{"id":10601,"date":"2024-02-06T14:30:02","date_gmt":"2024-02-06T05:30:02","guid":{"rendered":"https:\/\/blogs.mathworks.com\/japan-community\/?p=10601"},"modified":"2024-08-09T23:47:49","modified_gmt":"2024-08-09T14:47:49","slug":"large-language-models-with-matlab-jp","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/japan-community\/2024\/02\/06\/large-language-models-with-matlab-jp\/","title":{"rendered":"MATLAB \u3067\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb"},"content":{"rendered":"

\u203b\u3053\u306e\u6295\u7a3f\u306f 2024 \u5e74 1 \u6708 22 \u65e5\u306bArtificial Intelligence \u3078 \u6295\u7a3f<\/a>\u3055\u308c\u305f\u3082\u306e\u306e\u6284\u8a33\u3067\u3059\u3002<\/p>\n

\n

MATLAB \u3092 OpenAI\u2122 API \u306b\u63a5\u7d9a\u3057\u3066\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\u3092\u884c\u3046\u65b9\u6cd5<\/strong><\/p>\n

ChatGPT\u2122\u3001Generative AI\u3001\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM, Large Language Model\uff09\u3001\u3053\u308c\u3089\u306e\u9769\u65b0\u7684\u306a\u6280\u8853\u3092 MATLAB \u3068\u7d44\u307f\u5408\u308f\u305b\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u3053\u3068\u3092\u3054\u5b58\u3058\u3067\u3057\u3087\u3046\u304b\uff1f\u30d6\u30e9\u30a6\u30b6\u3067\u6a5f\u80fd\u3059\u308b MATLAB AI Chat Playground<\/a>\uff08\u95a2\u9023\u30d6\u30ed\u30b0\u6295\u7a3f<\/a>\uff09\u306b\u52a0\u3048\u3066\u3001MATLAB \u3092 OpenAI\u2122 Chat Completions API \u306b\u3082\u63a5\u7d9a\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n

\u3053\u306e\u30d6\u30ed\u30b0\u3067\u306f\u3001LLM \u306e\u80cc\u5f8c\u306b\u3042\u308b\u6280\u8853\u3068\u3001MATLAB \u3092 OpenAI API \u306b\u63a5\u7d9a\u3059\u308b\u65b9\u6cd5\u306b\u3064\u3044\u3066\u7d39\u4ecb\u3057\u307e\u3059\u3002\u307e\u305f\u3001Text Analytics Toolbox<\/a> \u3092\u6d3b\u7528\u3057\u3066\u3001\u611f\u60c5\u5206\u6790\u3084\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u69cb\u7bc9\u306a\u3069\u306e\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\uff08NLP\uff09\u30bf\u30b9\u30af<\/a>\u3092\u5b9f\u884c\u3059\u308b\u65b9\u6cd5\u3082\u3054\u7d39\u4ecb\u3057\u307e\u3059\u3002<\/p>\n

 <\/p>\n

LLM \u3063\u3066\u306a\u306b?<\/h1>\n

\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM\uff09\u306f\u3001\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u30e2\u30c7\u30eb\uff08\u30c7\u30a3\u30fc\u30d7\u30e9\u30fc\u30cb\u30f3\u30b0\u30e2\u30c7\u30eb\u306e\u7279\u5225\u306a\u30b1\u30fc\u30b9\uff09\u306b\u57fa\u3065\u3044\u3066\u3044\u307e\u3059\u3002\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u306f\u3001\u9023\u7d9a\u3057\u305f\u30c7\u30fc\u30bf\u5185\u306e\u95a2\u4fc2\u6027\u3092\u8ffd\u8de1\u3059\u308b\u3088\u3046\u306b\u8a2d\u8a08\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u305d\u308c\u3089\u306f\u81ea\u5df1\u6ce8\u610f\u30e1\u30ab\u30cb\u30ba\u30e0\u306b\u4f9d\u5b58\u3057\u3066\u3001\u5165\u529b\u3068\u51fa\u529b\u306e\u9593\u306e\u30b0\u30ed\u30fc\u30d0\u30eb\u306a\u4f9d\u5b58\u95a2\u4fc2\u3092\u6349\u3048\u307e\u3059\u3002LLM \u306f NLP \u306b\u9769\u547d\u3092\u3082\u305f\u3089\u3057\u307e\u3057\u305f\u3002\u306a\u305c\u306a\u3089\u3001\u4eba\u9593\u306e\u8a00\u8a9e\u306b\u5b58\u5728\u3059\u308b\u5358\u8a9e\u9593\u306e\u8907\u96d1\u306a\u95a2\u4fc2\u3084\u30cb\u30e5\u30a2\u30f3\u30b9\u3092\u6349\u3048\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u304b\u3089\u3067\u3059\u3002<\/p>\n

\u3088\u304f\u77e5\u3089\u308c\u3066\u3044\u308b\u30c8\u30e9\u30f3\u30b9\u30d5\u30a9\u30fc\u30de\u30fc\u30e2\u30c7\u30eb\u306b\u306f\u3001BERT \u3068 GPT \u306e\u30e2\u30c7\u30eb\u304c\u3042\u308a\u3001\u3069\u3061\u3089\u3082 MATLAB \u3067\u4f7f\u7528\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002MATLAB \u306b\u542b\u307e\u308c\u3066\u3044\u308b\u4e8b\u524d\u8a13\u7df4\u6e08\u307f\u306e BERT \u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3057\u305f\u3044\u5834\u5408\u306f\u3001bert<\/a> \u95a2\u6570\u3092\u4f7f\u7528\u3057\u307e\u3059\u3002\u3053\u306e\u30d6\u30ed\u30b0\u3067\u306f\u3001GPT \u30e2\u30c7\u30eb\u306b\u7126\u70b9\u3092\u5f53\u3066\u307e\u3059\u3002<\/p>\n

LLM\u30ea\u30dd\u30b8\u30c8\u30ea<\/h1>\n

MATLAB \u304b\u3089 LLM \u306b\u30a2\u30af\u30bb\u30b9\u3059\u308b\u305f\u3081\u306b\u5fc5\u8981\u306a\u30b3\u30fc\u30c9\u306f GitHub \u306e LLM \u30ea\u30dd\u30b8\u30c8\u30ea<\/a>\u306b\u3042\u308a\u307e\u3059\u3002\u3053\u306e\u30b3\u30fc\u30c9\u3092\u4f7f\u7528\u3057\u3066 MATLAB \u74b0\u5883\u304b\u3089ChatGPT API \u3092\u64cd\u4f5c\u3067\u304d\u307e\u3059\u3002gpt-3.5-turbo \u3084 gpt-4 \u3082\u30b5\u30dd\u30fc\u30c8\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n

\"GitHub<\/p>\n

\u30bb\u30c3\u30c8\u30a2\u30c3\u30d7<\/strong><\/p>\n

ChatGPT API \u3092\u64cd\u4f5c\u3059\u308b\u305f\u3081\u306b\u306f\u3001OpenAI API \u30ad\u30fc\u3092\u53d6\u5f97\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002API \u30ad\u30fc\u306e\u53d6\u5f97\u65b9\u6cd5\u3068 OpenAI API \u306e\u4f7f\u7528\u6599\u91d1\u306b\u3064\u3044\u3066\u8a73\u3057\u304f\u306f\u3001OpenAI API<\/a>\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\u3002\u73fe\u5728\u306e\u30d5\u30a9\u30eb\u30c0\u306b API \u30ad\u30fc\u3092\u30d5\u30a1\u30a4\u30eb\u3068\u3057\u3066\u4fdd\u5b58\u3057\u3066\u304a\u304f\u3068\u4fbf\u5229\u3067\u3059\u3002<\/p>\n

\"How

Save the OpenAI API key in your current folder.<\/p><\/div><\/p>\n

 <\/p>\n

Getting Started<\/strong><\/p>\n

MATLAB \u3067 LLM \u3092\u4f7f\u7528\u3057\u59cb\u3081\u308b\u305f\u3081\u306b\u3001\u307e\u305a OpenAI Chat \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u3092\u521d\u671f\u5316\u3057\u307e\u3059\u3002<\/p>\n

chat = openAIChat(systemPrompt,ApiKey=my_key);\r\n<\/pre>\n
\u6b21\u306e\u30bb\u30af\u30b7\u30e7\u30f3\u3067\u306f\u3001\u3055\u307e\u3056\u307e\u306a\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u306b\u5bfe\u3057\u3066\u30b7\u30b9\u30c6\u30e0\u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u3069\u306e\u3088\u3046\u306b\u6307\u5b9a\u3059\u308b\u304b\u3001\u304a\u3088\u3073\u30aa\u30d7\u30b7\u30e7\u30f3\u306e\u540d\u524d-\u5024\u5f15\u6570\u3092\u4f7f\u7528\u3057\u3066 OpenAI Chat \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u306e\u6a5f\u80fd\u3092\u5f37\u5316\u3059\u308b\u65b9\u6cd5\u3092\u7d39\u4ecb\u3057\u307e\u3059\u3002<\/p>\n
MATLAB Online \u3067\u59cb\u3081\u308b<\/strong><\/p>\n
MATLAB Online \u3067 LLM \u3092\u4f7f\u3044\u305f\u3044\u5834\u5408\u3082\u3042\u308b\u304b\u3082\u3057\u308c\u307e\u305b\u3093\u3002MATLAB \u30b3\u30fc\u30c9\u304c\u542b\u307e\u308c\u308b GitHub \u30ea\u30dd\u30b8\u30c8\u30ea\u306b\u306f\u300cOpen in MATLAB Online\u300d\u30dc\u30bf\u30f3\u304c\u3042\u308a\u307e\u3059\u3002\u3053\u306e\u30dc\u30bf\u30f3\u3092\u30af\u30ea\u30c3\u30af\u3059\u308b\u3068\u3001\u30ea\u30dd\u30b8\u30c8\u30ea\u304c\u76f4\u63a5 MATLAB Online \u3067\u958b\u304d\u307e\u3059\u3002\u4ee5\u4e0b\u306e\u30d3\u30c7\u30aa\u3092\u3054\u89a7\u306b\u306a\u3063\u3066\u300130\u79d2\u4ee5\u5185\u3067 LLM \u30ea\u30dd\u30b8\u30c8\u30ea\u3092 MATLAB Online \u3067\u958b\u304f\u65b9\u6cd5\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n
\"Open
Use LLMs in MATLAB Online.<\/p><\/div><\/p>\n
\u4f7f\u7528\u4f8b<\/h1>\n
\u3053\u306e\u30bb\u30af\u30b7\u30e7\u30f3\u3067\u306f\u3001MATLAB \u3067\u306e LLM \u306e\u4f7f\u7528\u4f8b\u3092\u7d39\u4ecb\u3057\u3001\u95a2\u9023\u30ea\u30f3\u30af\u3092\u8a18\u8f09\u3057\u307e\u3059\u3002\u4f7f\u7528\u4f8b\u306b\u306f\u611f\u60c5\u5206\u6790\u3001\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u69cb\u7bc9\u3001RAG (Retrieval Augmented Generation) \u306a\u3069\u304c\u3042\u308a\u307e\u3059\u3002Text Analytics Toolbox \u3092\u4f7f\u7528\u3057\u3066\u3001\u30c6\u30ad\u30b9\u30c8\u3092\u524d\u51e6\u7406\u3057\u3001\u5206\u6790\u3057\u3001\u305d\u3057\u3066\u300c\u610f\u5473\u306e\u3042\u308b\u300d\u5f62\u3067\u8868\u793a\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002\u3053\u3053\u3067\u306f\u3044\u304f\u3064\u304b\u306e\u95a2\u6570\u3092\u7d39\u4ecb\u3057\u307e\u3059\u304c\u3001\u8a73\u7d30\u306f\u30ea\u30f3\u30af\u5148\u3092\u30c1\u30a7\u30c3\u30af\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n
\u611f\u60c5\u5206\u6790<\/strong><\/p>\n
\u611f\u60c5\u5206\u6790\u3092\u884c\u3046\u7c21\u5358\u306a\u4f8b\u304b\u3089\u59cb\u3081\u307e\u3057\u3087\u3046\u3002\u611f\u60c5\u5206\u6790\u306f\u3001\u30c6\u30ad\u30b9\u30c8\u4e2d\u306e\u610f\u898b\u3084\u611f\u60c5\u306e\u5206\u985e\u306b\u53d6\u308a\u7d44\u3080\u3082\u306e\u3067\u3059\u3002\u30c6\u30ad\u30b9\u30c8\u306e\u611f\u60c5\u7684\u306a\u30c8\u30fc\u30f3\u306f\u3001\u30dd\u30b8\u30c6\u30a3\u30d6\u3001\u30cd\u30ac\u30c6\u30a3\u30d6\u3001\u307e\u305f\u306f\u30cb\u30e5\u30fc\u30c8\u30e9\u30eb\u3068\u3057\u3066\u5206\u985e\u3055\u308c\u308b\u3053\u3068\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n
\"Sentiment
Creating a sentiment analysis classifier.<\/p><\/div><\/p>\n
<\/h6>\n
\u307e\u305a\u30b7\u30b9\u30c6\u30e0\u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u6307\u5b9a\u3057\u307e\u3059\u3002\u30b7\u30b9\u30c6\u30e0\u30d7\u30ed\u30f3\u30d7\u30c8\u306f\u3001\u30a2\u30b7\u30b9\u30bf\u30f3\u30c8\u306b\u3069\u306e\u3088\u3046\u306b\u632f\u308b\u821e\u3046\u304b\u3092\u4f1d\u3048\u307e\u3059\u3002\u3053\u306e\u5834\u5408\u3001\u611f\u60c5\u5206\u6790\u5668\u3068\u3057\u3066\u632f\u308b\u821e\u3044\u307e\u3059\u3002\u307e\u305f\u3001\u30b7\u30b9\u30c6\u30e0\u306b\u611f\u60c5\u5206\u6790\u3092\u884c\u3046\u65b9\u6cd5\u306e\u7c21\u5358\u306a\u4f8b\u3082\u63d0\u4f9b\u3057\u307e\u3059\u3002<\/p>\n
systemPrompt = \"You are a sentiment analyser. You will look at a sentence and output\"+...\r\n    \" a single word that classifies that sentence as either 'positive' or 'negative'.\"+....\r\n    \"Examples: \\n\"+...\r\n    \"The project was a complete failure. \\n\"+...\r\n    \"negative \\n\\n\"+...  \r\n    \"The team successfully completed the project ahead of schedule.\"+...\r\n    \"positive \\n\\n\"+...\r\n    \"His attitude was terribly discouraging to the team. \\n\"+...\r\n    \"negative \\n\\n\";\r\n<\/pre>\n
<\/h6>\n
\u30b7\u30b9\u30c6\u30e0\u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u6e21\u3057\u3066 OpenAI Chat \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u3092\u521d\u671f\u5316\u3057\u307e\u3059\u3002<\/p>\n
chat = openAIChat(systemPrompt,ApiKey=my_key);<\/pre>\n
\u65b0\u3057\u3044\u6587\u3092\u6e21\u3057\u3066\u5206\u985e\u3055\u305b\u307e\u3059\u3002<\/p>\n
text = generate(chat,\"The team is feeling very motivated.\")<\/pre>\n
\u00a0 text = “positive”<\/span><\/p>\n
\u6b63\u3057\u304f\u30dd\u30b8\u30c6\u30a3\u30d6\u306a\u611f\u60c5\u3068\u3057\u3066\u5206\u985e\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n
 <\/p>\n
\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u69cb\u7bc9<\/strong><\/p>\n
\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306f\u4eba\u9593\u306e\u4f1a\u8a71\u3092\u30b7\u30df\u30e5\u30ec\u30fc\u30c8\u3057\u307e\u3059\u3002\u3064\u307e\u308a\u30e6\u30fc\u30b6\u30fc\u304c\u30af\u30a8\u30ea\u3092\u5165\u529b\u3057\u3001\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u304c\u81ea\u7136\u306a\u4eba\u9593\u306e\u8a00\u8a9e\u3067\u5fdc\u7b54\u3092\u751f\u6210\u3057\u307e\u3059\u3002<\/p>\n
<\/h6>\n
\"Building
Building a chatbot.<\/p><\/div><\/p>\n
LLM\u30ea\u30dd\u30b8\u30c8\u30ea\u306e\u300cExample: Build ChatBot<\/a>\u300d\u3092\u53c2\u8003\u306b\u3001\u6570\u5206\u3067\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u3092\u69cb\u7bc9\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u3092\u69cb\u7bc9\u3059\u308b\u6700\u521d\u306e2\u3064\u306e\u30b9\u30c6\u30c3\u30d7\u306f\u3001(1) \u30c1\u30e3\u30c3\u30c8\u3092\u5b9f\u884c\u3059\u308b\u305f\u3081\u306b openAIChat \u306e\u30a4\u30f3\u30b9\u30bf\u30f3\u30b9\u3092\u4f5c\u6210\u3059\u308b\u3053\u3068\u3068\u3001(2) \u4f1a\u8a71\u5c65\u6b74\u3092\u4fdd\u5b58\u3059\u308b\u305f\u3081\u306b openAIMessages \u95a2\u6570\u3092\u4f7f\u7528\u3059\u308b\u3053\u3068\u3067\u3059\u3002<\/p>\n
chat = openAIChat(\"You are a helpful assistant. You reply in a very concise way, keeping answers \"+...\r\n    \"limited to short sentences.\",ModelName=modelName,ApiKey=my_key);\r\nmessages = openAIMessages;\r\n<\/pre>\n
\u3055\u3089\u306b\u6570\u884c\u306e\u30b3\u30fc\u30c9\u3092\u8ffd\u52a0\u3057\u305f\u5f8c\u3001\u30e1\u30ad\u30b7\u30b3\u3067\u306e\u4f11\u6687\u8a08\u753b\u306b\u5f79\u7acb\u3064\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u3092\u69cb\u7bc9\u3057\u307e\u3057\u305f\u3002\u4ed6\u306e MATLAB \u95a2\u6570\uff08\u4f8b\u3048\u3070\u3001extractBetween<\/a>\uff09\u3092\u4f7f\u7528\u3057\u3066\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u5fdc\u7b54\u3092\u30d5\u30a9\u30fc\u30de\u30c3\u30c8\u3057\u307e\u3057\u305f\u3002\u4ee5\u4e0b\u306e\u56f3\u306f\u3001\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u3068\u306e\u7c21\u6f54\u3067\u3059\u304c\u5f79\u7acb\u3064\u30c1\u30e3\u30c3\u30c8\u3092\u793a\u3057\u3066\u3044\u307e\u3059\u3002\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306f\u4ee5\u524d\u306e\u30af\u30a8\u30ea\u304b\u3089\u306e\u60c5\u5831\u3092\u4fdd\u6301\u3057\u3066\u3044\u308b\u3053\u3068\u306b\u6ce8\u610f\u3057\u3066\u304f\u3060\u3055\u3044\u3002\u79c1\u306f\u8cea\u554f\u3067\u300c\u30e6\u30ab\u30bf\u30f3\u534a\u5cf6\u300d\u3092\u7e70\u308a\u8fd4\u3059\u5fc5\u8981\u306f\u3042\u308a\u307e\u305b\u3093\u3002<\/p>\n
<\/h6>\n
\"Conversation
User queries and chatbot responses for planning a Mexico vacation.<\/p><\/div><\/p>\n
 <\/p>\n
Retrieval Augmented Generation<\/strong><\/p>\n
Retrieval-augmented generation (RAG\uff09\u306f\u3001LLM \u306b\u3088\u3063\u3066\u5f97\u3089\u308c\u305f\u7d50\u679c\u3092\u5411\u4e0a\u3055\u305b\u308b\u305f\u3081\u306e\u6280\u8853\u3067\u3059\u3002\u5916\u90e8\u30bd\u30fc\u30b9\u304b\u3089\u60c5\u5831\u3092\u53d6\u5f97\u3059\u308b\u3053\u3068\u3067\u3001\u6b63\u78ba\u6027\u3068\u4fe1\u983c\u6027\u306e\u4e21\u65b9\u3092\u5f37\u5316\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002\u4f8b\u3048\u3070\u3001LLM \u306b\u4f9b\u7d66\u3055\u308c\u308b\u30d7\u30ed\u30f3\u30d7\u30c8\u306f\u3001\u3088\u308a\u6700\u65b0\u306e\u60c5\u5831\u3084\u6280\u8853\u7684\u306a\u60c5\u5831\u3067\u5f37\u5316\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n
<\/h6>\n
\"Workflow
Workflow for retrieval-augmented generation (RAG).<\/p><\/div><\/p>\n
 <\/p>\n
Example: Retrieval-Augmented Generation<\/a> \u3067\u306f\u3001\u96fb\u529b\u30b7\u30b9\u30c6\u30e0\u306b\u95a2\u3059\u308b\u6280\u8853\u5831\u544a\u66f8\u304b\u3089\u60c5\u5831\u3092\u53d6\u5f97\u3057\u3001\u6280\u8853\u7684\u306a\u554f\u3044\u5408\u308f\u305b\u306e\u305f\u3081\u306b ChatGPT \u3092\u5f37\u5316\u3059\u308b\u65b9\u6cd5\u3092\u793a\u3057\u3066\u3044\u307e\u3059\u3002\u3053\u3053\u3067\u306f\u3059\u3079\u3066\u306e\u8a73\u7d30\u3092\u793a\u3057\u3066\u3044\u307e\u305b\u3093\u304c\u3001\u91cd\u8981\u306a\u30b9\u30c6\u30c3\u30d7\u3092\u3044\u304f\u3064\u304b\u5f37\u8abf\u3057\u307e\u3059\u3002<\/p>\n
    \n
  1. MATLAB \u95a2\u6570 (e.g., websave<\/a>\u00a0\u3084\u00a0fileDatastore<\/a>) \u30aa\u30f3\u30e9\u30a4\u30f3\u6587\u66f8\u306e\u53d6\u5f97\u304a\u3088\u3073\u6574\u7406\u3057\u307e\u3059\u3002<\/li>\n
  2. Text Analytics Toolbox \u306e\u95a2\u6570 (e.g., splitParagraphs<\/a>, tokenizedDocument<\/a>, \u3084\u00a0bm25Similarity<\/a>) \u53d6\u5f97\u3057\u305f\u6587\u66f8\u304b\u3089\u4f7f\u7528\u3059\u308b\u30c6\u30ad\u30b9\u30c8\u3092\u6e96\u5099\u3057\u307e\u3059\u3002<\/li>\n
  3. \u30c6\u30ad\u30b9\u30c8\u306e\u6e96\u5099\u304c\u3067\u304d\u305f\u3089\u3001\u6307\u5b9a\u3055\u308c\u305f\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u3068 API \u30ad\u30fc\u3067\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u3092\u521d\u671f\u5316\u3057\u307e\u3059\u3002\n
    chat = openAIChat(\"You are a helpful assistant. You will get a \" + ...\r\n    \"context for each question, but only use the information \" + ...\r\n    \"in the context if that makes sense to answer the question. \" + ...\r\n    \"Let's think step-by-step, explaining how you reached the answer.\",ApiKey=my_key);\r\n<\/pre>\n<\/li>\n
  4. \u30af\u30a8\u30ea\u3092\u5b9a\u7fa9\u3057\u307e\u3059\u3002\u6b21\u306b\u3001\u30af\u30a8\u30ea\u306b\u57fa\u3065\u3044\u3066\u95a2\u9023\u6587\u66f8\u3092\u53d6\u5f97\u304a\u3088\u3073\u30d5\u30a3\u30eb\u30bf\u30ea\u30f3\u30b0\u3057\u307e\u3059\u3002\n
    query = \"What technical criteria can be used to streamline new approvals for grid-friendly DPV?\";\r\nselectedDocs = retrieveAndFilterRelevantDocs(allDocs,query);\r\n<\/pre>\n<\/li>\n
  5. \u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u5b9a\u7fa9\u3057\u3001\u5fdc\u7b54\u3092\u751f\u6210\u3057\u307e\u3059\u3002\n
    prompt = \"Context:\" ...\r\n    + join(selectedDocs, \" \") + newline +\"Answer the following question: \"+ query;\r\nresponse = generate(chat,prompt);\r\n<\/pre>\n
    \u53ef\u8996\u5316\u3057\u3084\u3059\u3044\u3088\u3046\u306b\u30c6\u30ad\u30b9\u30c8\u3092\u6574\u5f62\u3057\u307e\u3059\u3002<\/p>\n
    wrapText(response)\r\n<\/pre>\n
    ans =<\/span><\/p>\n

    \n“The technical criteria that can be used to streamline new approvals for grid-friendly DPV can include prudent screening criteria for systems that meet certain specifications.
    \nThese criteria can be based on factors such as DPV capacity penetration relative to minimum feeder daytime load.
    \nAdditionally, hosting capacity calculations can be used to estimate the point where DPV would induce technical impacts on system operations.
    \nThese screening criteria are commonly used in countries like India and the United States.”
    \n<\/span><\/p>\n<\/li>\n<\/ol>\n
    <\/h6>\n
    \u4ed6\u306e\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9<\/strong><\/p>\n
    \u4e0a\u8a18\u306b\u793a\u3057\u305f\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u306f\u3001LLM \u3092\u4f7f\u7528\u3057\u3066\u9054\u6210\u3067\u304d\u308b\u3053\u3068\u306e\u4e00\u4f8b\u306b\u904e\u304e\u307e\u305b\u3093\u3002\u4ed6\u306e\u6ce8\u76ee\u3059\u3079\u304d\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\uff08LLM\u30ea\u30dd\u30b8\u30c8\u30ea\u306e\u4f8b\u3092\u542b\u3080\uff09\u306b\u306f\u3001\u30c6\u30ad\u30b9\u30c8\u306e\u8981\u7d04\u3084\u95a2\u6570\u547c\u3073\u51fa\u3057\u304c\u3042\u308a\u307e\u3059\u3002\u307e\u305f\u3001LLM \u3092\u4f7f\u7528\u3057\u3066\u3001\u6a5f\u68b0\u7ffb\u8a33<\/a>\u306e\u3088\u3046\u306a\u4ed6\u306e\u591a\u304f\u306e\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\uff08NLP\uff09\u30bf\u30b9\u30af\u306b\u3082\u5229\u7528\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n
    MATLAB LLM\u30ea\u30dd\u30b8\u30c8\u30ea<\/a>\u3092\u4f55\u306b\u4f7f\u7528\u3057\u307e\u3059\u304b\uff1f\u4ee5\u4e0b\u306e\u30b3\u30e1\u30f3\u30c8\u6b04\u306b\u3054\u610f\u898b\u304f\u3060\u3055\u3044\u3002<\/em><\/p>\n
    \u30c6\u30ad\u30b9\u30c8\u8981\u7d04<\/strong>\u306f\u3001\u9577\u3044\u30c6\u30ad\u30b9\u30c8\u6587\u66f8\u304b\u3089\u77ed\u304f\u3001\u6b63\u78ba\u3067\u3001\u8aad\u307f\u3084\u3059\u3044\u8981\u7d04\u3092\u81ea\u52d5\u7684\u306b\u4f5c\u6210\u3059\u308b\u3053\u3068\u3067\u3059\u3002Example: Text Summarization<\/a>\u3067\u306f\u3001\u5927\u304d\u306a\u30c6\u30ad\u30b9\u30c8\u3092\u5c0f\u3055\u306a\u30c1\u30e3\u30f3\u30af\u306b\u5206\u5272\u3057\u3001\u30b9\u30c6\u30c3\u30d7\u30d0\u30a4\u30b9\u30c6\u30c3\u30d7\u3067\u5404\u30c1\u30e3\u30f3\u30af\u3092\u8981\u7d04\u3059\u308b\u65b9\u6cd5\u3092\u7d39\u4ecb\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n
    \u95a2\u6570\u547c\u3073\u51fa\u3057<\/strong>\u306f\u3001LLM \u306e NLP \u6a5f\u80fd\u3092\u5b9a\u7fa9\u3057\u305f\u95a2\u6570\u3068\u7d44\u307f\u5408\u308f\u305b\u308b\u3053\u3068\u304c\u3067\u304d\u308b\u5f37\u529b\u306a\u30c4\u30fc\u30eb\u3067\u3059\u3002\u3057\u304b\u3057\u3001ChatGPT \u306f\u95a2\u6570\u540d\u3092 “Hallucinate” \u3059\u308b\u3053\u3068\u304c\u3042\u308b\u306e\u3067\u3001\u4efb\u610f\u306b\u751f\u6210\u3055\u308c\u305f\u95a2\u6570\u3092\u5b9f\u884c\u3059\u308b\u3053\u3068\u306f\u907f\u3051\u3001\u5b9a\u7fa9\u3057\u305f\u95a2\u6570\u306e\u5b9f\u884c\u306e\u307f\u3092\u8a31\u53ef\u3057\u3066\u304f\u3060\u3055\u3044\u3002arXiv API \u304b\u3089\u5b66\u8853\u8ad6\u6587\u3092\u81ea\u52d5\u7684\u306b\u5206\u6790\u3059\u308b\u305f\u3081\u306b\u95a2\u6570\u547c\u3073\u51fa\u3057\u3092\u4f7f\u7528\u3059\u308b\u4f8b\u306b\u3064\u3044\u3066\u306f\u3001LLM\u3092\u4f7f\u7528\u3057\u305f\u95a2\u6570\u547c\u3073\u51fa\u3057<\/a>\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n
    Key Takeaways<\/h1>\n
      \n
    1. MATLAB \u3067\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\u30bf\u30b9\u30af\u3092\u5b9f\u884c\u3059\u308b\u305f\u3081\u306b GPT \u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3067\u304d\u308b\u65b0\u3057\u3044 GitHub \u30ea\u30dd\u30b8\u30c8\u30ea<\/a>\u304c\u3042\u308a\u307e\u3059\u3002<\/li>\n
    2. \u30ea\u30dd\u30b8\u30c8\u30ea\u306b\u306f\u3001\u611f\u60c5\u5206\u6790\u3084\u30c1\u30e3\u30c3\u30c8\u30dc\u30c3\u30c8\u306e\u69cb\u7bc9\u306a\u3069\u3001\u591a\u304f\u306e\u30bf\u30b9\u30af\u3092\u9054\u6210\u3059\u308b\u305f\u3081\u306e\u4f8b\u3068\u30e6\u30fc\u30b9\u30b1\u30fc\u30b9\u304c\u542b\u307e\u308c\u3066\u3044\u307e\u3059\u3002<\/li>\n
    3. \u00a0MATLAB \u30c4\u30fc\u30eb\u3092\u6d3b\u7528\u3057\u3001Text Analytics Toolbox<\/a> \u306e\u95a2\u6570\u3092\u4f7f\u7528\u3057\u3066\u3001\u30c6\u30ad\u30b9\u30c8\u306e\u53d6\u5f97\u3001\u7ba1\u7406\u3001\u6e96\u5099\u306a\u3069\u306e LLM \u6a5f\u80fd\u3092\u5f37\u5316\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"
      \"\"<\/div>\n
      \u203b\u3053\u306e\u6295\u7a3f\u306f 2024 \u5e74 1 \u6708 22 \u65e5\u306bArtificial Intelligence \u3078 \u6295\u7a3f\u3055\u308c\u305f\u3082\u306e\u306e\u6284\u8a33\u3067\u3059\u3002<\/p>\n
      MATLAB \u3092 OpenAI\u2122 API \u306b\u63a5\u7d9a\u3057\u3066\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\u3092\u884c\u3046\u65b9\u6cd5
      \nChatGPT\u2122\u3001Generative AI\u3001\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\uff08LLM, Large Language Model\uff09\u3001\u3053\u308c\u3089\u306e\u9769\u65b0\u7684\u306a\u6280\u8853\u3092 MATLAB…       read more >><\/a><\/p>\n","protected":false},"author":159,"featured_media":10619,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[141,61],"tags":[],"_links":{"self":[{"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/posts\/10601"}],"collection":[{"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/users\/159"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/comments?post=10601"}],"version-history":[{"count":15,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/posts\/10601\/revisions"}],"predecessor-version":[{"id":12052,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/posts\/10601\/revisions\/12052"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/media\/10619"}],"wp:attachment":[{"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/media?parent=10601"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/categories?post=10601"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.mathworks.com\/japan-community\/wp-json\/wp\/v2\/tags?post=10601"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}