Dynamiq\u662f\u4e00\u4e2a\u5f00\u6e90\u7684AI\u7f16\u6392\u6846\u67b6\uff0c\u4e13\u4e3a\u4ee3\u7406AI\u548c\u5927\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u5e94\u7528\u800c\u8bbe\u8ba1\u3002\u5b83\u65e8\u5728\u7b80\u5316AI\u9a71\u52a8\u5e94\u7528\u7a0b\u5e8f\u7684\u5f00\u53d1\uff0c\u7279\u522b\u662f\u5728\u68c0\u7d22\u589e\u5f3a\u751f\u6210\uff08RAG\uff09\u548cLLM\u4ee3\u7406\u7684\u7f16\u6392\u65b9\u9762\u3002Dynamiq\u63d0\u4f9b\u4e86\u4e30\u5bcc\u7684\u529f\u80fd\u6a21\u5757\u548c\u8be6\u7ec6\u7684\u6587\u6863\uff0c\u5e2e\u52a9\u5f00\u53d1\u8005\u5feb\u901f\u4e0a\u624b\u5e76\u9ad8\u6548\u5730\u6784\u5efa\u590d\u6742\u7684AI\u5e94\u7528\u3002<\/p>\n
<\/p>\n
Dynamiq\u662f\u4e00\u4e2a\u521b\u65b0\u7684 AI \u6846\u67b6\uff0c\u5b83\u901a\u8fc7\u5c06 LLMs \u7684\u63a8\u7406\u80fd\u529b\uff08\u5927\u8111\uff09\u4e0e\u6267\u884c\u5177\u4f53\u64cd\u4f5c\u7684\u529f\u80fd\uff08\u53cc\u624b\uff09\u76f8\u7ed3\u5408\uff0c\u4f7f AI \u80fd\u591f\u50cf\u4eba\u7c7b\u4e00\u6837\u7406\u89e3\u4efb\u52a1\u3001\u8fdb\u884c\u63a8\u7406\u5e76\u91c7\u53d6\u5b9e\u9645\u884c\u52a8\u6765\u89e3\u51b3\u73b0\u5b9e\u95ee\u9898<\/p>\n
\/\/ ReAct \u7684\u5b9a\u4e49\uff1a<\/strong><\/p>\n \/\/ ReAct agents \u7684\u5de5\u4f5c\u539f\u7406\uff1a<\/strong> \/\/ \u6846\u67b6\u7ec4\u6210\u90e8\u5206\uff1a<\/strong><\/p>\n \/\/ \u5b9e\u9645\u5e94\u7528\u793a\u4f8b\uff1a<\/strong> \/\/ Akshay \u5206\u4eab\u7684 Dynamiq \u6846\u67b6\uff1a<\/strong> \/\/ \u4e3b\u8981\u7279\u70b9\uff1a \u4e13\u4e1a\u9886\u57df\uff1a<\/p>\n \u5b9a\u4f4d\uff1a<\/p>\n <\/p>\n <\/p>\n \u6216\u8005\u4ece\u6e90\u4ee3\u7801\u6784\u5efa\uff1a<\/p>\n \u4ee5\u4e0b\u662f\u4e00\u4e2a\u7b80\u5355\u7684LLM\u5de5\u4f5c\u6d41\u793a\u4f8b\uff1a<\/p>\n \u4ee5\u4e0b\u662f\u4e00\u4e2a\u652f\u6301\u590d\u6742\u7f16\u7801\u4efb\u52a1\u7684ReAct\u4ee3\u7406\u793a\u4f8b\uff1a<\/p>\n \u4ee5\u4e0b\u662f\u4e00\u4e2a\u591a\u4ee3\u7406\u534f\u540c\u5de5\u4f5c\u7684\u793a\u4f8b\uff1a<\/p>\n Dynamiq\u652f\u6301\u68c0\u7d22\u589e\u5f3a\u751f\u6210\uff08RAG\uff09\uff0c\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u6b65\u9aa4\u5b9e\u73b0\uff1a<\/p>\n \u4ee5\u4e0b\u662f\u4e00\u4e2a\u7b80\u5355\u7684RAG\u5de5\u4f5c\u6d41\u793a\u4f8b\uff1a<\/p>\n \u4ee5\u4e0b\u662f\u4e00\u4e2a\u5e26\u8bb0\u5fc6\u529f\u80fd\u7684\u7b80\u5355\u804a\u5929\u673a\u5668\u4eba\u793a\u4f8b\uff1a<\/p>\n Dynamiq\u662f\u4e00\u4e2a\u5f00\u6e90\u7684AI\u7f16\u6392\u6846\u67b6\uff0c\u4e13\u4e3a\u4ee3\u7406AI\u548c\u5927\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u5e94\u7528\u800c\u8bbe\u8ba1\u3002\u5b83\u65e8\u5728\u7b80\u5316AI\u9a71\u52a8\u5e94\u7528\u7a0b\u5e8f\u7684\u5f00\u53d1\uff0c\u7279\u522b\u662f\u5728\u68c0\u7d22\u589e\u5f3a\u751f\u6210\uff08RAG\uff09\u548cLLM\u4ee3\u7406\u7684\u7f16\u6392\u65b9\u9762\u3002Dynamiq\u63d0\u4f9b\u4e86\u4e30\u5bcc\u7684\u529f\u80fd\u6a21\u5757\u548c\u8be6\u7ec6\u7684\u6587\u6863\uff0c\u5e2e\u52a9\u5f00\u53d1\u8005\u5feb\u901f\u4e0a\u624b…<\/p>\n","protected":false},"author":1,"featured_media":61300,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[230,201],"class_list":["post-14078","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","tag-aikaiyuanxiangmu","tag-aizhinengti"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/14078","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/comments?post=14078"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/posts\/14078\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media\/61300"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/media?parent=14078"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/categories?post=14078"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/de\/wp-json\/wp\/v2\/tags?post=14078"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\n
\n\u5b83\u6574\u5408\u4e86\u4e24\u4e2a\u5173\u952e\u7ec4\u4ef6\uff1a<\/p>\n\n
\n
\n\u4f5c\u8005\u7528\u4e00\u4e2a\u5224\u65ad\u662f\u5426\u9700\u8981\u5e26\u4f1e\u7684\u573a\u666f\u6765\u8bf4\u660e ReAct agent \u7684\u5de5\u4f5c\u6d41\u7a0b\uff1a<\/p>\n\n
\nDynamiq \u662f\u4e00\u4e2a\u9762\u5411\u65b0\u4e00\u4ee3 AI \u5f00\u53d1\u7684\u7efc\u5408\u6027\u6846\u67b6\uff0c\u4e13\u6ce8\u4e8e\u7b80\u5316 AI \u5e94\u7528\u7684\u5f00\u53d1\u6d41\u7a0b\uff0c\u5176\u4e3b\u8981\u7279\u70b9\u662f\u80fd\u591f\u7f16\u6392\u548c\u7ba1\u7406\u57fa\u4e8e RAG<\/a> \u548c LLM \u7684 Agents \u7cfb\u7edf\u3002<\/p>\n
\n\u5168\u80fd\u6027\uff1a\u4f5c\u4e3a\u4e00\u7ad9\u5f0f(“all-in-one”)\u6846\u67b6\uff0c\u6574\u5408\u4e86\u5f00\u53d1 AI \u5e94\u7528\u6240\u9700\u7684\u5404\u79cd\u5de5\u5177\u548c\u529f\u80fd<\/p>\n\n
\n
\u529f\u80fd\u5217\u8868<\/h2>\n
\n
\u4f7f\u7528\u5e2e\u52a9<\/h2>\n
\u5b89\u88c5\u548c\u914d\u7f6e<\/h3>\n
\n
pip install dynamiq\r\n<\/code><\/pre>\ngit clone https:\/\/github.com\/dynamiq-ai\/dynamiq.git\r\ncd dynamiq\r\npoetry install\r\n<\/code><\/pre>\n<\/li>\n<\/ol>\n\u4f7f\u7528\u793a\u4f8b<\/h3>\n
\u7b80\u5355LLM\u6d41\u7a0b<\/h4>\n
from dynamiq.nodes.llms.openai import OpenAI\r\nfrom dynamiq.connections import OpenAI as OpenAIConnection\r\nfrom dynamiq import Workflow<\/a>\r\nfrom dynamiq.prompts import Prompt, Message\r\n# \u5b9a\u4e49\u7ffb\u8bd1\u63d0\u793a\u6a21\u677f\r\nprompt_template = \"\"\"\r\nTranslate the following text into English: {{ text }}\r\n\"\"\"\r\nprompt = Prompt(messages=[Message(content=prompt_template, role=\"user\")])\r\n# \u8bbe\u7f6eLLM\u8282\u70b9\r\nllm = OpenAI(\r\nid=\"openai\",\r\nconnection=OpenAIConnection(api_key=\"$OPENAI_API_KEY\"),\r\nmodel=\"gpt-4o\",\r\ntemperature=0.3,\r\nmax_tokens=1000,\r\nprompt=prompt\r\n)\r\n# \u521b\u5efa\u5de5\u4f5c\u6d41\u5bf9\u8c61\r\nworkflow = Workflow()\r\nworkflow.flow.add_nodes(llm)\r\n# \u8fd0\u884c\u5de5\u4f5c\u6d41\r\nresult = workflow.run(input_data={\"text\": \"Hola Mundo!\"})\r\nprint(result.output)\r\n<\/code><\/pre>\nReAct\u4ee3\u7406<\/h4>\n
from dynamiq.nodes.llms.openai import OpenAI\r\nfrom dynamiq.connections import OpenAI as OpenAIConnection, E2B<\/a> as E2BConnection\r\nfrom dynamiq.nodes.agents.react import ReActAgent\r\nfrom dynamiq.nodes.tools.e2b_sandbox import E2BInterpreterTool\r\n# \u521d\u59cb\u5316E2B\u5de5\u5177\r\ne2b_tool = E2BInterpreterTool(connection=E2BConnection(api_key=\"$API_KEY\"))\r\n# \u8bbe\u7f6eLLM\r\nllm = OpenAI(\r\nid=\"openai\",\r\nconnection=OpenAIConnection(api_key=\"$API_KEY\"),\r\nmodel=\"gpt-4o\",\r\ntemperature=0.3,\r\nmax_tokens=1000,\r\n)\r\n# \u521b\u5efaReAct\u4ee3\u7406\r\nagent = ReActAgent(\r\nname=\"react-agent\",\r\nllm=llm,\r\ntools=[e2b_tool],\r\nrole=\"Senior Data Scientist\",\r\nmax_loops=10,\r\n)\r\n# \u8fd0\u884c\u4ee3\u7406\r\nresult = agent.run(input_data={\"input\": \"Add the first 10 numbers and tell if the result is prime.\"})\r\nprint(result.output.get(\"content\"))\r\n<\/code><\/pre>\n\u591a\u4ee3\u7406\u7f16\u6392<\/h3>\n
from dynamiq.connections import OpenAI as OpenAIConnection, ScaleSerp as ScaleSerpConnection, E2B as E2BConnection\r\nfrom dynamiq.nodes.llms import OpenAI\r\nfrom dynamiq.nodes.agents.orchestrators.adaptive import AdaptiveOrchestrator\r\nfrom dynamiq.nodes.agents.orchestrators.adaptive_manager import AdaptiveAgentManager\r\nfrom dynamiq.nodes.agents.react import ReActAgent\r\nfrom dynamiq.nodes.agents.reflection import ReflectionAgent\r\nfrom dynamiq.nodes.tools.e2b_sandbox import E2BInterpreterTool\r\nfrom dynamiq.nodes.tools.scale_serp import ScaleSerpTool\r\n# \u521d\u59cb\u5316\u5de5\u5177\r\npython_tool = E2BInterpreterTool(connection=E2BConnection(api_key=\"$E2B_API_KEY\"))\r\nsearch_tool = ScaleSerpTool(connection=ScaleSerpConnection(api_key=\"$SCALESERP_API_KEY\"))\r\n# \u521d\u59cb\u5316LLM\r\nllm = OpenAI(connection=OpenAIConnection(api_key=\"$OPENAI_API_KEY\"), model=\"gpt-4o\", temperature=0.1)\r\n# \u5b9a\u4e49\u4ee3\u7406\r\ncoding_agent = ReActAgent(\r\nname=\"coding-agent\",\r\nllm=llm,\r\ntools=[python_tool],\r\nrole=\"Expert agent with coding skills. Goal is to provide the solution to the input task using Python software engineering skills.\",\r\nmax_loops=15,\r\n)\r\nplanner_agent = ReflectionAgent(\r\nname=\"planner-agent\",\r\nllm=llm,\r\nrole=\"Expert agent with planning skills. Goal is to analyze complex requests and provide a detailed action plan.\",\r\n)\r\nsearch_agent = ReActAgent(\r\nname=\"search-agent\",\r\nllm=llm,\r\ntools=[search_tool],\r\nrole=\"Expert agent with web search skills. Goal is to provide the solution to the input task using web search and summarization skills.\",\r\nmax_loops=10,\r\n)\r\n# \u521d\u59cb\u5316\u81ea\u9002\u5e94\u4ee3\u7406\u7ba1\u7406\u5668\r\nagent_manager = AdaptiveAgentManager(llm=llm)\r\n# \u521b\u5efa\u7f16\u6392\u5668\r\norchestrator = AdaptiveOrchestrator(\r\nname=\"adaptive-orchestrator\",\r\nagents=[coding_agent, planner_agent, search_agent],\r\nmanager=agent_manager,\r\n)\r\n# \u5b9a\u4e49\u8f93\u5165\u4efb\u52a1\r\ninput_task = (\r\n\"Use coding skills to gather data about Nvidia and Intel stock prices for the last 10 years, \"\r\n\"calculate the average per year for each company, and create a table. Then craft a report \"\r\n\"and add a conclusion: what would have been better if I had invested $100 ten years ago?\"\r\n)\r\n# \u8fd0\u884c\u7f16\u6392\u5668\r\nresult = orchestrator.run(input_data={\"input\": input_task})\r\nprint(result.output.get(\"content\"))\r\n<\/code><\/pre>\nRAG\u6587\u6863\u7d22\u5f15\u548c\u68c0\u7d22<\/h3>\n
\n
from io import BytesIO\r\nfrom dynamiq import Workflow\r\nfrom dynamiq.connections import OpenAI as OpenAIConnection, Pinecone as PineconeConnection\r\nfrom dynamiq.nodes.converters import PyPDFConverter\r\nfrom dynamiq.nodes.splitters.document import DocumentSplitter\r\nfrom dynamiq.nodes.embedders import OpenAIDocumentEmbedder\r\nfrom dynamiq.nodes.writers import PineconeDocumentWriter\r\n# \u521d\u59cb\u5316\u5de5\u4f5c\u6d41\r\nrag_wf = Workflow()\r\n# PDF\u6587\u6863\u8f6c\u6362\u5668\r\nconverter = PyPDFConverter(document_creation_mode=\"one-doc-per-page\")\r\nrag_wf.flow.add_nodes(converter)\r\n# \u6587\u6863\u62c6\u5206\u5668\r\ndocument_splitter = (\r\nDocumentSplitter(split_by=\"sentence\", split_length=10, split_overlap=1)\r\n.inputs(documents=converter.outputs.documents)\r\n.depends_on(converter)\r\n)\r\nrag_wf.flow.add_nodes(document_splitter)\r\n# OpenAI\u5411\u91cf\u5d4c\u5165\r\nembedder = (\r\nOpenAIDocumentEmbedder(connection=OpenAIConnection(api_key=\"$OPENAI_API_KEY\"), model=\"text-embedding-3-small\")\r\n.inputs(documents=document_splitter.outputs.documents)\r\n.depends_on(document_splitter)\r\n)\r\nrag_wf.flow.add_nodes(embedder)\r\n# Pinecone\u5411\u91cf\u5b58\u50a8\r\nvector_store = (\r\nPineconeDocumentWriter(connection=PineconeConnection(api_key=\"$PINECONE_API_KEY\"), index_name=\"default\", dimension=1536)\r\n.inputs(documents=embedder.outputs.documents)\r\n.depends_on(embedder)\r\n)\r\nrag_wf.flow.add_nodes(vector_store)\r\n# \u51c6\u5907\u8f93\u5165PDF\u6587\u4ef6\r\nfile_paths = [\"example.pdf\"]\r\ninput_data = {\r\n\"files\": [BytesIO(open(path, \"rb\").read()) for path in file_paths],\r\n\"metadata\": [{\"filename\": path} for path in file_paths],\r\n}\r\n# \u8fd0\u884cRAG\u7d22\u5f15\u6d41\u7a0b\r\nrag_wf.run(input_data=input_data)\r\n<\/code><\/pre>\n\u5e26\u8bb0\u5fc6\u7684\u804a\u5929\u673a\u5668\u4eba<\/h3>\n
from dynamiq.connections import OpenAI as OpenAIConnection\r\nfrom dynamiq.memory import Memory\r\nfrom dynamiq.memory.backend.in_memory import InMemory\r\nfrom dynamiq.nodes.agents.simple import SimpleAgent\r\nfrom dynamiq.nodes.llms import OpenAI\r\n\r\nAGENT_ROLE = \"helpful assistant, goal is to provide useful information and answer questions\"\r\nllm = OpenAI(\r\nconnection=OpenAIConnection(api_key=\"$OPENAI_API_KEY\"),\r\nmodel=\"gpt-4o\",\r\ntemperature=0.1,\r\n)\r\n\r\nmemory = Memory(backend=InMemory())\r\nagent = SimpleAgent(\r\nname=\"Agent\",\r\nllm=llm,\r\nrole=AGENT_ROLE,\r\nid=\"agent\",\r\nmemory=memory,\r\n)\r\n\r\n\r\ndef main():\r\nprint(\"Welcome to the AI Chat! (Type 'exit' to end)\")\r\nwhile True:\r\nuser_input = input(\"You: \")\r\nif user_input.lower() == \"exit\":\r\nbreak\r\n\r\nresponse = agent.run({\"input\": user_input})\r\nresponse_content = response.output.get(\"content\")\r\nprint(f\"AI: {response_content}\")\r\n\r\n\r\nif __name__ == \"__main__\":\r\nmain()<\/pre>\n","protected":false},"excerpt":{"rendered":"