{"id":1563,"date":"2024-05-22T17:36:15","date_gmt":"2024-05-22T17:36:15","guid":{"rendered":"https:\/\/www.nicekj.com\/?p=1563"},"modified":"2024-05-22T17:36:44","modified_gmt":"2024-05-22T17:36:44","slug":"shiyong-python-diaojiao-openai-rang-ai-huidadegengwanmei","status":"publish","type":"post","link":"https:\/\/www.nicekj.com\/shiyong-python-diaojiao-openai-rang-ai-huidadegengwanmei.html","title":{"rendered":"\u4f7f\u7528 Python \u8c03\u6559 OpenAI \uff0c\u8ba9 AI \u56de\u7b54\u7684\u66f4\u5b8c\u7f8e"},"content":{"rendered":"

\u5728\u8fd9\u4e2a\u5feb\u901f\u53d1\u5c55\u7684\u6570\u5b57\u65f6\u4ee3\uff0cAI\u4eba\u5de5\u667a\u6167\u5df2\u7ecf\u6e10\u6e10\u6210\u4e3a\u6211\u4eec\u65e5\u5e38\u751f\u6d3b\u7684\u4e00\u90e8\u5206\u3002\u800cOpenAI\uff0c\u4f5c\u4e3aAI\u7814\u7a76\u7684\u673a\u6784\u4e4b\u4e00\uff0c\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u5de5\u5177\u548cLLM\u6a21\u578b\uff0c\u5982GPT-3.5\u3001GPT-4\uff0c\u6765\u5e2e\u52a9\u6211\u4eec\u66f4\u597d\u5730\u7406\u89e3\u548c\u4f7f\u7528AI\u3002<\/p>\n

\u5728\u672c\u6587\u4e2d\uff0c\u5c06\u4ee5\u7b80\u5355\u660e\u4e86\u7684\u65b9\u5f0f\u6765\u8bf4\u660e\u5982\u4f55\u63a7\u5236OpenAI\u6a21\u578b\u7684\u4e00\u4e9b\u5173\u952e\u8bbe\u5b9a\u3002\u4ece\u751f\u6210\u591a\u6837\u56de\u590d\u7684 n<\/code> \u53c2\u6570\uff0c\u5230\u63a7\u5236\u56de\u7b54\u957f\u5ea6\u7684 max_tokens<\/code> \uff0c\u518d\u5230\u5efa\u7acb\u8bcd\u6c47\u9ed1\u540d\u5355\u7684 stop<\/code> \u53c2\u6570\uff0c\u4ee5\u53ca\u6700\u540e\u7684 temperature<\/code> \u53c2\u6570\uff0c\u8fd9\u4e9b\u90fd\u662f\u8ba9AI\u751f\u6210\u7684\u5185\u5bb9\u66f4\u52a0\u8d34\u5408\u6211\u4eec\u9700\u6c42\u7684\u5b9e\u7528\u5de5\u5177\u3002\u8ba9\u6211\u4eec\u4e00\u8d77\u63a2\u7d22\u8fd9\u4e2a\u4ee4\u4eba\u5174\u594b\u7684AI\u4e16\u754c\u3002<\/p>\n

\u5982\u4f55\u7528OpenAI\u83b7\u5f97\u591a\u4e2a\u7b54\u6848\uff1f<\/h2>\n
\n

\u4f7f\u7528\u53c2\u6570 n<\/code> \u53ef\u9009\u62e9\u591a\u4e2a\u7b54\u6848\u3002<\/p>\n<\/blockquote>\n

\u5f53\u6211\u4eec\u4f7f\u7528ChatGPT\u65f6\uff0c\u901a\u5e38\u6211\u4eec\u4f1a\u5f97\u5230\u4e00\u4e2a\u7b54\u6848\u3002\u4f46\u5982\u679c\u4f60\u5bf9\u8fd9\u4e2a\u7b54\u6848\u4e0d\u592a\u6ee1\u610f\u600e\u4e48\u529e\uff1f\u6709\u65f6\u5019\uff0c\u6211\u4eec\u5e0c\u671b\u80fd\u770b\u5230\u4e0d\u540c\u7684\u7b54\u6848\uff0c\u4ece\u4e2d\u9009\u62e9\u4e00\u4e2a\u6700\u9002\u5408\u7684\u3002\u8fd9\u65f6\u5019\u4f60\u9700\u8981\u5728\u7b54\u6848\u7684\u4e0b\u65b9\u6309\u4e0b\u300c\u91cd\u65b0\u751f\u6210\u300d\u6309\u94ae\uff0c\u8ba9ChatGPT\u91cd\u65b0\u751f\u6210\u4e00\u7ec4\u65b0\u7684\u56de\u7b54\u3002<\/p>\n

\n

\"\u56fe\u7247\"<\/figure>\n<\/p>\n

ChatGPT\u7684\u72b6\u51b5<\/p>\n

\u597d\u6d88\u606f\u662f\u5728OpenAI\uff0c\u6211\u4eec\u53ef\u4ee5\u8ba9\u5b83\u4e00\u6b21\u7ed9\u6211\u4eec\u597d\u51e0\u4e2a\u7b54\u6848\u4f5c\u4e3a\u9009\u62e9\u3002\u8fd9\u5c31\u662f n<\/code> \u8fd9\u4e2a\u53c2\u6570\u7684\u4f5c\u7528\u3002\u901a\u8fc7\u8bbe\u7f6e n<\/code> \u53c2\u6570\uff0c\u6211\u4eec\u53ef\u4ee5\u544a\u8bc9OpenAI\u4e00\u6b21\u751f\u6210\u591a\u5c11\u4e2a\u7b54\u6848\u3002\u6bd4\u5982\u8bf4\uff0c\u8bbe\u7f6e n = 3<\/code> \u5c31\u4f1a\u4e00\u6b21\u5f97\u5230\u4e09\u4e2a\u4e0d\u540c\u7684\u7b54\u6848\u3002<\/p>\n

\u4f46\u8981\u6ce8\u610f\uff0c\u8fd9\u4e9b\u7b54\u6848\u90fd\u662f\u9488\u5bf9\u540c\u4e00\u4e2a\u95ee\u9898\uff0c\u56de\u7b54\u7684\u4e0d\u540c\u56de\u7b54\uff0c\u800c\u4e0d\u662f\u4e00\u6765\u4e00\u56de\u7684\u8fde\u7eed\u5bf9\u8bdd\u3002<\/p>\n

\u5982\u4f55\u505a\uff1f<\/h3>\n

\u5047\u8bbe\u4f60\u4f7f\u7528Python\u8bed\u8a00\u6765\u8ddfOpenAI\u6c9f\u901a\uff0c\u53ef\u4ee5\u8fd9\u6837\u5199\uff1a<\/p>\n

<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> openai<=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = openai.ChatCompletion.create(<\/span>\n model<\/span>= \"gpt-3.5-turbo\"<\/span>,<\/span>\n messages<\/span> = [{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u4eca\u5929\u662f\u4e2a\u6674\u6717\u7684\u5929\u6c14\u5427\"<\/span>}],<\/span>\n n<\/span> = 2<\/span>        <\/span>\n)<\/span>\n<\/span>\nprint(reply)<\/span>\n<\/span>\nfor choice in reply[\"choices\"]<\/span>:<\/span>\n    print(choice[\"index\"]<\/span>, choice[\"messages\"]<\/span>[\"content\"]<\/span>)<\/span>\n<\/span>\n<\/span>\n openai>=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = client.chat.completions.create(<\/span>\n  model<\/span>=\"gpt-3.5-turbo\"<\/span>,<\/span>\n  messages<\/span>=[{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u4eca\u5929\u662f\u4e2a\u6674\u6717\u7684\u5929\u6c14\u5427\"<\/span>}],<\/span>\n  n<\/span>=2<\/span><\/span>\n)<\/span>\n<\/span>\npprint(reply)<\/span>\n<\/span>\nfor choice in reply.choices:<\/span>\n    print(choice.index, choice.message.content)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u8fd9\u6bb5\u7a0b\u5e8f\u4f1a\u8ba9OpenAI\u7ed9\u6211\u4eec\u4e24\u4e2a\u4e0d\u540c\u7684\u56de\u7b54\u3002\u6211\u4eec\u900f\u8fc7\u4e00\u4e2afor loop \u56de\u5708\u628a\u8fd9\u4e9b\u56de\u7b54\u4e00\u4e2a\u4e00\u4e2a\u663e\u793a\u51fa\u6765\u3002\u6bcf\u4e2a\u56de\u7b54\u90fd\u6709\u4e00\u4e2a\u7f16\u53f7\uff0c\u4ece0\u5f00\u59cb\u3002<\/p>\n
\n
\"\u56fe\u7247\"<\/figure>\n<\/p>\n
\u8fd9\u6837\u4f60\u5c31\u53ef\u4ee5\u770b\u5230\u4e24\u4e2a\u4e0d\u540c\u7684\u7b54\u6848\u4e86\u3002\u4e0d\u8fc7\uff0c\u8fd9\u91cc\u5fc5\u9700\u8981\u8bb0\u5f97\u7684\u4e00\u70b9\u662f\uff0c\u5f53\u6211\u4eec\u8981\u6c42\u66f4\u591a\u7684\u7b54\u6848\u65f6\uff0c\u4f7f\u7528\u7684 token<\/code> \u6570\u91cf\u4e5f\u4f1a\u589e\u52a0\u3002token<\/code> \u662fOpenAI\u8ba1\u7b97\u4f7f\u7528\u91cf\u7684\u5355\u4f4d\uff0c\u6240\u4ee5\u591a\u4e2a\u7b54\u6848\u610f\u5473\u7740\u53ef\u80fd\u9700\u8981\u4ed8\u51fa\u66f4\u591a\u8d39\u7528\u3002<\/p>\n
\u5982\u4f55\u8bbe\u5b9aOpenAI\u56de\u7b54\u7684\u957f\u5ea6\u9650\u5236\uff1f<\/h2>\n
\n
\u4f7f\u7528max_tokens\u53c2\u6570\uff0c\u63a7\u5236\u56de\u590d\u5185\u5bb9\u957f\u5ea6<\/p>\n<\/blockquote>\n
\u56de\u7b54\u7684\u7b54\u6848\u8d8a\u957f\uff0c\u6240\u7528\u5230\u7684 token<\/code> \u6570\u8d8a\u591a\uff0c\u8fde\u5e26\u7684\u8d39\u7528\u4e5f\u4f1a\u8d8a\u9ad8\u3002\u56e0\u6b64\uff0c\u6709\u65f6\u5019\u6211\u4eec\u5728\u4f7f\u7528OpenAI\u65f6\u4e0d\u5e0c\u671b\u6a21\u578b\u7ed9\u6211\u4eec\u592a\u957f\u7684\u56de\u7b54\uff0c\u4ee5\u4fbf\u4e8e\u8282\u7701\u6210\u672c\u3002\u8fd9\u65f6\uff0c\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u4e00\u4e2a\u53eb\u505a max_tokens<\/code> \u7684\u8bbe\u5b9a\u6765\u9650\u5236\u56de\u590d\u7684\u957f\u5ea6\uff0c\u63a7\u5236 token<\/code> \u7684\u4f7f\u7528\u6570\u91cf\u3002\u4ee5\u4e0b\u7528\u4e24\u4e2a\u4f8b\u5b50\u8bf4\u660e\uff1a<\/p>\n
\u4e3e\u4f8b 1<\/h3>\n
\u6bd4\u5982\u8bf4\uff0c\u6211\u4eec\u53ea\u60f3\u8981\u4e00\u4e2a\u5f88\u77ed\u7684\u56de\u7b54\uff0c\u53ef\u4ee5\u8fd9\u6837\u8bbe\u5b9a\uff1a<\/p>\n
<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> openai<=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = openai.ChatCompletion.create(<\/span>\n model<\/span>= \"gpt-3.5-turbo\"<\/span>,<\/span>\n messages<\/span> = [{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u51ac\u5929\u4e3a\u4f55\u4f1a\u4e0b\u96ea\uff1f\"<\/span>}],<\/span>\n max_tokens<\/span> = 10<\/span>        <\/span>\n)<\/span>\n<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"message\"]<\/span>[\"content\"]<\/span>)<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"finish_reason\"]<\/span>)<\/span>\nprint(reply[\"usage\"]<\/span>[\"completion_tokens\"]<\/span>)<\/span>\n<\/span>\n openai>=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = client.chat.completions.create(<\/span>\n    model<\/span> = \"gpt-3.5-turbo\"<\/span>,<\/span>\n    messages<\/span> = [<\/span>\n        {\"role\"<\/span>:\"user\"<\/span>, \"content\"<\/span>: \"\u51ac\u5929\u4e3a\u4f55\u4f1a\u4e0b\u96ea\uff1f\"<\/span>}<\/span>\n    ],<\/span>\n    max_tokens<\/span> = 10<\/span><\/span>\n)<\/span>\n<\/span>\nprint(reply.choices[0]<\/span>.message.content)<\/span>\nprint(reply.choices[0]<\/span>.finish_reason)<\/span>\nprint(reply.usage.completion_tokens)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u8fd9\u6837\u8bbe\u5b9a\u540e\uff0cOpenAI\u53ea\u4f1a\u56de\u7b54\u4e00\u4e2a\u5f88\u77ed\u7684\u53e5\u5b50\uff0c\u5927\u698210\u4e2a\u8bcd\uff08tokens\uff09\u5de6\u53f3\u3002\u663e\u7136\uff0c\u8fd9\u4e2a\u56de\u7b54\u4e0d\u53ef\u80fd\u5b8c\u6574\uff0c\u56e0\u4e3a max_tokens<\/code> \u9650\u5236\u4e86\u5b83\u7684\u957f\u5ea6\u3002\u6211\u4eec\u53ef\u4ee5\u770b\u5230 finish_reason<\/code> \u7684\u7ed3\u679c\u4e3alength\uff0c\u5c31\u53ef\u4ee5\u5f97\u77e5\uff0c\u8fd9\u6b21\u7684\u56de\u7b54\u662f\u56e0\u4e3a\u957f\u5ea6\u53d7\u5230\u9650\u5236\u800c\u505c\u6b62\uff0c\u56de\u7b54\u5e76\u4e0d\u5b8c\u6574\u3002<\/p>\n
\n
\"\u56fe\u7247\"<\/figure>\n<\/p>\n
\u4e3e\u4f8b 2<\/h3>\n
\u8fd8\u6709\uff0c\u6211\u4eec\u5fc5\u987b\u6ce8\u610f\u4e0d\u8981\u8bbe\u5b9a\u4e00\u4e2a\u592a\u5927\u7684 max_tokens<\/code> \u503c\u3002\u4f8b\u5982\uff0cgpt-3.5-turbo\u6a21\u578b\u6700\u591a\u80fd\u5904\u74064097\u4e2a token<\/code> \u3002\u5982\u679c\u8bbe\u5b9a\u7684 max_tokens<\/code> \u592a\u5927\uff0c\u8d85\u8fc7\u8fd9\u4e2a\u8303\u56f4\uff0cAPI\u5c31\u4f1a\u51fa\u73b0\u9519\u8bef\uff0c\u56e0\u4e3a\u5b83\u8d85\u51fa\u4e86\u6a21\u578b\u7684\u5904\u7406\u80fd\u529b\u3002<\/p>\n
<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> openai<=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = openai.ChatCompletion.create(<\/span>\n model<\/span>= \"gpt-3.5-turbo\"<\/span>,<\/span>\n messages<\/span> = [{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u51ac\u5929\u4e3a\u4f55\u4f1a\u4e0b\u96ea\uff1f\"<\/span>}],<\/span>\n max_tokens<\/span> = 4090<\/span>        <\/span>\n)<\/span>\n<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"message\"]<\/span>[\"content\"]<\/span>)<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"finish_reason\"]<\/span>)<\/span>\nprint(reply[\"usage\"]<\/span>[\"completion_tokens\"]<\/span>)<\/span>\n<\/span>\n openai>=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = client.chat.completions.create(<\/span>\n    model<\/span> = \"gpt-3.5-turbo\"<\/span>,<\/span>\n    messages<\/span> = [<\/span>\n        {\"role\"<\/span>:\"user\"<\/span>, \"content\"<\/span>: \"\u51ac\u5929\u4e3a\u4f55\u4f1a\u4e0b\u96ea\uff1f\"<\/span>}<\/span>\n    ],<\/span>\n    max_tokens<\/span> = 4090<\/span><\/span>\n)<\/span>\n<\/span>\nprint(reply.choices[0]<\/span>.message.content)<\/span>\nprint(reply.choices[0]<\/span>.finish_reason)<\/span>\nprint(reply.usage.completion_tokens)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u6267\u884c\u8fd9\u4e2a\u4f8b\u5b50\uff0c\u4f1a\u4ea7\u751f\u5982\u4e0b\u7684\u7ed3\u679c\u3002\u544a\u8bc9\u6211\u4eec\u8fd9\u4e2a\u6a21\u578b\u6700\u591a\u53ea\u80fd\u5904\u74064097\u4e2a token<\/code> \uff0c\u4f46\u662f\u6211\u4eec\u5374\u8bbe\u5b9a\u4e86\u8d85\u8fc7\u8fd9\u4e2a\u9650\u5236\u7684 token<\/code> \u6570\uff08 This model\u2019s maximum context length is 4097 tokens. However, you requested 4110 tokens (20 in the messages, 4090 in the completion\uff09<\/p>\n
\n
\"\u56fe\u7247\"<\/figure>\n<\/p>\n
\u603b\u4f53\u6765\u8bf4\uff0c max_tokens<\/code> \u662f\u4e00\u4e2a\u5f88\u6709\u7528\u7684\u8bbe\u5b9a\uff0c\u53ef\u4ee5\u5e2e\u6211\u4eec\u63a7\u5236\u56de\u7b54\u7684\u957f\u5ea6\uff0c\u8fdb\u800c\u63a7\u5236\u7528\u91cf\u4e0e\u6210\u672c\u3002\u53cd\u4e4b\uff0c\u7531\u4e8e\u5b83\u7684\u9ed8\u8ba4\u503c\u4e3a\u65e0\u9650\u5927\uff0c\u82e5\u6ca1\u6709\u7279\u522b\u8bbe\u5b9a max_tokens<\/code> \u7684\u8bdd\uff0c\u5c31\u6ca1\u6709\u4efb\u4f55\u9650\u5236\u3002\u56e0\u6b64\u8981\u8bb0\u5f97\uff0c\u8bbe\u5b9a\u5f97\u592a\u9ad8\u6216\u592a\u4f4e\u90fd\u53ef\u80fd\u4f1a\u5f71\u54cd\u56de\u7b54\u7684\u8d28\u91cf\u3002<\/p>\n
\u5982\u4f55\u8bbe\u5b9aOpenAI\u6a21\u578b\u56de\u7b54\u7684\u7981\u7528\u8bed\uff1f<\/h2>\n
\n
stop\u53c2\u6570\uff0c\u53ef\u5efa\u7acb\u8bcd\u6c47\u9ed1\u540d\u5355<\/p>\n<\/blockquote>\n
\u5f53\u6211\u4eec\u4f7f\u7528OpenAI\u7684\u65f6\u5019\uff0c\u53ef\u80fd\u4f1a\u62c5\u5fc3\u5b83\u7684\u56de\u7b54\u4e2d\u51fa\u73b0\u4e00\u4e9b\u6211\u4eec\u4e0d\u60f3\u8981\u7684\u8bcd\u6c47\u3002\u90a3\u4e48\uff0c\u6211\u4eec\u600e\u4e48\u624d\u80fd\u907f\u514d\u8fd9\u79cd\u60c5\u51b5\u5462\uff1f\u8fd9\u65f6\u5019\uff0cOpenAI\u63d0\u4f9b\u6211\u4eec\u4e00\u4e2a\u53eb\u505a stop<\/code> \u7684\u53c2\u6570\u6765\u8bbe\u5b9a\u9ed1\u540d\u5355\u3002<\/p>\n
stop<\/code> \u53c2\u6570\u53ef\u4ee5\u8ba9\u6211\u4eec\u6307\u5b9a\u4e00\u4e9b\u4e0d\u60f3\u8ba9AI\u4f7f\u7528\u7684\u8bcd\u6c47\u3002\u9ed8\u8ba4\u60c5\u51b5\u4e0b\uff0c\u8fd9\u4e2a list\u5217\u8868\u662f\u7a7a\u7684\uff0c\u8fd9\u4e5f\u5c31\u610f\u5473\u7740\u6ca1\u6709\u4efb\u4f55\u8bcd\u6c47\u88ab\u7981\u6b62\u3002<\/p>\n
\u6211\u4eec\u6700\u591a\u53ef\u4ee5\u5728\u8fd9\u4e2a\u5217\u8868\u4e2d\u8bbe\u5b9a\u56db\u4e2a\u8bcd\u6c47\u3002\u4e00\u65e6AI\u5728\u56de\u7b54\u4e2d\u9047\u5230\u8fd9\u4e9b\u8bcd\u6c47\u65f6\uff0c\u5c31\u4f1a\u7acb\u523b\u505c\u6b62\u56de\u7b54\uff0c\u5e76\u8fd4\u56de\u76ee\u524d\u7684\u7ed3\u679c\u3002\u4ee5\u4e0b\u4e3e\u4f8b\u8bf4\u660e\uff1a<\/p>\n
\u4e3e\u4f8b<\/h3>\n
\u5047\u5982\u6211\u4eec\u4e0d\u5e0c\u671b\u56de\u7b54\u4e2d\u51fa\u73b0\u300c\u597d\u300d\u8fd9\u4e2a\u5b57\u8bcd\uff0c\u6211\u4eec\u53ef\u4ee5\u900f\u8fc7\u4e0b\u9762\u65b9\u5f0f\u8bbe\u5b9a\uff1a<\/p>\n
<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div> openai<=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = openai.ChatCompletion.create(<\/span>\n model<\/span>= \"gpt-3.5-turbo\"<\/span>,<\/span>\n messages<\/span> = [{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u6211\u5f88\u597d\uff0c\u60a8\u597d\u5417\uff1f\"<\/span>}],<\/span>\n stop<\/span> = [\"\u597d\"<\/span>]        <\/span>\n)<\/span>\n<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"message\"]<\/span>[\"content\"]<\/span>)<\/span>\nprint(reply[\"choices\"]<\/span>[0]<\/span>[\"finish_reason\"]<\/span>)<\/span>\n<\/span>\n<\/span>\n<\/span>\n openai>=1.0.0 \u65f6<\/span>\n<\/span>\nreply<\/span> = client.chat.completions.create(<\/span>\n  model<\/span>=\"gpt-3.5-turbo\"<\/span>,<\/span>\n  messages<\/span>=[{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u6211\u5f88\u597d\uff0c\u60a8\u597d\u5417\uff1f\"<\/span>}],<\/span>\n  stop<\/span>=['\u597d'<\/span>]<\/span>\n)<\/span>\n<\/span>\nprint(reply.choices[0]<\/span>.message.content)<\/span>\nprint(reply.choices[0]<\/span>.finish_reason)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u6211\u4eec\u5c06stop\u53c2\u6570\u8bbe\u5b9a\u4e3a\u300c\u597d\u300d\uff08 stop = [\u201c\u597d\u201d]<\/code> \uff09\u3002\u8fd9\u6837\u4e00\u6765\uff0c\u5982\u679cAI\u7684\u56de\u7b54\u4e2d\u5305\u542b\u4e86\u300c\u597d\u300d\u8fd9\u4e2a\u5b57\u8bcd\uff0c\u5b83\u5c31\u4f1a\u7acb\u5373\u505c\u6b62\u56de\u7b54\u3002\u8fd9\u4e2a\u529f\u80fd\u5bf9\u4e8e\u63a7\u5236AI\u56de\u7b54\u7684\u5185\u5bb9\u975e\u5e38\u6709\u7528\uff0c\u53ef\u4ee5\u786e\u4fdd\u4e0d\u4f1a\u51fa\u73b0\u4e0d\u9002\u5f53\u6216\u4e0d\u60f3\u770b\u5230\u7684\u8bcd\u6c47\u3002<\/p>\n
\u5982\u4f55\u901a\u8fc7\u300c\u6e29\u5ea6\u300d\u8bbe\u5b9a\u8ba9OpenAI\u56de\u7b54\u66f4\u6709\u8da3<\/h2>\n
\n
temperature\u53c2\u6570\uff0c\u53ef\u589e\u52a0\u56de\u7b54\u7684\u591a\u6837\u6027<\/p>\n<\/blockquote>\n
\u300ctemperature\u300d\u8fd9\u4e2a\u53c2\u6570\u5c31\u50cf\u662f\u8c03\u8282OpenAI\u56de\u7b54\u7684\u300c\u6e29\u5ea6\u300d\u8bbe\u5b9a\u5668\u3002\u5f53\u6211\u4eec\u8bb2\u6e29\u5ea6\u65f6\uff0c\u9ad8\u6e29\u901a\u5e38\u610f\u5473\u8457\u66f4\u591a\u6d3b\u529b\u548c\u70ed\u60c5\uff0c\u800c\u4f4e\u6e29\u5219\u610f\u5473\u8457\u51b7\u9759\u548c\u7a33\u5b9a\u3002\u8fd9\u4e2a\u6982\u5ff5\u4e5f\u9002\u7528\u4e8eOpenAI\u7684\u8bbe\u5b9a\u3002\u901a\u8fc7\u8c03\u6574 temperature<\/code> \uff0c\u6211\u4eec\u53ef\u4ee5\u63a7\u5236AI\u56de\u7b54\u7684\u968f\u673a\u6027\u548c\u521b\u9020\u6027\u3002\u8fd9\u4e2a\u53c2\u6570\u7684\u8303\u56f4\u4ece0\u52302\uff0c\u6570\u503c\u8d8a\u9ad8\uff0c\u56de\u7b54\u5c31\u8d8a\u591a\u53d8\u548c\u6709\u8da3\uff1b\u6570\u503c\u8d8a\u4f4e\uff0c\u56de\u7b54\u5c31\u8d8a\u7a33\u5b9a\u548c\u53ef\u9884\u6d4b\u3002<\/p>\n
\u8bbe\u5b9a\u4e3a\u6700\u4f4e\u503c\uff1a0<\/h3>\n
\u5f53\u6211\u4eec\u628a temperature<\/code> \u8bbe\u4e3a0\u65f6\uff1a<\/p>\n
<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>reply<\/span> = client.chat.completions.create(<\/span>\nmodel<\/span>=\"gpt-3.5-turbo\"<\/span>,<\/span>\nmessages<\/span>=[{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u55e8\uff01\u6211\u5f88\u597d\uff0c\u60a8\u597d\u5417\uff1f\"<\/span>}],<\/span>\ntemperature<\/span>=0<\/span>,<\/span>\nn<\/span>=2<\/span><\/span>\n)<\/span>\nfor choice in reply.choices:<\/span>\n print(choice.index, choice.message.content)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u8fd9\u6837\u8bbe\u5b9a\u540e\uff0c\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/p>\n
\n
\"\u56fe\u7247\"<\/figure>\n<\/p>\n
AI\u4ea7\u751f\u7684\u56de\u7b54\u4f1a\u975e\u5e38\u7a33\u5b9a\u548c\u4e00\u81f4\uff0c\u51e0\u4e4e\u6ca1\u6709\u592a\u591a\u53d8\u5316\u3002<\/p>\n
\u8bbe\u5b9a\u4e3a\u6700\u9ad8\u503c\uff1a2<\/h3>\n
\u53cd\u4e4b\uff0c\u5982\u679c\u5c06 temperature<\/code> \u8bbe\u4e3a2\uff1a<\/p>\n
<\/div>
ini<\/span>
\u590d\u5236\u4ee3\u7801<\/div><\/div><\/div>reply<\/span> = client.chat.completions.create(<\/span>\nmodel<\/span>=\"gpt-3.5-turbo\"<\/span>,<\/span>\nmessages<\/span>=[{\"role\"<\/span>: \"user\"<\/span>, \"content\"<\/span>: \"\u55e8\uff01\u6211\u5f88\u597d\uff0c\u60a8\u597d\u5417\uff1f\"<\/span>}],<\/span>\ntemperature<\/span>=2<\/span>,<\/span>\nn<\/span>=2<\/span><\/span>\n)<\/span>\nfor choice in reply.choices:<\/span>\n print(choice.index, choice.message.content)<\/span>\n<\/span>\n<\/code><\/pre>\n
\u5728\u8fd9\u4e2a\u8bbe\u5b9a\u4e0b\uff0cAI\u4ea7\u751f\u7684\u56de\u7b54\u4f1a\u975e\u5e38\u6d3b\u6cfc\u548c\u591a\u53d8\uff0c\u6709\u65f6\u751a\u81f3\u53ef\u80fd\u663e\u5f97\u6709\u70b9\u5947\u602a\u6216\u4e0d\u76f8\u5173\uff08\u5982\u4e0b\u56fe\u663e\u793a\uff09\u3002\u800c\u4e14\uff0c\u8fd9\u53ef\u80fd\u4f1a\u5bfc\u81f4AI\u9700\u8981\u66f4\u957f\u7684\u65f6\u95f4\u6765\u751f\u6210\u56de\u7b54\u3002<\/p>\n
\n
\"\u56fe\u7247\"<\/figure>\n<\/p>\n
\u5b9e\u9645\u4e0a\uff0c temperature<\/code> \u7684\u9ed8\u8ba4\u503c\u662f1\u3002\u5728\u8fd9\u4e2a\u8bbe\u5b9a\u4e0b\u7684\u56de\u7b54\u65e2\u4e0d\u4f1a\u592a\u50f5\u786c\uff0c\u4e5f\u4e0d\u4f1a\u592a\u8fc7\u6d3b\u8dc3\uff0c\u662f\u4e00\u4e2a\u8f83\u4e3a\u5747\u8861\u7684\u9009\u62e9\u3002\u56e0\u6b64\uff0c\u5982\u679c\u4f60\u4e0d\u786e\u5b9a\u4ece\u54ea\u91cc\u5f00\u59cb\uff0c\u53ef\u4ee5\u5148\u5c1d\u8bd5\u8bbe\u5b9a temperature<\/code> \u4e3a1\u5f00\u59cb\u6d4b\u8bd5\u3002<\/p>\n
\u7ed3\u8bed<\/h2>\n
\u900f\u8fc7\u672c\u6587\u7684\u4ecb\u7ecd\uff0c\u6211\u4eec\u4e00\u8d77\u63a2\u7d22\u4e86\u5982\u4f55\u63a7\u5236OpenAI\u6a21\u578b\u7684\u51e0\u4e2a\u91cd\u8981\u8bbe\u5b9a\u3002\u4ece\u8bbe\u5b9a\u751f\u6210\u56de\u590d\u6570\u91cf\u7684 n<\/code> \u53c2\u6570\uff0c\u5230\u63a7\u5236\u56de\u7b54\u957f\u5ea6\u7684 max_tokens<\/code> \uff0c\u518d\u5230\u5efa\u7acb\u8bcd\u6c47\u9ed1\u540d\u5355\u7684 stop<\/code> \u53c2\u6570\uff0c\u4ee5\u53ca\u8c03\u8282\u56de\u7b54\u98ce\u683c\u7684 temperature<\/code> \uff0c\u8fd9\u4e9b\u90fd\u662f\u4f7f\u6211\u4eec\u80fd\u591f\u66f4\u6709\u6548\u5730\u5229\u7528AI\u6280\u672f\u7684\u5f3a\u5927\u5de5\u5177\u3002\u8fd9\u4e9b\u8bbe\u5b9a\uff0c\u8ba9\u6211\u4eec\u53ef\u4ee5\u8c03\u6574\u4e0e\u63a7\u5236AI\u7684\u56de\u7b54\uff0c\u4f7f\u4e4b\u66f4\u8d34\u5408\u6211\u4eec\u7684\u9700\u6c42\u548c\u9884\u671f\uff0c\u65e0\u8bba\u662f\u5728\u8282\u7701\u6210\u672c\u3001\u907f\u514d\u4e0d\u9002\u5f53\u5185\u5bb9\uff0c\u8fd8\u662f\u521b\u9020\u66f4\u6709\u8da3\u548c\u591a\u6837\u7684\u56de\u7b54\u4e0a\u90fd\u5927\u6709\u88e8\u76ca\u3002<\/p>\n
\u5e0c\u671b\u8fd9\u7bc7\u6587\u7ae0\u80fd\u591f\u5e2e\u52a9\u60a8\u5bf9OpenAI\u7684\u529f\u80fd\u6709\u4e86\u66f4\u6df1\u5165\u7684\u7406\u89e3\uff0c\u65e0\u8bba\u76ee\u524d\u7684\u60a8\u662fAI\u9886\u57df\u7684\u521d\u5b66\u8005\u8fd8\u662f\u5df2\u7ecf\u6709\u4e00\u5b9a\u7ecf\u9a8c\u7684\u4f7f\u7528\u8005\u3002\u5728\u8fd9\u4e2a\u4e0d\u65ad\u53d8\u5316\u7684AI\u9886\u57df\uff0c\u4fdd\u6301\u5b66\u4e60\u548c\u5b9e\u9a8c\u7684\u5fc3\u662f\u975e\u5e38\u91cd\u8981\u7684\u3002\u968f\u8457\u6280\u672f\u7684\u8fdb\u6b65\uff0c\u4f60\u6211\u5c06\u4f1a\u770b\u5230\u66f4\u591a\u4ee4\u4eba\u53f9\u4e3a\u89c2\u6b62\u7684\u53ef\u80fd\u6027\u3002\u671f\u5f85AI\u5728\u672a\u6765\u7684\u53d1\u5c55\uff0c\u5e76\u79ef\u6781\u53c2\u4e0e\u5176\u4e2d\uff0c\u5171\u540c\u5851\u9020\u4e00\u4e2a\u66f4\u667a\u80fd\u3001\u66f4\u6709\u6548\u7684\u672a\u6765\u3002<\/p>","protected":false},"excerpt":{"rendered":"
\u5728\u8fd9\u4e2a\u5feb\u901f\u53d1\u5c55\u7684\u6570\u5b57\u65f6\u4ee3\uff0cAI\u4eba\u5de5\u667a\u6167\u5df2\u7ecf\u6e10\u6e10\u6210\u4e3a\u6211\u4eec\u65e5\u5e38\u751f\u6d3b\u7684\u4e00\u90e8\u5206\u3002\u800cOpenAI\uff0c\u4f5c\u4e3aAI\u7814\u7a76\u7684\u673a\u6784\u4e4b\u4e00\uff0c\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u5de5\u5177\u548cLLM\u6a21\u578b\uff0c\u5982GPT-3.5\u3001GPT-4\uff0c\u6765\u5e2e\u52a9\u6211\u4eec\u66f4\u597d\u5730\u7406\u89e3\u548c\u4f7f\u7528A<\/p>\n","protected":false},"author":1,"featured_media":11146,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_title":"","rank_math_description":"","rank_math_focus_keyword":"","views":"11","footnotes":""},"categories":[3],"tags":[129,136,126,127,128],"collection":[],"class_list":["post-1563","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-fenlei2","tag-129","tag-136","tag-gpt","tag-ai","tag-128"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1563","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/comments?post=1563"}],"version-history":[{"count":0,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/posts\/1563\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media\/11146"}],"wp:attachment":[{"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/media?parent=1563"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/categories?post=1563"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/tags?post=1563"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.nicekj.com\/nicekj2024\/wp\/v2\/collection?post=1563"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}