SmolAgents \u662f Hugging Face \u7684\u4e00\u4e2a\u5c16\u7aef\u5e93\uff0c\u5141\u8bb8\u5f00\u53d1\u4eba\u5458\u521b\u5efa\u80fd\u591f\u89e3\u51b3\u590d\u6742\u4efb\u52a1\u7684\u667a\u80fd\u3001\u7279\u5b9a\u9886\u57df\u7684\u4ee3\u7406\u3002\u672c\u535a\u5ba2\u5c06\u6df1\u5165\u4ecb\u7ecd SmolAgents\uff0c\u5305\u62ec\u5de5\u5177\u3001\u4ee3\u7801\u4ee3\u7406\u3001\u5b89\u5168\u4ee3\u7801\u6267\u884c\u548c\u5b9e\u9645\u5b9e\u73b0\u3002\u6211\u4eec\u8fd8\u5c06\u7ed3\u5408\u4ee3\u7801\u793a\u4f8b\u4f7f\u6982\u5ff5\u66f4\u6e05\u6670\u3002<\/p>\n
\u5728 AI \u9886\u57df\uff0c\u4ee3\u7406\u662f LLM \u8f93\u51fa\u63a7\u5236\u5de5\u4f5c\u6d41\u7684\u7a0b\u5e8f\u3002LLM \u5bf9\u7cfb\u7edf\u64cd\u4f5c\u7684\u5f71\u54cd\u7a0b\u5ea6\u51b3\u5b9a\u4e86\u5176\u4ee3\u7406\u7ea7\u522b\u3002\u8fd9\u79cd\u4ee3\u7406\u5b58\u5728\u4e8e\u4ee5\u4e0b\u8303\u56f4\u5185\uff1a<\/p>\n
\u5f53\u4efb\u52a1\u9700\u8981\u57fa\u4e8e\u52a8\u6001\u8f93\u5165\u8fdb\u884c\u8c03\u6574\u7684\u7075\u6d3b\u5de5\u4f5c\u6d41\u65f6\uff0c\u5b9e\u65bd\u4ee3\u7406\u662f\u6709\u76ca\u7684\u3002\u4f46\u662f\uff0c\u5bf9\u4e8e\u5177\u6709\u53ef\u9884\u6d4b\u548c\u5b9a\u4e49\u660e\u786e\u7684\u6d41\u7a0b\u7684\u4efb\u52a1\uff0c\u4f20\u7edf\u7684\u786e\u5b9a\u6027\u7f16\u7a0b\u53ef\u80fd\u5c31\u8db3\u591f\u4e86\u3002<\/p>\n
!pip install -q smolagents\n!pip install huggingface_hub<\/code><\/pre>\n2\u3001\u6784\u5efa\u6709\u6548\u7684\u4ee3\u7406<\/h2>\n
\u521b\u5efa\u5f3a\u5927\u7684\u4ee3\u7406\u6d89\u53ca\u51e0\u4e2a\u5173\u952e\u8003\u8651\u56e0\u7d20\uff1a<\/p>\n
\n- \u5b9a\u4e49\u660e\u786e\u7684\u76ee\u6807\uff1a\u4e3a\u4ee3\u7406\u5efa\u7acb\u7279\u5b9a\u7684\u4efb\u52a1\u6216\u76ee\u6807\uff0c\u4ee5\u786e\u4fdd\u4e13\u6ce8\u7684\u8868\u73b0\u3002<\/li>\n
- \u7ed3\u5408\u76f8\u5173\u5de5\u5177\uff1a\u4e3a\u4ee3\u7406\u914d\u5907\u4e0e\u5176\u76ee\u6807\u4e00\u81f4\u7684\u5de5\u5177\uff0c\u589e\u5f3a\u5176\u6267\u884c\u6307\u5b9a\u4efb\u52a1\u7684\u80fd\u529b\u3002<\/li>\n
- \u5b9e\u65bd\u8bb0\u5fc6\u673a\u5236\uff1a\u4f7f\u4ee3\u7406\u80fd\u591f\u4fdd\u7559\u4ee5\u524d\u4ea4\u4e92\u4e2d\u7684\u4e0a\u4e0b\u6587\uff0c\u4ece\u800c\u505a\u51fa\u66f4\u8fde\u8d2f\u548c\u660e\u667a\u7684\u54cd\u5e94\u3002<\/li>\n
- \u786e\u4fdd\u5b89\u5168\u7684\u4ee3\u7801\u6267\u884c\uff1a\u4fdd\u62a4\u6267\u884c\u73af\u5883\u4ee5\u9632\u6b62\u672a\u7ecf\u6388\u6743\u7684\u64cd\u4f5c\u6216\u5b89\u5168\u6f0f\u6d1e\u3002<\/li>\n<\/ul>\n
3\u3001\u5728 smolagents \u4e2d\u4f7f\u7528\u5de5\u5177<\/h2>\n
\u5de5\u5177\u5bf9\u4e8e\u6269\u5c55\u4ee3\u7406\u7684\u529f\u80fd\u81f3\u5173\u91cd\u8981\u3002\u5728 smolagents \u4e2d\uff0c\u5de5\u5177\u672c\u8d28\u4e0a\u662f LLM \u53ef\u4ee5\u5728\u4ee3\u7406\u7cfb\u7edf\u4e2d\u4f7f\u7528\u7684\u529f\u80fd\u3002\u4e3a\u4e86\u5b9e\u73b0\u8fd9\u4e00\u70b9\uff0c\u5de5\u5177\u88ab\u5c01\u88c5\u5728\u63d0\u4f9b\u5143\u6570\u636e\u7684\u7c7b\u4e2d\uff0c\u5e2e\u52a9 LLM \u4e86\u89e3\u5b83\u4eec\u7684\u7528\u6cd5\u3002\u4f8b\u5982\uff1a<\/p>\n
from smolagents import Tool\n\nclass HFModelDownloadsTool(Tool):\n name = \"model_download_counter\"\n description = \"Returns the most downloaded model of a given task on the Hugging Face Hub.\"\n inputs = {\n \"task\": {\n \"type\": \"string\",\n \"description\": \"The task category (e.g., text-classification, depth-estimation).\",\n }\n }\n output_type = \"string\"\n\n def forward(self, task: str):\n from huggingface_hub import list_models\n model = next(iter(list_models(filter=task, sort=\"downloads\", direction=-1)))\n return model.id\n\nmodel_downloads_tool = HFModelDownloadsTool()<\/code><\/pre>\n\u6b64\u7ed3\u6784\u5141\u8bb8 LLM \u6709\u6548\u5730\u4e0e\u5de5\u5177\u4ea4\u4e92\uff0c\u5229\u7528\u5176\u529f\u80fd\u5b8c\u6210\u7279\u5b9a\u4efb\u52a1\u3002<\/p>\n
\u8981\u6784\u5efa\u5de5\u5177\uff0c\u5f00\u53d1\u4eba\u5458\u53ef\u4ee5\u5c06 smolagents \u63d0\u4f9b\u7684 Tool<\/code> \u7c7b\u5b50\u7c7b\u5316\u3002\u8fd9\u79cd\u65b9\u6cd5\u5141\u8bb8\u5c06\u51fd\u6570\u7684\u903b\u8f91\u4e0e\u5176\u5143\u6570\u636e\u4e00\u8d77\u5c01\u88c5\uff0c\u4ece\u800c\u4fc3\u8fdb\u4e0e\u4ee3\u7406\u7684\u65e0\u7f1d\u96c6\u6210\u3002\u6216\u8005\uff0c\u5bf9\u4e8e\u66f4\u7b80\u5355\u7684\u5de5\u5177\uff0c\u53ef\u4ee5\u4f7f\u7528 @tool<\/code> \u88c5\u9970\u5668\u4ee5\u66f4\u5c11\u7684\u6837\u677f\u4ee3\u7801\u5b9e\u73b0\u7c7b\u4f3c\u7684\u7ed3\u679c\u3002<\/p>\n\u4e00\u65e6\u5b9a\u4e49\uff0c\u5de5\u5177\u5c31\u53ef\u4ee5\u5728 Hugging Face Hub \u4e0a\u5171\u4eab\uff0c\u4ece\u800c\u5b9e\u73b0\u91cd\u7528\u548c\u534f\u4f5c\u3002\u901a\u8fc7\u5728\u5de5\u5177\u5b9e\u4f8b\u4e0a\u8c03\u7528 push_to_hub()<\/code> \u65b9\u6cd5\uff0c\u5f00\u53d1\u4eba\u5458\u53ef\u4ee5\u5c06\u4ed6\u4eec\u7684\u5de5\u5177\u4e0a\u4f20\u5230 Hub\uff0c\u4f7f\u5176\u53ef\u4ee5\u52a0\u8f7d\u5230\u5176\u4ed6\u4ee3\u7406\u4e2d\u3002\u8fd9\u4fc3\u8fdb\u4e86\u4e00\u4e2a\u534f\u4f5c\u751f\u6001\u7cfb\u7edf\uff0c\u5176\u4e2d\u5de5\u5177\u53ef\u4ee5\u8f7b\u677e\u5730\u5206\u53d1\u548c\u4f7f\u7528\u5728\u5404\u4e2a\u9879\u76ee\u4e2d\u3002<\/p>\n4\u3001\u5b9a\u4e49\u4e00\u4e2a\u7b80\u5355\u7684\u5de5\u5177<\/h2>\n
\u53ef\u4ee5\u4f7f\u7528 @tool<\/code> \u88c5\u9970\u5668\u521b\u5efa\u4e00\u4e2a\u5de5\u5177\u3002<\/p>\nfrom smoltools import tool\n\n@tool\ndef add_numbers(a: int, b: int) -> int:\n \"\"\"Adds two numbers.\"\"\"\n return a + b\n\n# Using the tool\nresult = add_numbers(3, 5)\nprint(result) # Output: 8<\/code><\/pre>\n\u53ef\u4ee5\u5c06\u5de5\u5177\u63a8\u9001\u5230 Hugging Face Hub \u4ee5\u4f9b\u91cd\u7528\u3002<\/p>\n
add_numbers.push_to_hub(\"your-username\/add_numbers_tool\")<\/code><\/pre>\n\u4ece Hub \u52a0\u8f7d\u5de5\u5177\uff1a<\/p>\n
from smoltools import Tool\n\n# Load the tool from the Hub\ntool_from_hub = Tool.from_hub(\"your-username\/add_numbers_tool\")\n\n# Use the loaded tool\nprint(tool_from_hub.run(3, 5)) # Output: 8<\/code><\/pre>\n5\u3001\u4f7f\u7528 OpenTelemetry \u68c0\u67e5\u4ee3\u7406\u8fd0\u884c<\/h2>\n
\u76d1\u63a7\u548c\u8c03\u8bd5\u4ee3\u7406\u64cd\u4f5c\u5bf9\u4e8e\u4fdd\u6301\u6027\u80fd\u548c\u53ef\u9760\u6027\u81f3\u5173\u91cd\u8981\u3002smolagents \u4e0e OpenTelemetry \u96c6\u6210\uff0c\u4ee5\u4fc3\u8fdb\u5bf9\u4ee3\u7406\u8fd0\u884c\u7684\u5168\u9762\u8bb0\u5f55\u548c\u68c0\u67e5\u3002\u901a\u8fc7\u4f7f\u7528 OpenTelemetry \u68c0\u6d4b\u4ee3\u7406\uff0c\u5f00\u53d1\u4eba\u5458\u53ef\u4ee5\u8ddf\u8e2a\u64cd\u4f5c\u3001\u76d1\u63a7\u6027\u80fd\u5e76\u8bc6\u522b\u6f5c\u5728\u95ee\u9898\u3002\u8be5\u8fc7\u7a0b\u5305\u62ec\uff1a<\/p>\n
\u5b89\u88c5\u6240\u9700\u7684\u8f6f\u4ef6\u5305\uff1a<\/p>\n
pip install smolagents arize-phoenix opentelemetry-sdk opentelemetry-exporter-otlp openinference-instrumentation-smolagents\n<\/code><\/pre>\n\u8fd0\u884c\u6536\u96c6\u5668\uff1a<\/p>\n
python -m phoenix.server.main serve<\/code><\/pre>\n\u8bbe\u7f6e Instrumentor\uff1a<\/p>\n
from opentelemetry import trace\nfrom opentelemetry.sdk.trace import TracerProvider\nfrom opentelemetry.sdk.trace.export import BatchSpanProcessor\nfrom openinference.instrumentation.smolagents import SmolagentsInstrumentor\nfrom opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter\nfrom opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor\n\nendpoint = \"http:\/\/0.0.0.0:6006\/v1\/traces\"\ntrace_provider = TracerProvider()\ntrace_provider.add_span_processor(SimpleSpanProcessor(OTLPSpanExporter(endpoint)))\n\nSmolagentsInstrumentor().instrument(tracer_provider=trace_provider)<\/code><\/pre>\n\u6b64\u8bbe\u7f6e\u53ef\u5b9e\u73b0\u5b9e\u65f6\u76d1\u63a7\u548c\u8fd0\u884c\u540e\u5206\u6790\uff0c\u4ece\u800c\u63d0\u4f9b\u6709\u5173\u4ee3\u7406\u884c\u4e3a\u7684\u5b9d\u8d35\u89c1\u89e3\u3002<\/p>\n
6\u3001\u786e\u4fdd\u5b89\u5168\u7684\u4ee3\u7801\u6267\u884c<\/h2>\n
\u9274\u4e8e\u4ee3\u7406\u53ef\u80fd\u4f1a\u6267\u884c\u7531 LLM \u751f\u6210\u7684\u4ee3\u7801\uff0c\u56e0\u6b64\u786e\u4fdd\u5b89\u5168\u7684\u6267\u884c\u73af\u5883\u81f3\u5173\u91cd\u8981\u3002smolagents \u63d0\u4f9b\u4e86\u9632\u8303\u6f5c\u5728\u5b89\u5168\u98ce\u9669\u7684\u673a\u5236\uff1a<\/p>\n
\n- \u672c\u5730 Python \u89e3\u91ca\u5668\uff1a\u4e00\u79cd\u5b9a\u5236\u7684\u89e3\u91ca\u5668\uff0c\u53ef\u5c06\u5bfc\u5165\u9650\u5236\u4e3a\u7528\u6237\u5b9a\u4e49\u7684\u5217\u8868\uff0c\u9650\u5236\u64cd\u4f5c\u6570\u91cf\u4ee5\u9632\u6b62\u8d44\u6e90\u8017\u5c3d\uff0c\u5e76\u7981\u6b62\u672a\u5b9a\u4e49\u7684\u64cd\u4f5c\u3002<\/li>\n
- E2B \u4ee3\u7801\u6267\u884c\u5668\uff1a\u4e3a\u4e86\u589e\u5f3a\u5b89\u5168\u6027\uff0csmolagents \u4e0e E2B \u96c6\u6210\uff0c\u5141\u8bb8\u5728\u6c99\u76d2\u9694\u79bb\u5bb9\u5668\u4e2d\u6267\u884c\u4ee3\u7801\u3002\u6b64\u8bbe\u7f6e\u53ef\u786e\u4fdd\u6267\u884c\u7684\u4ee3\u7801\u4e0d\u4f1a\u5f71\u54cd\u672c\u5730\u73af\u5883\uff0c\u4ece\u800c\u63d0\u4f9b\u989d\u5916\u7684\u4fdd\u62a4\u5c42\u3002<\/li>\n<\/ul>\n
\u5b9e\u65bd\u8fd9\u4e9b\u5b89\u5168\u63aa\u65bd\u5bf9\u4e8e\u7ef4\u62a4\u8fd0\u884c\u4ee3\u7406\u5de5\u4f5c\u6d41\u7684\u7cfb\u7edf\u7684\u5b8c\u6574\u6027\u548c\u5b89\u5168\u6027\u81f3\u5173\u91cd\u8981\u3002<\/p>\n
7\u3001\u4ee3\u7801\u4ee3\u7406<\/h2>\n
\u4ee3\u7801\u4ee3\u7406\u662f smolagents \u4e2d\u7684\u4e00\u79cd\u4ee3\u7406\uff0c\u5b83\u5728\u64cd\u4f5c\u7684\u6bcf\u4e2a\u6b65\u9aa4\u4e2d\u7f16\u5199\u548c\u6267\u884c Python \u4ee3\u7801\u7247\u6bb5\u3002\u8fd9\u79cd\u65b9\u6cd5\u5141\u8bb8\u4ee3\u7406\u901a\u8fc7\u52a8\u6001\u7f16\u5199\u548c\u8fd0\u884c\u4ee3\u7801\u6765\u6267\u884c\u590d\u6742\u4efb\u52a1\uff0c\u5229\u7528 Python \u5e7f\u6cdb\u7684\u5e93\u548c\u5de5\u5177\u751f\u6001\u7cfb\u7edf\u3002<\/p>\n
\u9ed8\u8ba4\u60c5\u51b5\u4e0b\uff0c\u8fd9\u4e9b\u4ee3\u7801\u7247\u6bb5\u7684\u6267\u884c\u53d1\u751f\u5728\u672c\u5730\u73af\u5883\u4e2d\u3002\u4f46\u662f\uff0c\u4e3a\u4e86\u786e\u4fdd\u5b89\u5168\u6027\u5e76\u9632\u6b62\u4e0e\u6267\u884c\u4e0d\u53d7\u4fe1\u4efb\u7684\u4ee3\u7801\u76f8\u5173\u7684\u6f5c\u5728\u98ce\u9669\uff0csmolagents \u63d0\u4f9b\u4e86\u5b89\u5168\u4ee3\u7801\u6267\u884c\u673a\u5236\u3002\u8fd9\u5305\u62ec\u4e0e E2B \u7b49\u670d\u52a1\u96c6\u6210\uff0c\u8fd9\u4e9b\u670d\u52a1\u5141\u8bb8\u5728\u6c99\u76d2\u9694\u79bb\u7684\u5bb9\u5668\u4e2d\u6267\u884c\u4ee3\u7801\uff0c\u4ece\u800c\u4fdd\u62a4\u672c\u5730\u73af\u5883\u514d\u53d7\u672a\u7ecf\u6388\u6743\u7684\u64cd\u4f5c\u6216\u5b89\u5168\u6f0f\u6d1e\u7684\u5f71\u54cd\u3002<\/p>\n
from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel\n\nagent = CodeAgent(\n tools=[retriever_tool, DuckDuckGoSearchTool()], model=HfApiModel(), max_iterations=4, verbose=True\n)\nagent.run(\"How to download a model from Huggingface?\")<\/code><\/pre>\n\u6b64\u4ee3\u7406\u5c06\u81ea\u5b9a\u4e49\u68c0\u7d22\u5de5\u5177\u4e0e DuckDuckGo \u641c\u7d22\u76f8\u7ed3\u5408\uff0c\u4f7f\u5176\u80fd\u591f\u4ece\u7ed3\u6784\u5316\u548c\u975e\u7ed3\u6784\u5316\u6765\u6e90\u83b7\u53d6\u7cbe\u786e\u7b54\u6848\u3002<\/p>\n
from smolagents import Tool\nfrom langchain.docstore.document import Document\nfrom langchain_community.retrievers import BM25Retriever\n\nclass RetrieverTool(Tool):\n name = \"retriever\"\n description = \"Uses semantic search to retrieve the parts of transformers documentation that could be most relevant to answer your query.\"\n inputs = {\n \"query\": {\n \"type\": \"string\",\n \"description\": \"The query to perform. This should be semantically close to your target documents. Use the affirmative form rather than a question.\",\n }\n }\n output_type = \"string\"\n\n def __init__(self, docs, **kwargs):\n super().__init__(**kwargs)\n self.retriever = BM25Retriever.from_documents(\n docs, k=10\n )\n\n def forward(self, query: str) -> str:\n assert isinstance(query, str), \"Your search query must be a string\"\n\n docs = self.retriever.invoke(\n query,\n )\n return \"\\nRetrieved documents:\\n\" + \"\".join(\n [\n f\"\\n\\n===== Document {str(i)} =====\\n\" + doc.page_content\n for i, doc in enumerate(docs)\n ]\n )<\/code><\/pre>\n\u8981\u4f7f\u7528\u6b64\u5de5\u5177\uff1a<\/p>\n
retriever_tool = RetrieverTool(docs_processed)\nretriever_tool.forward(\"How to download a model from Huggingface?\")<\/code><\/pre>\n8\u3001\u57fa\u4e8e RAG \u7684\u68c0\u7d22<\/h2>\n
\u68c0\u7d22\u589e\u5f3a\u751f\u6210 (RAG) \u7cfb\u7edf\u4ece SmolAgents \u4e2d\u53d7\u76ca\u532a\u6d45\u3002\u4f46\u662f\uff0c\u666e\u901a RAG \u7cfb\u7edf\u6709\u5c40\u9650\u6027\uff0c\u4f8b\u5982\u4f9d\u8d56\u4e8e\u5355\u4e2a\u68c0\u7d22\u6b65\u9aa4\u3002SmolAgents \u901a\u8fc7\u4ee5\u4e0b\u65b9\u5f0f\u89e3\u51b3\u4e86\u8fd9\u4e9b\u95ee\u9898\uff1a<\/p>\n
\n- \u6539\u8fdb\u7684\u8bed\u4e49\u76f8\u4f3c\u6027\u8bc4\u5206\u3002<\/li>\n
- \u8fed\u4ee3\u68c0\u7d22\u4ee5\u63d0\u9ad8\u51c6\u786e\u6027\u3002<\/li>\n<\/ul>\n
import datasets\nfrom langchain.docstore.document import Document\nfrom langchain.text_splitter import RecursiveCharacterTextSplitter\nfrom langchain_community.retrievers import BM25Retriever\n\nknowledge_base = datasets.load_dataset(\"m-ric\/huggingface_doc\", split=\"train\")\nknowledge_base = knowledge_base.filter(lambda row: row[\"source\"].startswith(\"huggingface\/transformers\"))\n\nsource_docs = [\n Document(page_content=doc[\"text\"], metadata={\"source\": doc[\"source\"].split(\"\/\")[1]})\n for doc in knowledge_base\n]\n\ntext_splitter = RecursiveCharacterTextSplitter(\n chunk_size=500,\n chunk_overlap=50,\n add_start_index=True,\n strip_whitespace=True,\n separators=[\"\\n\\n\", \"\\n\", \".\", \" \", \"\"]\n)\ndocs_processed = text_splitter.split_documents(source_docs)<\/code><\/pre>\n\u5c06\u5176\u4e0e\u68c0\u7d22\u5de5\u5177\u7ed3\u5408\u4f7f\u7528\uff1a<\/p>\n
retriever_tool = RetrieverTool(docs_processed)\nagent = CodeAgent(\n tools=[retriever_tool], model=HfApiModel(), max_iterations=4, verbose=True\n)\nagent_output = agent.run(\"For a transformers model training, which is slower, the forward or the backward pass?\")\nprint(\"Final output:\")\nprint(agent_output)<\/code><\/pre>\n9\u3001\u89e3\u51b3 Vanilla RAG \u6311\u6218<\/h2>\n
Vanilla RAG \u4e2d\u7684\u6311\u6218\u5305\u62ec\uff1a<\/p>\n
\n- \u5355\u4e00\u68c0\u7d22\u6b65\u9aa4\uff1a\u5f53\u7ed3\u679c\u4e0d\u4f73\u65f6\uff0c\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u591f\u7528\u3002<\/li>\n
- \u8bed\u4e49\u8bc4\u5206\u9650\u5236\uff1a\u67e5\u8be2\u548c\u6587\u6863\u4e4b\u95f4\u7684\u8bc4\u5206\u4e0d\u4e00\u81f4\u3002<\/li>\n<\/ul>\n
SmolAgents \u901a\u8fc7\u4ee5\u4e0b\u65b9\u5f0f\u89e3\u51b3\u8fd9\u4e9b\u95ee\u9898\uff1a<\/p>\n
\n- \u591a\u6b65\u9aa4\u68c0\u7d22\u3002<\/li>\n
- \u4f18\u5316\u8bed\u4e49\u5339\u914d\u3002<\/li>\n<\/ul>\n
10\u3001\u7ed3\u675f\u8bed<\/h2>\n
Hugging Face \u7684 SmolAgents \u63d0\u4f9b\u4e86\u4e00\u4e2a\u5f3a\u5927\u7684\u6846\u67b6\uff0c\u7528\u4e8e\u6784\u5efa\u667a\u80fd\u3001\u7279\u5b9a\u4e8e\u4efb\u52a1\u7684\u4ee3\u7406\u3002\u51ed\u501f\u5176\u65e0\u7f1d\u96c6\u6210\u7684\u5de5\u5177\u3001\u5b89\u5168\u7684\u4ee3\u7801\u6267\u884c\u548c\u9ad8\u7ea7\u68c0\u7d22\u529f\u80fd\uff0c\u5b83\u662f AI \u5f00\u53d1\u4eba\u5458\u4e0d\u53ef\u6216\u7f3a\u7684\u5e93\u3002\u7acb\u5373\u8bd5\u7528\u5e76\u5c06\u60a8\u7684\u9879\u76ee\u63d0\u5347\u5230\u4e00\u4e2a\u65b0\u7684\u6c34\u5e73\uff01<\/p>\n
\n\n","protected":false},"excerpt":{"rendered":"
SmolAgents \u662f Hugging Face \u7684\u4e00\u4e2a\u5c16\u7aef\u5e93\uff0c\u5141\u8bb8\u5f00\u53d1\u4eba\u5458\u521b\u5efa\u80fd\u591f\u89e3\u51b3\u590d\u6742\u4efb\u52a1\u7684\u667a\u80fd\u3001\u7279\u5b9a\u9886\u57df\u7684\u4ee3\u7406\u3002\u672c\u535a\u5ba2\u5c06\u6df1\u5165\u4ecb\u7ecd SmolAgents\uff0c\u5305\u62ec\u5de5\u5177\u3001\u4ee3\u7801\u4ee3\u7406\u3001\u5b89\u5168\u4ee3\u7801\u6267\u884c\u548c\u5b9e\u9645\u5b9e\u73b0\u3002\u6211\u4eec\u8fd8\u5c06\u7ed3\u5408\u4ee3\u7801\u793a\u4f8b\u4f7f\u6982\u5ff5\u66f4\u6e05\u6670\u3002 1\u3001\u4ee3\u7406\u7b80\u4ecb \u5728 AI \u9886\u57df\uff0c\u4ee3\u7406\u662f LLM \u8f93\u51fa\u63a7\u5236\u5de5\u4f5c\u6d41\u7684\u7a0b\u5e8f\u3002LLM \u5bf9\u7cfb\u7edf\u64cd\u4f5c\u7684\u5f71\u54cd\u7a0b\u5ea6\u51b3\u5b9a\u4e86\u5176\u4ee3\u7406\u7ea7\u522b\u3002\u8fd9\u79cd\u4ee3\u7406\u5b58\u5728\u4e8e\u4ee5\u4e0b\u8303\u56f4\u5185\uff1a \u65e0\u4ee3\u7406 ()\uff1aLLM \u8f93\u51fa\u4e0d\u5f71\u54cd\u7a0b\u5e8f\u6d41\u7a0b\u3002 \u4f4e\u4ee3\u7406 ()\uff1aLLM \u8f93\u51fa\u786e\u5b9a\u5de5\u4f5c\u6d41\u4e2d\u7684\u6761\u4ef6\u5206\u652f\u3002 \u4e2d\u4ee3\u7406 ()\uff1aLLM \u8f93\u51fa\u51b3\u5b9a\u6267\u884c\u54ea\u4e9b\u529f\u80fd\u6216\u5de5\u5177\u3002 \u9ad8\u4ee3\u7406 ()\uff1aLLM \u8f93\u51fa\u63a7\u5236\u8fed\u4ee3\u8fc7\u7a0b\u5e76\u53ef\u4ee5\u542f\u52a8\u5176\u4ed6\u4ee3\u7406\u5de5\u4f5c\u6d41\u3002 \u5f53\u4efb\u52a1\u9700\u8981\u57fa\u4e8e\u52a8\u6001\u8f93\u5165\u8fdb\u884c\u8c03\u6574\u7684\u7075\u6d3b\u5de5\u4f5c\u6d41\u65f6\uff0c\u5b9e\u65bd\u4ee3\u7406\u662f\u6709\u76ca\u7684\u3002\u4f46\u662f\uff0c\u5bf9\u4e8e\u5177\u6709\u53ef\u9884\u6d4b\u548c\u5b9a\u4e49\u660e\u786e\u7684\u6d41\u7a0b\u7684\u4efb\u52a1\uff0c\u4f20\u7edf\u7684\u786e\u5b9a\u6027\u7f16\u7a0b\u53ef\u80fd\u5c31\u8db3\u591f\u4e86\u3002 !pip install -q smolagents !pip install huggingface_hub 2\u3001\u6784\u5efa\u6709\u6548\u7684\u4ee3\u7406 \u521b\u5efa\u5f3a\u5927\u7684\u4ee3\u7406\u6d89\u53ca\u51e0\u4e2a\u5173\u952e\u8003\u8651\u56e0\u7d20\uff1a \u5b9a\u4e49\u660e\u786e\u7684\u76ee\u6807\uff1a\u4e3a\u4ee3\u7406\u5efa\u7acb\u7279\u5b9a\u7684\u4efb\u52a1\u6216\u76ee\u6807\uff0c\u4ee5\u786e\u4fdd\u4e13\u6ce8\u7684\u8868\u73b0\u3002 \u7ed3\u5408\u76f8\u5173\u5de5\u5177\uff1a\u4e3a\u4ee3\u7406\u914d\u5907\u4e0e\u5176\u76ee\u6807\u4e00\u81f4\u7684\u5de5\u5177\uff0c\u589e\u5f3a\u5176\u6267\u884c\u6307\u5b9a\u4efb\u52a1\u7684\u80fd\u529b\u3002 \u5b9e\u65bd\u8bb0\u5fc6\u673a\u5236\uff1a\u4f7f\u4ee3\u7406\u80fd\u591f\u4fdd\u7559\u4ee5\u524d\u4ea4\u4e92\u4e2d\u7684\u4e0a\u4e0b\u6587\uff0c\u4ece\u800c\u505a\u51fa\u66f4\u8fde\u8d2f\u548c\u660e\u667a\u7684\u54cd\u5e94\u3002 \u786e\u4fdd\u5b89\u5168\u7684\u4ee3\u7801\u6267\u884c\uff1a\u4fdd\u62a4\u6267\u884c\u73af\u5883\u4ee5\u9632\u6b62\u672a\u7ecf\u6388\u6743\u7684\u64cd\u4f5c\u6216\u5b89\u5168\u6f0f\u6d1e\u3002 3\u3001\u5728 smolagents \u4e2d\u4f7f\u7528\u5de5\u5177 \u5de5\u5177\u5bf9\u4e8e\u6269\u5c55\u4ee3\u7406\u7684\u529f\u80fd\u81f3\u5173\u91cd\u8981\u3002\u5728 smolagents \u4e2d\uff0c\u5de5\u5177\u672c\u8d28\u4e0a\u662f LLM \u53ef\u4ee5\u5728\u4ee3\u7406\u7cfb\u7edf\u4e2d\u4f7f\u7528\u7684\u529f\u80fd\u3002\u4e3a\u4e86\u5b9e\u73b0\u8fd9\u4e00\u70b9\uff0c\u5de5\u5177\u88ab\u5c01\u88c5\u5728\u63d0\u4f9b\u5143\u6570\u636e\u7684\u7c7b\u4e2d\uff0c\u5e2e\u52a9 LLM \u4e86\u89e3\u5b83\u4eec\u7684\u7528\u6cd5\u3002\u4f8b\u5982\uff1a from smolagents import Tool class HFModelDownloadsTool(Tool): […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[13],"tags":[],"class_list":["post-53751","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts\/53751","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/comments?post=53751"}],"version-history":[{"count":0,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/posts\/53751\/revisions"}],"wp:attachment":[{"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/media?parent=53751"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/categories?post=53751"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/fwq.ai\/blog\/wp-json\/wp\/v2\/tags?post=53751"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}