    2063. 

  2063. 

  2064. 

  2065. 

  2066. 

  2067. 

  2068. 

  2069. 

  2070. 

  2071. 

  2072. 

  2073. 

  2074. 

  2075.                 
    2076. 

  2076.               </article>
    2077. 

  2077.             </div>
    2078. 

  2078.           
    2079. 

  2079.           
    2080. 

  2080. <script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
    2081. 

  2081.         </div>
    2082. 

  2082.         
    2083. 

  2083.       </main>
    2084. 

  2084.       
    2085. 

  2085.         <footer class="md-footer">
    2086. 

  2086.   
    2087. 

  2087.   <div class="md-footer-meta md-typeset">
    2088. 

  2088.     <div class="md-footer-meta__inner md-grid">
    2089. 

  2089.       <div class="md-copyright">
    2090. 

  2090.   
    2091. 

  2091.     <div class="md-copyright__highlight">
    2092. 

  2092.       © EnterpriseKB-RAG 2026. Internal knowledge base demo site.
    2093. 

  2093.     </div>
    2094. 

  2094.   
    2095. 

  2095.   
    2096. 

  2096.     Made with
    2097. 

  2097.     <a href="https://squidfunk.github.io/mkdocs-material/" target="_blank" rel="noopener">
    2098. 

  2098.       Material for MkDocs
    2099. 

  2099.     </a>
    2100. 

  2100.   
    2101. 

  2101. </div>
    2102. 

  2102.       
    2103. 

  2103.     </div>
    2104. 

  2104.   </div>
    2105. 

  2105. </footer>
    2106. 

  2106.       
    2107. 

  2107.     </div>
    2108. 

  2108.     <div class="md-dialog" data-md-component="dialog">
    2109. 

  2109.       <div class="md-dialog__inner md-typeset"></div>
    2110. 

  2110.     </div>
    2111. 

  2111.     
    2112. 

  2112.     
    2113. 

  2113.     
    2114. 

  2114.       
    2115. 

  2115.       <script id="__config" type="application/json">{"base": "..", "features": [], "search": "../assets/javascripts/workers/search.973d3a69.min.js", "tags": null, "translations": {"clipboard.copied": "\u5df2\u590d\u5236", "clipboard.copy": "\u590d\u5236", "search.result.more.one": "\u5728\u8be5\u9875\u4e0a\u8fd8\u6709 1 \u4e2a\u7b26\u5408\u6761\u4ef6\u7684\u7ed3\u679c", "search.result.more.other": "\u5728\u8be5\u9875\u4e0a\u8fd8\u6709 # \u4e2a\u7b26\u5408\u6761\u4ef6\u7684\u7ed3\u679c", "search.result.none": "\u6ca1\u6709\u627e\u5230\u7b26\u5408\u6761\u4ef6\u7684\u7ed3\u679c", "search.result.one": "\u627e\u5230 1 \u4e2a\u7b26\u5408\u6761\u4ef6\u7684\u7ed3\u679c", "search.result.other": "# \u4e2a\u7b26\u5408\u6761\u4ef6\u7684\u7ed3\u679c", "search.result.placeholder": "\u952e\u5165\u4ee5\u5f00\u59cb\u641c\u7d22", "search.result.term.missing": "\u7f3a\u5c11", "select.version": "\u9009\u62e9\u5f53\u524d\u7248\u672c"}, "version": null}</script>
    2116. 

  2116.     
    2117. 

  2117.     
    2118. 

  2118.       <script src="../assets/javascripts/bundle.f55a23d4.min.js"></script>
    2119. 

  2119.       
    2120. 

  2120.         <script src="../javascripts/extra.js"></script>
    2121. 

  2121.       
    2122. 

  2122.         <script src="../javascripts/mathjax.js"></script>
    2123. 

  2123.       
    2124. 

  2124.         <!--        <script src="https://polyfill.io/v3/polyfill.min.js?features=es6"></script>-->
    2125. 

  2125.       
    2126. 

  2126.         <script src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>
    2127. 

  2127.       
    2128. 

  2128.     
    2129. 

  2129.   </body>
    2130. 

  2130. </html>
    2131.