{"id":28627,"date":"2025-03-14T00:33:17","date_gmt":"2025-03-13T16:33:17","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=28627"},"modified":"2025-03-14T00:33:17","modified_gmt":"2025-03-13T16:33:17","slug":"shiyong-ollamallamai","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/pt\/shiyong-ollamallamai\/","title":{"rendered":"\u4f7f\u7528 Ollama+LlamaIndex \u642d\u5efa\u672c\u5730 RAG \u5e94\u7528"},"content":{"rendered":"<h2>\u7b80\u4ecb<\/h2>\n<p>\u672c\u6587\u6863\u5c06\u8be6\u7ec6\u4ecb\u7ecd\u5982\u4f55\u4f7f\u7528 LlamaIndex \u6846\u67b6\u6765\u642d\u5efa\u672c\u5730 RAG\uff08Retrieval-Augmented Generation\uff09\u5e94\u7528\u3002\u901a\u8fc7\u96c6\u6210 LlamaIndex\uff0c\u53ef\u4ee5\u5728\u672c\u5730\u73af\u5883\u4e2d\u6784\u5efa\u4e00\u4e2a <a href=\"https:\/\/www.kdjingpai.com\/pt\/rag\/\">RAG<\/a> \u7cfb\u7edf\uff0c\u7ed3\u5408\u68c0\u7d22\u4e0e\u751f\u6210\u7684\u80fd\u529b\uff0c\u4ee5\u63d0\u9ad8\u4fe1\u606f\u68c0\u7d22\u7684\u6548\u7387\u548c\u751f\u6210\u5185\u5bb9\u7684\u76f8\u5173\u6027\u3002\u53ef\u4ee5\u81ea\u5b9a\u4e49\u672c\u5730\u77e5\u8bc6\u5e93\u8def\u5f84\uff0c\u901a\u8fc7 LlamaIndex \u6784\u5efa\u7d22\u5f15\uff0c\u7136\u540e\u5229\u7528\u7d22\u5f15\u8fdb\u884c\u4e0a\u4e0b\u6587\u5bf9\u8bdd\u3002<\/p>\n<blockquote><p>\u6ce8: \u672c\u6587\u6863\u5305\u542b\u6838\u5fc3\u4ee3\u7801\u7247\u6bb5\u548c\u8be6\u7ec6\u89e3\u91ca\u3002\u5b8c\u6574\u4ee3\u7801\u53ef\u89c1\u00a0<a href=\"https:\/\/github.com\/datawhalechina\/handy-ollama\/blob\/main\/notebook\/C7\/LlamaIndex_RAG\/%E4%BD%BF%E7%94%A8LlamaIndex%E6%90%AD%E5%BB%BA%E6%9C%AC%E5%9C%B0RAG%E5%BA%94%E7%94%A8.ipynb\">notebook<\/a>\u00a0\u3002<\/p><\/blockquote>\n<p>&nbsp;<\/p>\n<h2>1. \u6a21\u578b\u4e0b\u8f7d<\/h2>\n<p>\u672c\u4f8b\u4e2d\u4f7f\u7528\u7684\u662f\u00a0<code>llama3.1<\/code>\u00a0\u6a21\u578b\uff0c\u53ef\u4ee5\u6839\u636e\u81ea\u8eab\u7535\u8111\u914d\u7f6e\uff0c\u4f7f\u7528\u5408\u9002\u7684\u6a21\u578b\u3002<\/p>\n<pre><code><a href=\"https:\/\/www.kdjingpai.com\/pt\/ollama\/\">ollama<\/a> pull llama3.1\r\nollama pull nomic-embed-text\r\n<\/code><\/pre>\n<p>&nbsp;<\/p>\n<h2>2. \u5b89\u88c5\u4f9d\u8d56<\/h2>\n<pre><code>pip install llama-index-llms-ollama\r\npip install llama-index-embeddings-ollama\r\npip install -U llama-index-readers-file\r\n<\/code><\/pre>\n<p>&nbsp;<\/p>\n<h2>3. \u52a0\u8f7d\u6570\u636e<\/h2>\n<p>\u52a0\u8f7d\u5f53\u524d\u76ee\u5f55\u4e0b data \u6587\u4ef6\u5939\u4e2d\u6240\u6709\u7684\u6587\u6863\uff0c\u5e76\u52a0\u8f7d\u5230\u5185\u5b58\u4e2d\u3002<\/p>\n<pre><code>documents = SimpleDirectoryReader(\"data\").load_data()\r\n<\/code><\/pre>\n<p>&nbsp;<\/p>\n<h2>4. \u6784\u5efa\u7d22\u5f15<\/h2>\n<pre><code>Settings.embed_model = OllamaEmbedding(model_name=\"nomic-embed-text\")\r\nSettings.llm = Ollama(model=\"llama3.1\", request_timeout=360.0)\r\nindex = VectorStoreIndex.from_documents(\r\ndocuments,\r\n)\r\n<\/code><\/pre>\n<ul>\n<li><code>Settings.embed_model<\/code>\u00a0\uff1a \u5168\u5c40\u7684\u00a0<code>embed_model<\/code>\u00a0\u5c5e\u6027\u3002\u793a\u4f8b\u4ee3\u7801\u4e2d\u5c06\u521b\u5efa\u597d\u7684\u5d4c\u5165\u6a21\u578b\u8d4b\u503c\u7ed9\u5168\u5c40\u7684\u00a0<code>embed_model<\/code>\u00a0\u5c5e\u6027\uff1b<\/li>\n<li><code>Settings.llm<\/code>\u00a0\uff1a \u5168\u5c40\u7684\u00a0<code>llm<\/code>\u00a0\u5c5e\u6027\u3002\u793a\u4f8b\u4ee3\u7801\u4e2d\u5c06\u521b\u5efa\u597d\u7684\u8bed\u8a00\u6a21\u578b\u8d4b\u503c\u7ed9\u5168\u5c40\u7684\u00a0<code>llm<\/code>\u00a0\u5c5e\u6027\uff1b<\/li>\n<li><code>VectorStoreIndex.from_documents<\/code>\uff1a\u4f7f\u7528\u4e4b\u524d\u52a0\u8f7d\u7684\u6587\u6863\u6784\u5efa\u7d22\u5f15\uff0c\u5e76\u8f6c\u6362\u6210\u5411\u91cf\uff0c\u4fbf\u4e8e\u5feb\u901f\u68c0\u7d22\u3002<\/li>\n<\/ul>\n<p>\u901a\u8fc7\u00a0<code>Settings<\/code>\u00a0\u5168\u5c40\u5c5e\u6027\u7684\u8bbe\u7f6e\uff0c\u5728\u540e\u9762\u7684\u7d22\u5f15\u6784\u5efa\u4ee5\u53ca\u67e5\u8be2\u7684\u8fc7\u7a0b\u4e2d\u5c31\u4f1a\u9ed8\u8ba4\u4f7f\u7528\u76f8\u5e94\u7684\u6a21\u578b\u3002<\/p>\n<p>&nbsp;<\/p>\n<h2>5. \u67e5\u8be2\u6570\u636e<\/h2>\n<pre><code>query_engine = index.as_query_engine()\r\nresponse = query_engine.query(\"Datawhale\u662f\u4ec0\u4e48?\")\r\nprint(response)\r\n<\/code><\/pre>\n<ul>\n<li><code>index.as_query_engine()<\/code>\uff1a\u6839\u636e\u4e4b\u524d\u6784\u5efa\u597d\u7684\u7d22\u5f15\uff0c\u521b\u5efa\u67e5\u8be2\u5f15\u64ce\u3002\u8be5\u67e5\u8be2\u5f15\u64ce\u53ef\u4ee5\u63a5\u6536\u67e5\u8be2\uff0c\u8fd4\u56de\u68c0\u7d22\u540e\u7684\u54cd\u5e94\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>6. \u68c0\u7d22\u4e0a\u4e0b\u6587\u8fdb\u884c\u5bf9\u8bdd<\/h2>\n<p>\u7531\u4e8e\u68c0\u7d22\u5230\u7684\u4e0a\u4e0b\u6587\u53ef\u80fd\u4f1a\u5360\u7528\u5927\u91cf\u53ef\u7528\u7684 LLM \u4e0a\u4e0b\u6587\uff0c\u56e0\u6b64\u9700\u8981\u4e3a\u804a\u5929\u5386\u53f2\u8bb0\u5f55\u914d\u7f6e\u8f83\u5c0f <a href=\"https:\/\/www.kdjingpai.com\/pt\/tokenization\/\">token<\/a> \u9650\u5236\uff01<\/p>\n<pre><code># \u68c0\u7d22\u4e0a\u4e0b\u6587\u8fdb\u884c\u5bf9\u8bdd\r\nfrom llama_index.core.memory import ChatMemoryBuffer\r\nmemory = ChatMemoryBuffer.from_defaults(token_limit=1500)\r\nchat_engine = index.as_chat_engine(\r\nchat_mode=\"context\",\r\nmemory=memory,\r\nsystem_prompt=(\r\n\"You are a chatbot, able to have normal interactions.\"\r\n),\r\n)\r\nresponse = chat_engine.chat(\"Datawhale\u662f\u4ec0\u4e48\uff1f\")\r\nprint(response)\r\n<\/code><\/pre>\n<p><code>chat_mode<\/code>\u00a0\u53ef\u4ee5\u6839\u636e\u4f7f\u7528\u573a\u666f\uff0c\u9009\u62e9\u5408\u9002\u7684\u6a21\u5f0f\uff0c\u652f\u6301\u7684\u6a21\u5f0f\u5982\u4e0b\uff1a<\/p>\n<ul>\n<li><code>best<\/code>\uff08\u9ed8\u8ba4\uff09\uff1a\u4f7f\u7528\u5e26\u6709\u67e5\u8be2\u5f15\u64ce\u5de5\u5177\u7684\u4ee3\u7406\uff08react \u6216 openai\uff09\uff1b<\/li>\n<li><code>context<\/code>\uff1a\u4f7f\u7528\u68c0\u7d22\u5668\u83b7\u53d6\u4e0a\u4e0b\u6587\uff1b<\/li>\n<li><code>condense_question<\/code>\uff1a\u5c06\u95ee\u9898\u8fdb\u884c\u6d53\u7f29\uff1b<\/li>\n<li><code>condense_plus_context<\/code>\uff1a\u5c06\u95ee\u9898\u8fdb\u884c\u6d53\u7f29\u5e76\u4f7f\u7528\u68c0\u7d22\u5668\u83b7\u53d6\u4e0a\u4e0b\u6587\uff1b<\/li>\n<li><code>simple<\/code>\uff1a\u76f4\u63a5\u4f7f\u7528 LLM \u7684\u7b80\u5355\u804a\u5929\u5f15\u64ce\uff1b<\/li>\n<li><code><a href=\"https:\/\/www.kdjingpai.com\/pt\/react\/\">react<\/a><\/code>\uff1a\u4f7f\u7528\u5e26\u6709\u67e5\u8be2\u5f15\u64ce\u5de5\u5177\u7684 react \u4ee3\u7406\uff1b<\/li>\n<li><code>openai<\/code>\uff1a\u4f7f\u7528\u5e26\u6709\u67e5\u8be2\u5f15\u64ce\u5de5\u5177\u7684 openai \u4ee3\u7406\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>7. \u5411\u91cf\u7d22\u5f15\u7684\u5b58\u50a8\u548c\u52a0\u8f7d<\/h2>\n<ul>\n<li><code>storage_context.persist<\/code>\u00a0\u5b58\u50a8\u5411\u91cf\u7d22\u5f15\u3002<\/li>\n<li><code>load_index_from_storage<\/code>\u00a0\u52a0\u8f7d\u5411\u91cf\u7d22\u5f15\u3002<\/li>\n<\/ul>\n<pre><code># \u5b58\u50a8\u5411\u91cf\u7d22\u5f15\r\npersist_dir = 'data\/'\r\nindex.storage_context.persist(persist_dir=persist_dir)\r\n# \u52a0\u8f7d\u5411\u91cf\u7d22\u5f15\r\nfrom llama_index.core import StorageContext, load_index_from_storage\r\nstorage_context = StorageContext.from_defaults(persist_dir=persist_dir)\r\nindex= load_index_from_storage(storage_context)\r\n<\/code><\/pre>\n<p>&nbsp;<\/p>\n<h2>8. streamlit \u5e94\u7528<\/h2>\n<p>\u672c\u4f8b\u5185\u5bb9\u8fd8\u5b9e\u73b0\u4e86\u4e00\u4e2a streamlit \u5e94\u7528\uff0c\u5177\u4f53\u53ef\u4ee5\u67e5\u770b\u00a0<a href=\"https:\/\/github.com\/AXYZdong\/handy-ollama\/blob\/main\/notebook\/C7\/LlamaIndex_RAG\/app.py\">app.py<\/a><\/p>\n<p>\u6240\u9700\u4f9d\u8d56\u5982\u4e0b\uff1a<\/p>\n<pre><code>llama_index==0.10.62\r\nstreamlit==1.36.0\r\n<\/code><\/pre>\n<p>Note<\/p>\n<p>\u5728\u00a0<code>app.py<\/code>\u00a0\u4e2d\uff0c\u4e3a\u4e86\u5728\u8fde\u7eed\u5bf9\u8bdd\u4e2d\u4e0d\u91cd\u65b0\u52a0\u8f7d\u6a21\u578b\uff0c\u53ef\u4ee5\u914d\u7f6e\u73af\u5883\u53d8\u91cf\u00a0<code>OLLAMA_NUM_PARALLEL<\/code>\u00a0\u548c\u00a0<code>OLLAMA_MAX_LOADED_MODELS<\/code>\uff0c\u56e0\u4e3a\u652f\u6301\u591a\u4e2a\u6a21\u578b\u52a0\u8f7d\uff0c\u6240\u4ee5\u81f3\u5c11\u9700\u8981\u989d\u5916\u7684 8G \u5185\u5b58\u3002<\/p>\n<p><code>OLLAMA_NUM_PARALLEL<\/code>\uff1a\u540c\u65f6\u5904\u7406\u5355\u4e2a\u6a21\u578b\u7684\u591a\u4e2a\u8bf7\u6c42\u3002<\/p>\n<p><code>OLLAMA_MAX_LOADED_MODELS<\/code>\uff1a\u540c\u65f6\u52a0\u8f7d\u591a\u4e2a\u6a21\u578b\u3002<\/p>\n<p>\u793a\u4f8b\u5c55\u793a<\/p>\n<ol>\n<li>\u5355\u6587\u672c\u95ee\u7b54<\/li>\n<\/ol>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter size-full wp-image-28629\" title=\"\u4f7f\u7528 Ollama+LlamaIndex \u642d\u5efa\u672c\u5730 RAG \u5e94\u7528-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/5ef4994d70cd1db.png\" alt=\"\u4f7f\u7528 Ollama+LlamaIndex \u642d\u5efa\u672c\u5730 RAG \u5e94\u7528-1\" width=\"1872\" height=\"956\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/5ef4994d70cd1db.png 1872w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/5ef4994d70cd1db-768x392.png 768w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/5ef4994d70cd1db-1536x784.png 1536w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/5ef4994d70cd1db-18x9.png 18w\" sizes=\"auto, (max-width: 1872px) 100vw, 1872px\" \/><\/p>\n<ol start=\"2\">\n<li>\u591a\u6587\u672c\u95ee\u7b54<\/li>\n<\/ol>\n<p><a href=\"https:\/\/github.com\/datawhalechina\/handy-ollama\/blob\/main\/docs\/images\/C7-4-2.png\"> <img loading=\"lazy\" decoding=\"async\" class=\"aligncenter size-full wp-image-28628\" title=\"\u4f7f\u7528 Ollama+LlamaIndex \u642d\u5efa\u672c\u5730 RAG \u5e94\u7528-2\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/cc59a4983c2f2d8.png\" alt=\"\u4f7f\u7528 Ollama+LlamaIndex \u642d\u5efa\u672c\u5730 RAG \u5e94\u7528-2\" width=\"1872\" height=\"956\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/cc59a4983c2f2d8.png 1872w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/cc59a4983c2f2d8-768x392.png 768w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/cc59a4983c2f2d8-1536x784.png 1536w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/03\/cc59a4983c2f2d8-18x9.png 18w\" sizes=\"auto, (max-width: 1872px) 100vw, 1872px\" \/> <\/a><\/p>\n<p>&nbsp;<\/p>\n<p>\u53c2\u8003\u6587\u732e\uff1a<a href=\"https:\/\/docs.llamaindex.ai\/en\/stable\/getting_started\/starter_example_local\/?h=ollama#load-data-and-build-an-index\">LlamaIndex \u6587\u6863<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u7b80\u4ecb \u672c\u6587\u6863\u5c06\u8be6\u7ec6\u4ecb\u7ecd\u5982\u4f55\u4f7f\u7528 LlamaIndex \u6846\u67b6\u6765\u642d\u5efa\u672c\u5730 RAG\uff08Retrieval-Augmented Generation\uff09\u5e94\u7528\u3002\u901a\u8fc7\u96c6\u6210 LlamaIndex\uff0c\u53ef\u4ee5\u5728\u672c\u5730\u73af\u5883\u4e2d\u6784\u5efa\u4e00\u4e2a RAG \u7cfb\u7edf\uff0c\u7ed3\u5408\u68c0\u7d22\u4e0e\u751f\u6210\u7684\u80fd\u529b\uff0c&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[182],"tags":[],"class_list":["post-28627","post","type-post","status-publish","format-standard","hentry","category-shicao"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts\/28627","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/comments?post=28627"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts\/28627\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/media?parent=28627"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/categories?post=28627"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/tags?post=28627"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}