{"id":648290,"date":"2025-10-15T05:48:09","date_gmt":"2025-10-15T04:48:09","guid":{"rendered":"https:\/\/blog.jetbrains.com\/pycharm\/2025\/08\/fine-tuning-and-deploying-gpt-models-using-hugging-face-transformers\/"},"modified":"2025-10-15T05:48:16","modified_gmt":"2025-10-15T04:48:16","slug":"fine-tuning-and-deploying-gpt-models-using-hugging-face-transformers","status":"publish","type":"pycharm","link":"https:\/\/blog.jetbrains.com\/ja\/pycharm\/2025\/10\/fine-tuning-and-deploying-gpt-models-using-hugging-face-transformers\/","title":{"rendered":"Hugging Face Transformers \u306b\u3088\u308b GPT \u30e2\u30c7\u30eb\u306e\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3068\u30c7\u30d7\u30ed\u30a4"},"content":{"rendered":"<p>Hugging Face \u306f\u73fe\u5728\u3001\u6a5f\u68b0\u5b66\u7fd2\u306e\u7814\u7a76\u8005\u3084\u611b\u597d\u5bb6\u306e\u9593\u3067\u5e83\u304f\u77e5\u3089\u308c\u3066\u3044\u308b\u540d\u524d\u3067\u3059\u3002 \u305d\u306e\u6700\u5927\u306e\u6210\u679c\u306e 1 \u3064\u306b\u306f\u3001\u30c6\u30ad\u30b9\u30c8\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30fc\u30d3\u30b8\u30e7\u30f3\u3001\u97f3\u58f0\u3001\u304a\u3088\u3073\u52d5\u753b\u306b\u304a\u3051\u308b\u6a5f\u68b0\u5b66\u7fd2\u30e2\u30c7\u30eb\u7528\u306e\u30e2\u30c7\u30eb\u5b9a\u7fa9\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3067\u3042\u308b <a href=\"https:\/\/huggingface.co\/docs\/transformers\/en\/index\" target=\"_blank\" rel=\"noopener\">Transformers<\/a> \u304c\u3042\u308a\u307e\u3059\u3002 <a href=\"https:\/\/huggingface.co\/models\" target=\"_blank\" rel=\"noopener\">Hugging Face Hub<\/a> \u306b\u6700\u5148\u7aef\u306e\u6a5f\u68b0\u5b66\u7fd2\u30e2\u30c7\u30eb\u306e\u5e83\u7bc4\u306a\u30ea\u30dd\u30b8\u30c8\u30ea\u304c\u516c\u958b\u3055\u308c\u3066\u304a\u308a\u3001Transformers \u306f\u5927\u591a\u6570\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306b\u5bfe\u5fdc\u3057\u3066\u3044\u308b\u3053\u3068\u304b\u3089\u3001\u63a8\u8ad6\u3084\u30e2\u30c7\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306b\u5e83\u304f\u4f7f\u7528\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n<h2 class=\"wp-block-heading\">AI \u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u7406\u7531<\/h2>\n<p>AI \u30e2\u30c7\u30eb\u306e\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306f\u3001\u7279\u5b9a\u306e\u30bf\u30b9\u30af\u3084\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u5408\u308f\u305b\u3066\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u8abf\u6574\u3057\u3001\u6c4e\u7528\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408\u3088\u308a\u3082\u9ad8\u3044\u7cbe\u5ea6\u3068\u52b9\u7387\u3092\u5b9f\u73fe\u3059\u308b\u4e0a\u3067\u975e\u5e38\u306b\u91cd\u8981\u3067\u3059\u3002 \u4e8b\u524d\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6e08\u307f\u306e\u30e2\u30c7\u30eb\u306b\u624b\u3092\u52a0\u3048\u3066\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u3068\u3001\u30bc\u30ed\u304b\u3089\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u5fc5\u8981\u304c\u306a\u304f\u306a\u308b\u305f\u3081\u3001\u6642\u9593\u3068\u30ea\u30bd\u30fc\u30b9\u3092\u7bc0\u7d04\u3067\u304d\u307e\u3059\u3002 \u307e\u305f\u3001\u7279\u5b9a\u5206\u91ce\u5185\u306e\u7279\u6b8a\u306a\u69d8\u5f0f\u3084\u30cb\u30e5\u30a2\u30f3\u30b9\u3001\u30a8\u30c3\u30b8\u30b1\u30fc\u30b9\u306e\u51e6\u7406\u3082\u6539\u5584\u3055\u308c\u308b\u305f\u3081\u3001\u4fe1\u983c\u6027\u306e\u9ad8\u3044\u8abf\u6574\u3055\u308c\u305f\u51fa\u529b\u3092\u5f97\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u3053\u306e\u30d6\u30ed\u30b0\u8a18\u4e8b\u3067\u306f\u3001\u6570\u5b66\u7684\u306a\u63a8\u8ad6\u306b\u3088\u3063\u3066 GPT \u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u3053\u3068\u3067\u3001\u6570\u5b66\u306e\u554f\u984c\u3092\u3088\u308a\u7684\u78ba\u306b\u51e6\u7406\u3067\u304d\u308b\u3088\u3046\u306b\u3057\u307e\u3059\u3002<\/p>\n<h2 class=\"wp-block-heading\">Hugging Face \u306e\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3059\u308b<\/h2>\n<p>PyCharm \u3092\u4f7f\u7528\u3059\u308b\u969b\u306b\u306f\u3001Hugging Face \u306e\u30e2\u30c7\u30eb\u3092\u7c21\u5358\u306b\u53c2\u7167\u3057\u3066\u8ffd\u52a0\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002 \u65b0\u898f\u306e Python \u30d5\u30a1\u30a4\u30eb\u5185\u3067\u3001\u4e0a\u90e8\u306e <em>Code<\/em>\uff08\u30b3\u30fc\u30c9\uff09\u30e1\u30cb\u30e5\u30fc\u304b\u3089 <em>Insert HF Model<\/em>\uff08HF \u30e2\u30c7\u30eb\u306e\u633f\u5165\uff09\u3092\u9078\u629e\u3057\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594074\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-40.png\" alt=\"Hugging Face \u306e\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3059\u308b\" width=\"946\" height=\"1070\" \/><\/figure>\n<p>\u8868\u793a\u3055\u308c\u305f\u30e1\u30cb\u30e5\u30fc\u5185\u3067\u306f\u3001\u30ab\u30c6\u30b4\u30ea\u3092\u6307\u5b9a\u3059\u308b\u304b\u3001\u4e0a\u90e8\u306e\u691c\u7d22\u30d0\u30fc\u306b\u5165\u529b\u3059\u308b\u3053\u3068\u3067\u30e2\u30c7\u30eb\u3092\u53c2\u7167\u3067\u304d\u307e\u3059\u3002 \u30e2\u30c7\u30eb\u3092\u9078\u629e\u3059\u308b\u3068\u3001\u30e2\u30c7\u30eb\u306e\u8aac\u660e\u304c\u53f3\u5074\u306b\u8868\u793a\u3055\u308c\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594085\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-41.png\" alt=\"Hugging Face \u306e\u30e2\u30c7\u30eb\u3092\u8a73\u3057\u304f\u898b\u308b\" width=\"1600\" height=\"923\" \/><\/figure>\n<p><em>Use Model<\/em>\uff08\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3059\u308b\uff09\u3092\u30af\u30ea\u30c3\u30af\u3059\u308b\u3068\u3001\u30b3\u30fc\u30c9\u30b9\u30cb\u30da\u30c3\u30c8\u304c\u30d5\u30a1\u30a4\u30eb\u306b\u8ffd\u52a0\u3055\u308c\u307e\u3059\u3002 \u3053\u306e\u64cd\u4f5c\u3060\u3051\u3067\u3001Hugging Face \u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3057\u59cb\u3081\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594096\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-42.png\" alt=\"PyCharm \u3067 Hugging Face \u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3059\u308b\" width=\"1600\" height=\"312\" \/><\/figure>\n<h2 class=\"wp-block-heading\">GPT\uff08Generative Pre-Trained Transformer\uff09\u30e2\u30c7\u30eb<\/h2>\n<p>GPT \u30e2\u30c7\u30eb\u306f <a href=\"https:\/\/huggingface.co\/models\" target=\"_blank\" rel=\"noopener\">Hugging Face Hub<\/a> \u3067\u5927\u304d\u306a\u4eba\u6c17\u3092\u96c6\u3081\u3066\u3044\u307e\u3059\u304c\u3001\u5b9f\u969b\u306b\u306f\u3069\u3093\u306a\u3082\u306e\u306a\u306e\u3067\u3057\u3087\u3046\u304b\uff1f GPT \u306f\u3001\u81ea\u7136\u8a00\u8a9e\u3092\u7406\u89e3\u3057\u3066\u9ad8\u54c1\u8cea\u306a\u30c6\u30ad\u30b9\u30c8\u3092\u751f\u6210\u3059\u308b\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6e08\u307f\u306e\u30e2\u30c7\u30eb\u3067\u3059\u3002 \u4e3b\u306b\u30c6\u30ad\u30b9\u30c8\u542b\u610f\u3001\u8cea\u554f\u5fdc\u7b54\u3001\u610f\u5473\u7684\u985e\u4f3c\u6027\u3001\u304a\u3088\u3073\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u5206\u985e\u306b\u95a2\u9023\u3059\u308b\u30bf\u30b9\u30af\u306b\u4f7f\u7528\u3055\u308c\u3066\u3044\u307e\u3059\u3002 \u6700\u3082\u6709\u540d\u306a\u4f8b\u306f\u3001<a href=\"https:\/\/openai.com\/index\/chatgpt\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u304c\u4f5c\u3063\u305f ChatGPT<\/a> \u3067\u3059\u3002<\/p>\n<p>\u591a\u6570\u306e OpenAI GPT \u30e2\u30c7\u30eb\u304c <a href=\"https:\/\/huggingface.co\/models\" target=\"_blank\" rel=\"noopener\">Hugging Face Hub<\/a> \u3067\u63d0\u4f9b\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u305d\u308c\u3089\u306e\u30e2\u30c7\u30eb\u3092 Transformers \u3067\u4f7f\u7528\u3057<em>\u3001<\/em>\u72ec\u81ea\u306e\u30c7\u30fc\u30bf\u3067\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u3066\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u7fd2\u3057\u307e\u3057\u3087\u3046<em>\u3002<\/em><\/p>\n<h2 class=\"wp-block-heading\">Transformers \u3092\u4f7f\u7528\u3059\u308b\u30e1\u30ea\u30c3\u30c8<\/h2>\n<p>Transformers \u306f Hugging Face \u304c\u63d0\u4f9b\u3059\u308b\u4ed6\u306e\u30c4\u30fc\u30eb\u3068\u9023\u643a\u3057\u3001\u4efb\u610f\u306e\u8907\u96d1\u306a\u30c7\u30a3\u30fc\u30d7\u30e9\u30fc\u30cb\u30f3\u30b0\u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u305f\u3081\u306e\u9ad8\u5ea6\u306a\u30c4\u30fc\u30eb\u3092\u63d0\u4f9b\u3057\u3066\u3044\u307e\u3059\u3002 \u305d\u308c\u3089\u306e\u30c4\u30fc\u30eb\u306f\u7279\u5b9a\u30e2\u30c7\u30eb\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3084\u30c8\u30fc\u30af\u30f3\u5316\u624b\u6cd5\u306e\u5b8c\u5168\u306a\u7406\u89e3\u3092\u5fc5\u8981\u3068\u305b\u305a\u3001\u30e2\u30c7\u30eb\u3092\u300c\u30d7\u30e9\u30b0\u30a2\u30f3\u30c9\u30d7\u30ec\u30a4\u300d\u3067\u4e92\u63db\u6027\u306e\u3042\u308b\u4efb\u610f\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u3068\u9023\u643a\u3055\u305b\u3001\u305d\u308c\u3068\u540c\u6642\u306b\u30c8\u30fc\u30af\u30f3\u5316\u3068\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u5927\u898f\u6a21\u306a\u30ab\u30b9\u30bf\u30de\u30a4\u30ba\u3092\u53ef\u80fd\u306b\u3057\u307e\u3059\u3002<\/p>\n<h2 class=\"wp-block-heading\">Transformers \u306e\u5b9f\u969b\u306e\u52d5\u4f5c<\/h2>\n<p>Transformers \u306e\u5b9f\u969b\u306e\u52d5\u4f5c\u3092\u8a73\u3057\u304f\u898b\u308b\u305f\u3081\u3001\u305d\u308c\u3092\u4f7f\u7528\u3057\u3066 GPT \u30e2\u30c7\u30eb\u3068\u3069\u306e\u3088\u3046\u306b\u5bfe\u8a71\u3067\u304d\u308b\u306e\u304b\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<h3 class=\"wp-block-heading\">\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u4e8b\u524d\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6e08\u307f\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3057\u3066\u63a8\u8ad6\u3059\u308b<\/h3>\n<p>OpenAI GPT-2 \u30e2\u30c7\u30eb\u3092\u9078\u629e\u3057\u3066\u30b3\u30fc\u30c9\u306b\u8ffd\u52a0\u3057\u305f\u3089\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import pipeline\n\n\npipe = pipeline(\"text-generation\", model=\"openai-community\/gpt2\")<\/pre>\n<p>\u3053\u306e\u30b3\u30fc\u30c9\u3092\u4f7f\u7528\u3059\u308b\u306b\u306f\u3001\u3044\u304f\u3064\u304b\u306e\u6e96\u5099\u304c\u5fc5\u8981\u3067\u3059\u3002 \u307e\u305a\u306f\u6a5f\u68b0\u5b66\u7fd2\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002 \u3053\u306e\u4f8b\u3067\u306f\u3001<a href=\"https:\/\/pytorch.org\/get-started\/locally\/\" target=\"_blank\" rel=\"noopener\">PyTorch<\/a> \u3092\u9078\u629e\u3057\u307e\u3057\u305f\u3002 PyCharm \u306e <em>Python Packages<\/em>\uff08Python \u30d1\u30c3\u30b1\u30fc\u30b8\uff09\u30a6\u30a3\u30f3\u30c9\u30a6\u3067\u7c21\u5358\u306b\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3067\u304d\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594107\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-43.png\" alt=\"PyCharm \u3067 PyTorch \u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3059\u308b\" width=\"920\" height=\"654\" \/><\/figure>\n<p>\u6b21\u306b\u3001`torch` \u30aa\u30d7\u30b7\u30e7\u30f3\u3092\u4f7f\u7528\u3057\u3066 Transformers \u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002 \u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u306b\u306f\u30bf\u30fc\u30df\u30ca\u30eb\u3092\u4f7f\u7528\u3057\u307e\u3059\u3002\u5de6\u5074\u306b\u3042\u308b\u30dc\u30bf\u30f3\u3092\u4f7f\u7528\u3059\u308b\u304b\u3001<em>\u2325 F12 <\/em>\uff08macOS\uff09\u307e\u305f\u306f <em>Alt + F12<\/em> \u30b7\u30e7\u30fc\u30c8\u30ab\u30c3\u30c8\u30ad\u30fc\u3092\u4f7f\u7528\u3057\u3066\u30bf\u30fc\u30df\u30ca\u30eb\u3092\u958b\u3044\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594118\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-44.png\" alt=\"PyCharm \u306e\u30bf\u30fc\u30df\u30ca\u30eb\u3067 Transformers \u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3059\u308b\" width=\"838\" height=\"502\" \/><\/figure>\n<p>\u3053\u3053\u3067\u306f uv \u3092\u4f7f\u7528\u3059\u308b\u305f\u3081\u3001\u30bf\u30fc\u30df\u30ca\u30eb\u3067\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u4f7f\u7528\u3057\u3066 uv \u3092\u4f9d\u5b58\u95a2\u4fc2\u3068\u3057\u3066\u8ffd\u52a0\u3057\u3001\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"bash\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">uv add \u201ctransformers[torch]\u201d\nuv sync<\/pre>\n<p>pip \u3092\u4f7f\u7528\u3057\u3066\u3044\u308b\u5834\u5408\u306f\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"bash\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">pip install \u201ctransformers[torch]\u201d<\/pre>\n<p>\u307e\u305f\u3001\u5f8c\u3067\u5fc5\u8981\u306b\u306a\u308b python-dotenv\u3001datasets<em>\u3001<\/em>notebook\u3001ipywidgets<em> \u306a\u3069\u306e\u30e9\u30a4\u30d6\u30e9\u30ea\u3082\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3057\u307e\u3059\u3002 <\/em>\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u306b\u306f\u4e0a\u8a18\u306e\u3044\u305a\u308c\u304b\u306e\u65b9\u6cd5\u3092\u4f7f\u7528\u3067\u304d\u307e\u3059\u3002<br \/>\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u5f8c\u306f GPU \u30c7\u30d0\u30a4\u30b9\u3092\u8ffd\u52a0\u3057\u3066\u30e2\u30c7\u30eb\u306e\u901f\u5ea6\u3092\u4e0a\u3052\u308b\u3053\u3068\u3092\u304a\u52e7\u3081\u3057\u307e\u3059\u3002 GPU \u30c7\u30d0\u30a4\u30b9\u306f\u3001\u30de\u30b7\u30f3\u306b\u642d\u8f09\u3055\u308c\u3066\u3044\u308b\u3082\u306e\u306b\u5fdc\u3058\u305f\u30c7\u30d0\u30a4\u30b9\u30d1\u30e9\u30e1\u30fc\u30bf\u30fc\u3092\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u8a2d\u5b9a\u3059\u308b\u3053\u3068\u306b\u3088\u3063\u3066\u8ffd\u52a0\u3067\u304d\u307e\u3059<em>\u3002 <\/em>\u3053\u3053\u3067\u306f Mac M2 \u30de\u30b7\u30f3\u3092\u4f7f\u7528\u3057\u3066\u3044\u308b\u305f\u3081\u3001<code class=\"EnlighterJSRAW\" data-enlighter-language=\"python\">device=\"mps\"<\/code> \u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u8a2d\u5b9a\u3067\u304d\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">pipe = pipeline(\"text-generation\", model=\"openai-community\/gpt2\", device=\"mps\")<\/pre>\n<p>CUDA GPU \u304c\u642d\u8f09\u3055\u308c\u3066\u3044\u308b\u5834\u5408\u306f\u3001<code class=\"EnlighterJSRAW\" data-enlighter-language=\"python\">device=\"cuda\"<\/code> \u306b\u8a2d\u5b9a\u3059\u308b\u3053\u3068\u3082\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306e\u30bb\u30c3\u30c8\u30a2\u30c3\u30d7\u304c\u5b8c\u4e86\u3057\u305f\u306e\u3067\u3001\u7c21\u5358\u306a\u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u4f7f\u7528\u3057\u3066\u8a66\u3057\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import pipeline\n\n\npipe = pipeline(\"text-generation\", model=\"openai-community\/gpt2\", device=\"mps\")\n\n\nprint(pipe(\"A rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width?\", max_new_tokens=200))<\/pre>\n<p>\u4e0a\u90e8\u306e <em>Run<\/em>\uff08\u5b9f\u884c\uff09\u30dc\u30bf\u30f3\uff08<img decoding=\"async\" loading=\"lazy\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/09\/AD_4nXf6ZDm7vSGyFlO0DzXegK6WP9JxsStUiJA-bkRZ0mwPsUsmn8M70emV5Sr8f17-fEK6z9V1EQKWEm3RPHdT8n8uqG18faVmQn5y09psVInQLU0CZQKXAEg2q7m7AOsh4hPU7G8gcQ.png\" width=\"30\" height=\"23\" \/>\uff09\u3067\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u5b9f\u884c\u3057\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594129\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-45.png\" alt=\"PyCharm \u3067\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u5b9f\u884c\u3059\u308b\" width=\"1050\" height=\"282\" \/><\/figure>\n<p>\u7d50\u679c\u306f\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">[{'generated_text': 'A rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width?nnA rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width?nnA rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width?nnA rectangle has a perimeter of 20 cm. If the width is 6 cm, what is the width? A rectangle has a perimeter'}]<\/pre>\n<p>\u307e\u3063\u305f\u304f\u63a8\u8ad6\u3067\u304d\u3066\u304a\u3089\u305a\u3001\u7121\u610f\u5473\u306a\u5185\u5bb9\u304c\u7f85\u5217\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u8b66\u544a\u304c\u8868\u793a\u3055\u308c\u308b\u53ef\u80fd\u6027\u3082\u3042\u308a\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"bash\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">Setting `pad_token_id` to `eos_token_id`:50256 for open-end generation.<\/pre>\n<p>\u3053\u308c\u306f\u30c7\u30d5\u30a9\u30eb\u30c8\u306e\u8a2d\u5b9a\u3067\u3059\u3002\u4ee5\u4e0b\u3092\u624b\u52d5\u3067\u8ffd\u52a0\u3057\u3066\u3053\u306e\u30e1\u30c3\u30bb\u30fc\u30b8\u306e\u8868\u793a\u3092\u6d88\u3059\u3053\u3068\u3082\u3067\u304d\u307e\u3059\u304c\u3001\u3053\u306e\u6bb5\u968e\u3067\u306f\u3042\u307e\u308a\u6c17\u306b\u3059\u308b\u5fc5\u8981\u306f\u3042\u308a\u307e\u305b\u3093\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">print(pipe(\"A rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width?\", max_new_tokens=200, pad_token_id=pipe.tokenizer.eos_token_id))<\/pre>\n<p>\u521d\u671f\u72b6\u614b\u3067 GPT-2 \u304c\u3069\u306e\u3088\u3046\u306b\u52d5\u4f5c\u3059\u308b\u304b\u3092\u78ba\u8a8d\u3057\u307e\u3057\u305f\u306e\u3067\u3001\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306b\u3088\u3063\u3066\u6570\u5b66\u7684\u306a\u63a8\u8ad6\u3092\u6539\u5584\u3067\u304d\u308b\u304b\u3069\u3046\u304b\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<h3 class=\"wp-block-heading\">Hugging Face Hub \u304b\u3089\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u8aad\u307f\u8fbc\u3093\u3067\u6e96\u5099\u3059\u308b<\/h3>\n<p>GPT \u30e2\u30c7\u30eb\u306e\u6539\u826f\u3092\u59cb\u3081\u308b\u306b\u306f\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u304c\u5fc5\u8981\u3067\u3059\u3002 Hugging Face Hub \u304b\u3089\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u53d6\u5f97\u65b9\u6cd5\u3092\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<p>Hugging Face \u30a2\u30ab\u30a6\u30f3\u30c8\u306b\u307e\u3060\u767b\u9332\u3057\u3066\u3044\u306a\u3044\u5834\u5408\u306f\u3001\u767b\u9332\u3092\u5b8c\u4e86\u3057\u3066\u304b\u3089<a href=\"https:\/\/huggingface.co\/docs\/hub\/security-tokens#user-access-tokens\" target=\"_blank\" rel=\"noopener\">\u30a2\u30af\u30bb\u30b9\u30c8\u30fc\u30af\u30f3\u3092\u4f5c\u6210<\/a>\u3057\u307e\u3059\u3002 \u3053\u306e\u6642\u70b9\u3067\u306f `read` \u30c8\u30fc\u30af\u30f3\u306e\u307f\u304c\u5fc5\u8981\u3067\u3059\u3002 \u4ee5\u4e0b\u306e\u3088\u3046\u306b\u30c8\u30fc\u30af\u30f3\u3092 `.env` \u30d5\u30a1\u30a4\u30eb\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"bash\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">HF_TOKEN=your-hugging-face-access-token<\/pre>\n<p>\u6570\u5b66\u7684\u306a\u63a8\u8ad6\u3092\u8aac\u660e\u3059\u308b\u30c6\u30ad\u30b9\u30c8\u304c\u542b\u307e\u308c\u308b\u3001\u3053\u306e <a href=\"https:\/\/huggingface.co\/datasets\/Cheukting\/math-meta-reasoning-cleaned\" target=\"_blank\" rel=\"noopener\">Math Reasoning Dataset<\/a> \u3092\u4f7f\u7528\u3057\u307e\u3059\u3002 \u6570\u5b66\u306e\u554f\u984c\u3092\u3088\u308a\u52b9\u7387\u3088\u304f\u89e3\u3051\u308b\u3088\u3046\u3001\u3053\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4f7f\u3063\u3066 GPT \u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u307e\u3059\u3002<\/p>\n<p>\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306b\u4f7f\u7528\u3059\u308b\u65b0\u3057\u3044 Jupyter \u30ce\u30fc\u30c8\u30d6\u30c3\u30af\u3092\u4f5c\u6210\u3057\u307e\u3057\u3087\u3046\u3002Jupyter \u30ce\u30fc\u30c8\u30d6\u30c3\u30af\u306a\u3089\u3001\u8907\u6570\u306e\u7570\u306a\u308b\u30b3\u30fc\u30c9\u30b9\u30cb\u30da\u30c3\u30c8\u3092 1 \u3064\u305a\u3064\u5b9f\u884c\u3057\u306a\u304c\u3089\u9032\u884c\u3092\u76e3\u8996\u3067\u304d\u308b\u305f\u3081\u3067\u3059\u3002<\/p>\n<p>\u6700\u521d\u306e\u30bb\u30eb\u3067\u306f\u3001\u4ee5\u4e0b\u306e\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4f7f\u7528\u3057\u3066 Hugging Face Hub \u304b\u3089\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u8aad\u307f\u8fbc\u307f\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from datasets import load_dataset\nfrom dotenv import load_dotenv\nimport os\n\n\nload_dotenv()\ndataset = load_dataset(\"Cheukting\/math-meta-reasoning-cleaned\", token=os.getenv(\"HF_TOKEN\"))\ndataset<\/pre>\n<p>\u3053\u306e\u30bb\u30eb\u3092\u5b9f\u884c\u3059\u308b\u3068\uff08\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306e\u901f\u5ea6\u306b\u3088\u3063\u3066\u306f\u5c11\u3057\u6642\u9593\u304c\u304b\u304b\u308a\u307e\u3059\uff09\u3001\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u304c\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3055\u308c\u307e\u3059\u3002 \u5b8c\u4e86\u3057\u305f\u3089\u3001\u7d50\u679c\u3092\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">DatasetDict({\n    train: Dataset({\n        features: ['id', 'text', 'token_count'],\n        num_rows: 987485\n    })\n})\n<\/pre>\n<p>\u8208\u5473\u304c\u3042\u3063\u3066\u30c7\u30fc\u30bf\u3092\u78ba\u8a8d\u3057\u305f\u3044\u5834\u5408\u306f\u3001PyCharm \u3067\u30d7\u30ec\u30d3\u30e5\u30fc\u3067\u304d\u307e\u3059\u3002 \u53f3\u4e0b\u306e\u30dc\u30bf\u30f3\u3092\u4f7f\u7528\u3057\u3066\u3001<em>Jupyter Variables<\/em>\uff08Jupyter \u5909\u6570\uff09\u30a6\u30a3\u30f3\u30c9\u30a6\u3092\u958b\u304d\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594140\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-46.png\" alt=\"PyCharm \u3067 Jupyter Variables\uff08Jupyter \u5909\u6570\uff09\u3092\u958b\u304f\" width=\"1052\" height=\"740\" \/><\/figure>\n<p><em>dataset<\/em> \u3092\u5c55\u958b\u3059\u308b\u3068\u3001<em>dataset[\u2018train\u2019]<\/em> \u306e\u6a2a\u306b <em>View as DataFrame<\/em>\uff08DataFrame \u3068\u3057\u3066\u8868\u793a\uff09\u30aa\u30d7\u30b7\u30e7\u30f3\u304c\u73fe\u308c\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594152\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-47.png\" alt=\"PyCharm \u306e Jupyter Variables\uff08Jupyter \u5909\u6570\uff09\" width=\"980\" height=\"882\" \/><\/figure>\n<p>\u305d\u306e\u30aa\u30d7\u30b7\u30e7\u30f3\u3092\u30af\u30ea\u30c3\u30af\u3059\u308b\u3068\u3001<em>Data View<\/em>\uff08\u30c7\u30fc\u30bf\u30d3\u30e5\u30fc\uff09\u30c4\u30fc\u30eb\u30a6\u30a3\u30f3\u30c9\u30a6\u3067\u30c7\u30fc\u30bf\u3092\u78ba\u8a8d\u3067\u304d\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594163\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-48.png\" alt=\"PyCharm \u306e Data View\uff08\u30c7\u30fc\u30bf\u30d3\u30e5\u30fc\uff09\u30c4\u30fc\u30eb\" width=\"980\" height=\"1102\" \/><\/figure>\n<p>\u6b21\u306b\u3001\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u30c6\u30ad\u30b9\u30c8\u3092\u30c8\u30fc\u30af\u30f3\u5316\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import GPT2Tokenizer\n\n\ntokenizer = GPT2Tokenizer.from_pretrained(\"openai-community\/gpt2\")\ntokenizer.pad_token = tokenizer.eos_token\n\n\ndef tokenize_function(examples):\n   return tokenizer(examples['text'], truncation=True, padding='max_length', max_length=512)\n\n\ntokenized_datasets = dataset.map(tokenize_function, batched=True)<\/pre>\n<p>\u3053\u3053\u3067\u306f GPT-2 \u30c8\u30fc\u30af\u30ca\u30a4\u30b6\u30fc\u3092\u4f7f\u7528\u3057\u3001<code>pad_token<\/code> \u304c\u30d5\u30a1\u30a4\u30eb\u306e\u7d42\u7aef\u3092\u793a\u3059\u30c8\u30fc\u30af\u30f3\u3067\u3042\u308b <code>eos_token<\/code> \u306b\u306a\u308b\u3088\u3046\u306b\u8a2d\u5b9a\u3057\u307e\u3059\u3002 \u8a2d\u5b9a\u304c\u5b8c\u4e86\u3057\u305f\u3089\u3001\u95a2\u6570\u3092\u4f7f\u7528\u3057\u3066\u30c6\u30ad\u30b9\u30c8\u3092\u30c8\u30fc\u30af\u30f3\u5316\u3057\u307e\u3059\u3002 \u521d\u56de\u5b9f\u884c\u6642\u306b\u306f\u5c11\u3057\u6642\u9593\u304c\u304b\u304b\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\u304c\u3001\u305d\u308c\u4ee5\u964d\u306f\u30ad\u30e3\u30c3\u30b7\u30e5\u3055\u308c\u3001\u30bb\u30eb\u3092\u518d\u5b9f\u884c\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u5834\u5408\u306b\u7d20\u65e9\u304f\u5b9f\u884c\u3067\u304d\u308b\u3088\u3046\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\n<p>\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u306f\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306b\u4f7f\u7528\u3067\u304d\u308b\u7d04 100 \u4e07\u306e\u884c\u304c\u542b\u307e\u308c\u3066\u3044\u307e\u3059\u3002 \u3053\u306e\u3059\u3079\u3066\u3092\u51e6\u7406\u3067\u304d\u308b\u306e\u306b\u5341\u5206\u306a\u8a08\u7b97\u80fd\u529b\u304c\u3042\u308b\u5834\u5408\u306f\u3001\u3059\u3079\u3066\u306e\u884c\u3092\u4f7f\u7528\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002 \u3053\u306e\u30c7\u30e2\u3067\u306f\u30ce\u30fc\u30c8\u30d1\u30bd\u30b3\u30f3\u4e0a\u3067\u30ed\u30fc\u30ab\u30eb\u306b\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u3057\u3066\u3044\u308b\u305f\u3081\u3001\u3054\u304f\u4e00\u90e8\u306e\u307f\u3092\u4f7f\u7528\u3059\u308b\u3053\u3068\u306b\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">tokenized_datasets_split = tokenized_datasets[\"train\"].shard(num_shards=100, index=0).train_test_split(test_size=0.2, shuffle=True)\ntokenized_datasets_split<\/pre>\n<p>\u3053\u3053\u3067\u306f\u30c7\u30fc\u30bf\u306e 1% \u306e\u307f\u3092\u4f7f\u7528\u3057\u3001\u305d\u306e\u5f8c\u306b <code>train_test_split<\/code> \u3092\u5b9f\u884c\u3057\u3066\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092 2 \u3064\u306b\u5206\u5272\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">DatasetDict({\n    train: Dataset({\n        features: ['id', 'text', 'token_count', 'input_ids', 'attention_mask'],\n        num_rows: 7900\n    })\n    test: Dataset({\n        features: ['id', 'text', 'token_count', 'input_ids', 'attention_mask'],\n        num_rows: 1975\n    })\n})\n<\/pre>\n<p>\u3053\u308c\u3067 GPT-2 \u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b\u6e96\u5099\u304c\u6574\u3044\u307e\u3057\u305f\u3002<\/p>\n<h3 class=\"wp-block-heading\">GPT \u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3059\u308b<\/h3>\n<p>\u6b21\u306e\u7a7a\u306e\u30bb\u30eb\u3067\u306f\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u5f15\u6570\u3092\u8a2d\u5b9a\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import TrainingArguments\ntraining_args = TrainingArguments(\n   output_dir='.\/results',\n   num_train_epochs=5,\n   per_device_train_batch_size=8,\n   per_device_eval_batch_size=8,\n   warmup_steps=100,\n   weight_decay=0.01,\n   save_steps = 500,\n   logging_steps=100,\n   dataloader_pin_memory=False\n)<\/pre>\n<p>\u3053\u308c\u3089\u306e\u5f15\u6570\u306e\u307b\u3068\u3093\u3069\u306f\u3001\u30e2\u30c7\u30eb\u306e\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3067\u306f\u3054\u304f\u4e00\u822c\u7684\u306a\u3082\u306e\u3067\u3059\u3002 \u305f\u3060\u3057\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30fc\u306e\u74b0\u5883\u306b\u5fdc\u3058\u3066\u82e5\u5e72\u306e\u8abf\u6574\u3092\u884c\u3046\u3053\u3068\u3092\u304a\u52e7\u3081\u3057\u307e\u3059\u3002<\/p>\n<ul>\n<li>\u30d0\u30c3\u30c1\u30b5\u30a4\u30ba\uff08batch_size\uff09 \u2013 \u30d0\u30c3\u30c1\u30b5\u30a4\u30ba\u304c\u5927\u304d\u3044\u307b\u3069\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u901f\u5ea6\u304c\u5411\u4e0a\u3059\u308b\u305f\u3081\u3001\u6700\u9069\u306a\u30d0\u30c3\u30c1\u30b5\u30a4\u30ba\u3092\u898b\u3064\u3051\u308b\u3053\u3068\u304c\u91cd\u8981\u3067\u3059\u3002 \u305f\u3060\u3057\u3001CPU \u3084 GPU \u3067\u4f7f\u7528\u53ef\u80fd\u306a\u30e1\u30e2\u30ea\u91cf\u306b\u306f\u9650\u5ea6\u304c\u3042\u308b\u305f\u3081\u3001\u4e0a\u9650\u304c\u3042\u308b\u3053\u3068\u306b\u6c17\u3065\u304f\u304b\u3068\u601d\u3044\u307e\u3059\u3002<\/li>\n<li>\u30a8\u30dd\u30c3\u30af\u6570\uff08epochs\uff09 \u2013 \u30a8\u30dd\u30c3\u30af\u6570\u304c\u591a\u304f\u306a\u308b\u307b\u3069\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306b\u304b\u304b\u308b\u6642\u9593\u304c\u9577\u304f\u306a\u308a\u307e\u3059\u3002 \u5fc5\u8981\u306a\u30a8\u30dd\u30c3\u30af\u6570\u3092\u6307\u5b9a\u3067\u304d\u307e\u3059\u3002<\/li>\n<li>save_steps\uff08\u4fdd\u5b58\u30b9\u30c6\u30c3\u30d7\u6570\uff09 \u2013 \u4fdd\u5b58\u30b9\u30c6\u30c3\u30d7\u6570\u306f\u3001\u30c1\u30a7\u30c3\u30af\u30dd\u30a4\u30f3\u30c8\u304c\u30c7\u30a3\u30b9\u30af\u306b\u4fdd\u5b58\u3055\u308c\u308b\u983b\u5ea6\u3092\u6c7a\u5b9a\u3057\u307e\u3059\u3002 \u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u306e\u901f\u5ea6\u304c\u9045\u304f\u3001\u4e88\u671f\u305b\u305a\u505c\u6b62\u3059\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u5834\u5408\u306f\u3001\u4fdd\u5b58\u983b\u5ea6\u3092\u4e0a\u3052\u308b\uff08\u5024\u3092\u4f4e\u3081\u306b\u8a2d\u5b9a\u3059\u308b\uff09\u3053\u3068\u3092\u304a\u52e7\u3081\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<p>\u8a2d\u5b9a\u3092\u8abf\u6574\u3057\u305f\u3089\u3001\u6b21\u306e\u30bb\u30eb\u306b trainer \u3092\u7d44\u307f\u8fbc\u307f\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import Trainer, DataCollatorForLanguageModeling\n\n\nmodel = GPT2LMHeadModel.from_pretrained(\"openai-community\/gpt2\")\ndata_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False)\n\n\ntrainer = Trainer(\n   model=model,\n   args=training_args,\n   train_dataset=tokenized_datasets_split['train'],\n   eval_dataset=tokenized_datasets_split['test'],\n   data_collator=data_collator,\n)\n\n\ntrainer.train(resume_from_checkpoint=False)<\/pre>\n<p>\u3053\u3053\u3067\u306f `resume_from_checkpoint=False` \u3092\u8a2d\u5b9a\u3057\u3066\u3044\u307e\u3059\u304c\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u304c\u4e2d\u65ad\u3057\u305f\u5834\u5408\u306b\u6700\u7d42\u30c1\u30a7\u30c3\u30af\u30dd\u30a4\u30f3\u30c8\u304b\u3089\u7d9a\u884c\u3067\u304d\u308b\u3088\u3046\u306b `True` \u306b\u8a2d\u5b9a\u3059\u308b\u3053\u3068\u3082\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u304c\u5b8c\u4e86\u3057\u305f\u3089\u3001\u30e2\u30c7\u30eb\u3092\u8a55\u4fa1\u3057\u3066\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">trainer.evaluate(tokenized_datasets_split['test'])\ntrainer.save_model(\".\/trained_model\")<\/pre>\n<p>\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6e08\u307f\u306e\u30e2\u30c7\u30eb\u3092\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u4f7f\u7528\u3067\u304d\u308b\u3088\u3046\u306b\u306a\u308a\u307e\u3057\u305f\u3002 \u4e8b\u524d\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u6e08\u307f\u306e\u30e2\u30c7\u30eb\u3067\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3092\u4f7f\u7528\u3057\u305f `model.py` \u306b\u623b\u308a\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">from transformers import pipeline\n\n\npipe = pipeline(\"text-generation\", model=\"openai-community\/gpt2\", device=\"mps\")\n\n\nprint(pipe(\"A rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width?\", max_new_tokens=200, pad_token_id=pipe.tokenizer.eos_token_id))<\/pre>\n<p>\u6b21\u306b\u3001`model=\u201dopenai-community\/gpt2\u2033` \u3092 `model=\u201d.\/trained_model\u201d` \u306b\u5909\u66f4\u3057\u3001\u305d\u306e\u7d50\u679c\u3092\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">[{'generated_text': \"A rectangle has a perimeter of 20 cm. If the length is 6 cm, what is the width?nAlright, let me try to solve this problem as a student, and I'll let my thinking naturally fall into the common pitfall as described.nn---nn**Step 1: Attempting the Problem (falling into the pitfall)**nnWe have a rectangle with perimeter 20 cm. The length is 6 cm. We want the width.nnFirst, I need to find the area under the rectangle.nnLet\u2019s set \\( A = 20 - 12 \\), where \\( A \\) is the perimeter.nn**Area under a rectangle:**  n\\[nA = (20-12)^2 + ((-12)^2)^2 = 20^2 + 12^2 = 24n\\]nnSo, \\( 24 = (20-12)^2 = 27 \\).nnNow, I\u2019ll just divide both sides by 6 to find the area under the rectangle.n\"}]<\/pre>\n<p>\u6b8b\u5ff5\u306a\u304c\u3089\u3001\u307e\u3060\u554f\u984c\u3092\u89e3\u6c7a\u3067\u304d\u307e\u305b\u3093\u3002 \u305f\u3060\u3057\u3001\u4ee5\u524d\u306b\u306f\u4f7f\u7528\u3055\u308c\u306a\u304b\u3063\u305f\u6570\u5f0f\u3068\u63a8\u8ad6\u304c\u73fe\u308c\u307e\u3057\u305f\u3002 \u5fc5\u8981\u3067\u3042\u308c\u3070\u3001\u307e\u3060\u4f7f\u7528\u3057\u3066\u3044\u306a\u3044\u30c7\u30fc\u30bf\u3092\u4f7f\u7528\u3057\u3066\u3082\u3046\u5c11\u3057\u3060\u3051\u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u6b21\u306e\u30bb\u30af\u30b7\u30e7\u30f3\u3067\u306f\u3001Hugging Face \u3068 FastAPI \u304c\u63d0\u4f9b\u3059\u308b\u4e21\u65b9\u306e\u30c4\u30fc\u30eb\u3092\u4f7f\u7528\u3057\u3066\u3001\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u30e2\u30c7\u30eb\u3092 API \u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u65b9\u6cd5\u3092\u78ba\u8a8d\u3057\u307e\u3059\u3002<\/p>\n<h2 class=\"wp-block-heading\">\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u30e2\u30c7\u30eb\u306e\u30c7\u30d7\u30ed\u30a4<\/h2>\n<p>\u30e2\u30c7\u30eb\u3092\u30b5\u30fc\u30d0\u30fc\u30d0\u30c3\u30af\u30a8\u30f3\u30c9\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u306b\u306f\u3001FastAPI \u3092\u4f7f\u7528\u3059\u308b\u306e\u304c\u6700\u3082\u7c21\u5358\u3067\u3059\u3002 FastAPI \u3092\u4f7f\u7528\u3057\u3066\u6a5f\u68b0\u5b66\u7fd2\u30e2\u30c7\u30eb\u3092\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u65b9\u6cd5\u306b\u3064\u3044\u3066\u306f\u524d\u56de\u306e<a href=\"https:\/\/blog.jetbrains.com\/pycharm\/2024\/09\/how-to-use-fastapi-for-machine-learning\/\">\u30d6\u30ed\u30b0\u8a18\u4e8b<\/a>\u3067\u53d6\u308a\u4e0a\u3052\u307e\u3057\u305f\u306e\u3067\u3001 \u3053\u3053\u3067\u306f\u540c\u3058\u30ec\u30d9\u30eb\u3067\u8a73\u3057\u304f\u66f8\u304b\u305a\u3001\u7c21\u5358\u306b\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u30e2\u30c7\u30eb\u3092\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u65b9\u6cd5\u3092\u8aac\u660e\u3057\u307e\u3059\u3002<\/p>\n<p><a href=\"https:\/\/www.jetbrains.com\/junie\/\" target=\"_blank\" rel=\"noopener\">Junie<\/a> \u3092\u5229\u7528\u3057\u3066\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4f5c\u6210\u3057\u307e\u3057\u305f\uff08<a href=\"https:\/\/github.com\/Cheukting\/fine-tune-gpt2\/tree\/main\/app\" target=\"_blank\" rel=\"noopener\">\u3053\u3061\u3089<\/a>\u3067\u78ba\u8a8d\u3067\u304d\u307e\u3059\uff09\u3002 \u3053\u306e\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4f7f\u7528\u3059\u308b\u3068\u3001FastAPI \u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3067\u30b5\u30fc\u30d0\u30fc\u30d0\u30c3\u30af\u30a8\u30f3\u30c9\u3092\u30c7\u30d7\u30ed\u30a4\u3067\u304d\u307e\u3059\u3002<\/p>\n<p>\u3044\u304f\u3064\u304b\u306e\u65b0\u3057\u3044\u4f9d\u5b58\u95a2\u4fc2\u3092\u8ffd\u52a0\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"bash\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">uv add fastapi pydantic uvicorn\nuv sync<\/pre>\n<p>`main.py` \u306e\u4e2d\u3067\u30b9\u30af\u30ea\u30d7\u30c8\u306e\u8208\u5473\u6df1\u3044\u30dd\u30a4\u30f3\u30c8\u3092\u3044\u304f\u3064\u304b\u898b\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\"># Initialize FastAPI app\napp = FastAPI(\n   title=\"Text Generation API\",\n   description=\"API for generating text using a fine-tuned model\",\n   version=\"1.0.0\"\n)\n\n\n# Initialize the model pipeline\ntry:\n   pipe = pipeline(\"text-generation\", model=\"..\/trained_model\", device=\"mps\")\nexcept Exception as e:\n   # Fallback to CPU if MPS is not available\n   try:\n       pipe = pipeline(\"text-generation\", model=\"..\/trained_model\", device=\"cpu\")\n   except Exception as e:\n       print(f\"Error loading model: {e}\")\n       pipe = None<\/pre>\n<p>\u3053\u306e\u30b9\u30af\u30ea\u30d7\u30c8\u306f\u30a2\u30d7\u30ea\u3092\u521d\u671f\u5316\u3057\u305f\u5f8c\u3001\u30e2\u30c7\u30eb\u3092\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306b\u8aad\u307f\u8fbc\u3082\u3046\u3068\u3057\u307e\u3059\u3002 Metal GPU \u304c\u4f7f\u7528\u3067\u304d\u306a\u3044\u5834\u5408\u306f\u3001CPU \u3092\u4ee3\u308f\u308a\u306b\u4f7f\u7528\u3057\u307e\u3059\u3002 Metal GPU \u3067\u306f\u306a\u304f\u3001CUDA GPU \u3092\u304a\u6301\u3061\u306e\u5834\u5408\u306f `mps` \u3092 `cuda` \u306b\u5909\u66f4\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\"># Request model\nclass TextGenerationRequest(BaseModel):\n   prompt: str\n   max_new_tokens: int = 200\n  \n# Response model\nclass TextGenerationResponse(BaseModel):\n   generated_text: str<\/pre>\n<p>Pydantic \u306e `BaseModel` \u3092\u7d99\u627f\u3057\u3066\u65b0\u3057\u3044\u30af\u30e9\u30b9\u304c 2 \u3064\u4f5c\u6210\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\u307e\u305f\u3001<em>Endpoints<\/em>\uff08\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\uff09\u30c4\u30fc\u30eb\u30a6\u30a3\u30f3\u30c9\u30a6\u3067\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3092\u691c\u67fb\u3059\u308b\u3053\u3068\u3082\u3067\u304d\u307e\u3059<em>\u3002 <\/em>11 \u884c\u76ee\u306e `app = FastAPI` \u306e\u6a2a\u306b\u3042\u308b\u5730\u7403\u5100\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066\u3001<em>Show All Endpoints<\/em>\uff08\u3059\u3079\u3066\u306e\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3092\u8868\u793a\uff09\u3092\u9078\u629e\u3057\u307e\u3059\u3002<\/p>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594174\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-49.png\" alt=\"PyCharm \u3067\u3059\u3079\u3066\u306e\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3092\u8868\u793a\u3059\u308b\" width=\"1600\" height=\"833\" \/><\/figure>\n<p>\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u306f 3 \u3064\u3042\u308a\u307e\u3059\u3002 \u30eb\u30fc\u30c8\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u306f\u5358\u306a\u308b\u30a6\u30a7\u30eb\u30ab\u30e0\u30e1\u30c3\u30bb\u30fc\u30b8\u3067\u3042\u308b\u305f\u3081\u3001\u4ed6\u306e 2 \u3064\u306b\u6ce8\u76ee\u3059\u308b\u3053\u3068\u306b\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">@app.post(\"\/generate\", response_model=TextGenerationResponse)\nasync def generate_text(request: TextGenerationRequest):\n   \"\"\"\n   Generate text based on the provided prompt.\n  \n   Args:\n       request: TextGenerationRequest containing the prompt and generation parameters\n      \n   Returns:\n       TextGenerationResponse with the generated text\n   \"\"\"\n   if pipe is None:\n       raise HTTPException(status_code=500, detail=\"Model not loaded properly\")\n  \n   try:\n       result = pipe(\n           request.prompt,\n           max_new_tokens=request.max_new_tokens,\n           pad_token_id=pipe.tokenizer.eos_token_id\n       )\n      \n       # Extract the generated text from the result\n       generated_text = result[0]['generated_text']\n      \n       return TextGenerationResponse(generated_text=generated_text)\n   except Exception as e:\n       raise HTTPException(status_code=500, detail=f\"Error generating text: {str(e)}\")\n<\/pre>\n<p>`\/generate` \u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u306f request \u30d7\u30ed\u30f3\u30d7\u30c8\u3092\u53ce\u96c6\u3057\u3001\u30e2\u30c7\u30eb\u3092\u4f7f\u7528\u3057\u3066\u5fdc\u7b54\u30c6\u30ad\u30b9\u30c8\u3092\u751f\u6210\u3057\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">@app.get(\"\/health\")\nasync def health_check():\n   \"\"\"Check if the API and model are working properly.\"\"\"\n   if pipe is None:\n       raise HTTPException(status_code=500, detail=\"Model not loaded\")\n   return {\"status\": \"healthy\", \"model_loaded\": True}<\/pre>\n<p>`\/health` \u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u306f\u3001\u30e2\u30c7\u30eb\u304c\u6b63\u3057\u304f\u8aad\u307f\u8fbc\u307e\u308c\u3066\u3044\u308b\u304b\u3069\u3046\u304b\u3092\u78ba\u8a8d\u3057\u307e\u3059\u3002 \u3053\u308c\u306f\u3001\u30af\u30e9\u30a4\u30a2\u30f3\u30c8\u30b5\u30a4\u30c9\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u304c\u78ba\u8a8d\u3092\u884c\u3063\u305f\u5f8c\u306b\u4ed6\u306e\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3092 UI \u3067\u4f7f\u7528\u3067\u304d\u308b\u3088\u3046\u306b\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u5834\u5408\u306b\u4fbf\u5229\u3067\u3059\u3002<\/p>\n<p>`run.py` \u3067\u306f\u3001<a href=\"https:\/\/www.uvicorn.org\/\" target=\"_blank\" rel=\"noopener\">uvicorn<\/a> \u3092\u4f7f\u7528\u3057\u3066\u30b5\u30fc\u30d0\u30fc\u3092\u5b9f\u884c\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">import uvicorn\n\n\nif __name__ == \"__main__\":\n   uvicorn.run(\"main:app\", host=\"0.0.0.0\", port=8000, reload=True)<\/pre>\n<p>\u3053\u306e\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u5b9f\u884c\u3059\u308b\u3068\u3001<a href=\"http:\/\/0.0.0.0:8000\/\" target=\"_blank\" rel=\"noopener\">http:\/\/0.0.0.0:8000\/<\/a> \u3067\u30b5\u30fc\u30d0\u30fc\u304c\u8d77\u52d5\u3057\u307e\u3059\u3002<\/p>\n<p>\u30b5\u30fc\u30d0\u30fc\u304c\u7a3c\u50cd\u3057\u59cb\u3081\u305f\u3089\u3001<a href=\"http:\/\/0.0.0.0:8000\/docs\" target=\"_blank\" rel=\"noopener\">http:\/\/0.0.0.0:8000\/docs<\/a> \u306b\u30a2\u30af\u30bb\u30b9\u3057\u3066\u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3092\u30c6\u30b9\u30c8\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n<h2 class=\"wp-block-heading\"><img decoding=\"async\" loading=\"lazy\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/09\/AD_4nXf6PXwz_Vz7VEQoyZs20NJ9TsO36oWJPf0w4iMjwHZ_EBP1Pk9c_8aWR2ybGE-wsmArM1zAQl1s8jHEr09I0g1A3boGD1Kt4i4CemufHZTHnATjIWrJ8x2ZUYg4Q7E4b3tc2XDmmg.png\" width=\"624\" height=\"269\" \/><\/h2>\n<p>\u3053\u308c\u3092 `\/generate` \u30a8\u30f3\u30c9\u30dd\u30a4\u30f3\u30c8\u3067\u8a66\u3057\u3066\u307f\u307e\u3057\u3087\u3046\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">{\n  \"prompt\": \"5 people give each other a present. How many presents are given altogether?\",\n  \"max_new_tokens\": 300\n}<\/pre>\n<p>\u4ee5\u4e0b\u306e\u3088\u3046\u306a\u5fdc\u7b54\u304c\u5f97\u3089\u308c\u307e\u3059\u3002<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"generic\" data-enlighter-theme=\"\" data-enlighter-highlight=\"\" data-enlighter-linenumbers=\"\" data-enlighter-lineoffset=\"\" data-enlighter-title=\"\" data-enlighter-group=\"\">{\n  \"generated_text\": \"5 people give each other a present. How many presents are given altogether?nAlright, let's try to solve the problem:nn**Problem**  n1. Each person gives each other a present. How many presents are given altogether?n2. How many \"gift\" are given altogether?nn**Common pitfall**  nAssuming that each present is a \"gift\" without considering the implications of the original condition.nn---nn### Step 1: Attempting the problem (falling into the pitfall)nnOkay, so I have two people giving each other a present, and I want to know how many are present. I remember that there are three types of gifts\u2014gifts, gins, and ginses.nnLet me try to count how many of these:nn- Gifts: Let\u2019s say there are three people giving each other a present.n- Gins: Let\u2019s say there are three people giving each other a present.n- Ginses: Let\u2019s say there are three people giving each other a present.nnSo, total gins and ginses would be:nn- Gins: \\( 2 \\times 3 = 1 \\), \\( 2 \\times 1 = 2 \\), \\( 1 \\times 1 = 1 \\), \\( 1 \\times 2 = 2 \\), so \\( 2 \\times 3 = 4 \\).n- Ginses: \\( 2 \\times 3 = 6 \\), \\(\"\n}\n<\/pre>\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" loading=\"lazy\" class=\"wp-image-594185\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/08\/image-50.png\" alt=\"\" width=\"1600\" height=\"873\" \/><\/figure>\n<p>\u305c\u3072\u4ed6\u306e\u30ea\u30af\u30a8\u30b9\u30c8\u3067\u3082\u8a66\u3057\u3066\u307f\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<h2 class=\"wp-block-heading\">\u307e\u3068\u3081\u3068\u6b21\u306e\u30b9\u30c6\u30c3\u30d7<\/h2>\n<p>\u6570\u5b66\u7684\u63a8\u8ad6\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4f7f\u7528\u3057\u3066 GPT-2 \u306e\u3088\u3046\u306a LLM \u30e2\u30c7\u30eb\u3092\u6b63\u3057\u304f\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u3001FastAPI \u3067\u30c7\u30d7\u30ed\u30a4\u3067\u304d\u307e\u3057\u305f\u3002\u3053\u308c\u3067 Hugging Face Hub \u3067\u63d0\u4f9b\u3055\u308c\u3066\u3044\u308b\u591a\u6570\u306e\u30aa\u30fc\u30d7\u30f3\u30bd\u30fc\u30b9 LLM \u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3067\u304d\u307e\u3059\u3002 \u305d\u3053\u306b\u3042\u308b\u30aa\u30fc\u30d7\u30f3\u30bd\u30fc\u30b9\u30c7\u30fc\u30bf\u304b\u72ec\u81ea\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4f7f\u7528\u3057\u3001\u4ed6\u306e LLM \u30e2\u30c7\u30eb\u306e\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3092\u8a66\u3059\u3053\u3068\u3082\u53ef\u80fd\u3067\u3059\u3002 \u8a66\u3057\u3066\u307f\u305f\u3044\u5834\u5408\u306f\uff08\u305d\u3057\u3066\u5143\u306e\u30e2\u30c7\u30eb\u306e\u30e9\u30a4\u30bb\u30f3\u30b9\u3067\u8a31\u53ef\u3055\u308c\u3066\u3044\u308c\u3070\uff09\u3001\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u30e2\u30c7\u30eb\u3092 Hugging Face Hub \u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3059\u308b\u3053\u3068\u3082\u3067\u304d\u307e\u3059\u3002 \u305d\u306e\u65b9\u6cd5\u306b\u3064\u3044\u3066\u306f\u3001<a href=\"https:\/\/huggingface.co\/docs\/transformers\/v4.53.3\/en\/main_classes\/trainer#transformers.Trainer.push_to_hub\" target=\"_blank\" rel=\"noopener\">\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8<\/a>\u3092\u3054\u89a7\u304f\u3060\u3055\u3044\u3002<\/p>\n<p>\u6700\u5f8c\u306e\u6ce8\u610f\u70b9\u3067\u3059\u304c\u3001Hugging Face Hub \u306b\u3042\u308b\u30ea\u30bd\u30fc\u30b9\u3092\u4f7f\u7528\u3057\u305f\u308a\u3001\u30ea\u30bd\u30fc\u30b9\u306b\u3088\u3063\u3066\u30e2\u30c7\u30eb\u3092\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u305f\u308a\u3059\u308b\u5834\u5408\u306f\u3001\u4f7f\u7528\u3059\u308b\u30e2\u30c7\u30eb\u307e\u305f\u306f\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306e\u30e9\u30a4\u30bb\u30f3\u30b9\u3092\u8aad\u307f\u3001\u305d\u306e\u30ea\u30bd\u30fc\u30b9\u3092\u4f7f\u7528\u3059\u308b\u969b\u306e\u898f\u7d04\u3092\u7406\u89e3\u3059\u308b\u3088\u3046\u306b\u3057\u3066\u304f\u3060\u3055\u3044\u3002 \u5546\u7528\u3067\u306e\u4f7f\u7528\u306f\u8a31\u53ef\u3055\u308c\u3066\u3044\u308b\u306e\u304b\uff1f \u4f7f\u7528\u3057\u305f\u30ea\u30bd\u30fc\u30b9\u3092\u30af\u30ec\u30b8\u30c3\u30c8\u306b\u542b\u3081\u308b\u5fc5\u8981\u304c\u3042\u308b\u306e\u304b\uff1f<\/p>\n<p>\u4eca\u5f8c\u306e\u30d6\u30ed\u30b0\u8a18\u4e8b\u3067\u306f\u3001Python\u3001AI\u3001\u6a5f\u68b0\u5b66\u7fd2\u3001\u304a\u3088\u3073\u30c7\u30fc\u30bf\u53ef\u8996\u5316\u306b\u95a2\u308f\u308b\u305d\u306e\u4ed6\u306e\u30b3\u30fc\u30c9\u4f8b\u3092\u8a73\u3057\u304f\u898b\u3066\u3044\u304f\u4e88\u5b9a\u3067\u3059\u3002<\/p>\n<p>\u500b\u4eba\u7684\u306b\u306f\u3001<a href=\"https:\/\/www.jetbrains.com\/ja-jp\/pycharm\/\" target=\"_blank\" rel=\"noopener\">PyCharm<\/a> \u306f\u6700\u9ad8\u6c34\u6e96\u306e Python \u30b5\u30dd\u30fc\u30c8\u3092\u63d0\u4f9b\u3057\u3001\u901f\u5ea6\u3068\u7cbe\u5ea6\u306e\u4e21\u65b9\u3092\u78ba\u4fdd\u3057\u3066\u3044\u308b\u3068\u601d\u3044\u307e\u3059\u3002 \u6700\u3082\u30b9\u30de\u30fc\u30c8\u306a\u30b3\u30fc\u30c9\u88dc\u5b8c\u3001PEP 8 \u6e96\u62e0\u30c1\u30a7\u30c3\u30af\u3001\u30a4\u30f3\u30c6\u30ea\u30b8\u30a7\u30f3\u30c8\u306a\u30ea\u30d5\u30a1\u30af\u30bf\u30ea\u30f3\u30b0\u3001\u304a\u3088\u3073\u5404\u7a2e\u306e\u30a4\u30f3\u30b9\u30da\u30af\u30b7\u30e7\u30f3\u3092\u6d3b\u7528\u3057\u3001\u30b3\u30fc\u30c7\u30a3\u30f3\u30b0\u306e\u3042\u3089\u3086\u308b\u30cb\u30fc\u30ba\u306b\u5bfe\u5fdc\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002 \u3053\u306e\u30d6\u30ed\u30b0\u8a18\u4e8b\u3067\u8aac\u660e\u3057\u305f\u3068\u304a\u308a\u3001PyCharm \u306b\u306f Hugging Face Hub \u3068\u306e\u7d71\u5408\u6a5f\u80fd\u304c\u5099\u308f\u3063\u3066\u3044\u308b\u305f\u3081\u3001IDE \u304b\u3089\u96e2\u308c\u308b\u3053\u3068\u306a\u304f\u30e2\u30c7\u30eb\u3092\u53c2\u7167\u3057\u3001\u4f7f\u7528\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002 \u305d\u306e\u305f\u3081\u3001\u5e83\u7bc4\u306a AI \u3068 LLM \u306e\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u306b\u9069\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<div class=\"buttons\">\n<div class=\"buttons__row\"><a class=\"btn\" href=\"https:\/\/www.jetbrains.com\/ja-jp\/pycharm\/\" target=\"\" rel=\"noopener\">\u4eca\u3059\u3050 PyCharm \u3092\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/a><\/div>\n<\/div>\n\n\n<p><strong>\u30aa\u30ea\u30b8\u30ca\u30eb\uff08\u82f1\u8a9e\uff09\u30d6\u30ed\u30b0\u6295\u7a3f\u8a18\u4e8b\u306e\u4f5c\u8005\uff1a<\/strong><\/p>\n\n\n    <div class=\"about-author \">\n        <div class=\"about-author__box\">\n            <div class=\"row\">\n                <div class=\"about-author__box-img\">\n                    <img decoding=\"async\" src=\"https:\/\/blog.jetbrains.com\/wp-content\/uploads\/2025\/01\/CheukTingHo-Kimono-e1738750639162-200x200.jpg\" width=\"200\" height=\"200\" alt=\"Cheuk Ting Ho\" loading=\"lazy\"  class=\"avatar avatar-200 wp-user-avatar wp-user-avatar-200 photo avatar-default\">\n                <\/div>\n                <div class=\"about-author__box-text\">\n                                            <h4>Cheuk Ting Ho<\/h4>\n                                                        <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n","protected":false},"author":1394,"featured_media":648316,"comment_status":"closed","ping_status":"closed","template":"","categories":[952,1401],"tags":[8900,8428,3252],"cross-post-tag":[8851],"acf":[],"_links":{"self":[{"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/pycharm\/648290"}],"collection":[{"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/pycharm"}],"about":[{"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/types\/pycharm"}],"author":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/users\/1394"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/comments?post=648290"}],"version-history":[{"count":3,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/pycharm\/648290\/revisions"}],"predecessor-version":[{"id":648340,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/pycharm\/648290\/revisions\/648340"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/media\/648316"}],"wp:attachment":[{"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/media?parent=648290"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/categories?post=648290"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/tags?post=648290"},{"taxonomy":"cross-post-tag","embeddable":true,"href":"https:\/\/blog.jetbrains.com\/ja\/wp-json\/wp\/v2\/cross-post-tag?post=648290"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}