{"id":29580,"date":"2024-10-28T01:55:33","date_gmt":"2024-10-28T01:55:33","guid":{"rendered":"http:\/\/atmokpo.com\/w\/?p=29580"},"modified":"2024-11-26T06:52:12","modified_gmt":"2024-11-26T06:52:12","slug":"%ed%97%88%ea%b9%85%ed%8e%98%ec%9d%b4%ec%8a%a4-%ed%8a%b8%eb%a0%8c%ec%8a%a4%ed%8f%ac%eb%a8%b8-%ed%99%9c%ec%9a%a9%ea%b0%95%ec%a2%8c-dialogpt-%eb%ac%b8%ec%9e%a5-%ec%83%9d%ec%84%b1","status":"publish","type":"post","link":"https:\/\/atmokpo.com\/w\/29580\/","title":{"rendered":"\ud5c8\uae45\ud398\uc774\uc2a4 \ud2b8\ub80c\uc2a4\ud3ec\uba38 \ud65c\uc6a9\uac15\uc88c, DialoGPT \ubb38\uc7a5 \uc0dd\uc131"},"content":{"rendered":"

<\/p>\n

\uc624\ub298\ub0a0 \uc778\uacf5\uc9c0\ub2a5\uc758 \uac00\uc7a5 \ube60\ub974\uac8c \uc131\uc7a5\ud558\ub294 \ubd84\uc57c \uc911 \ud558\ub098\ub294 \uc790\uc5f0\uc5b4 \ucc98\ub9ac(NLP)\uc785\ub2c8\ub2e4. \ub2e4\uc591\ud55c \uc5b8\uc5b4 \ubaa8\ub378\ub4e4\uc774 \ubc1c\uc804\ud558\uba74\uc11c, \ud14d\uc2a4\ud2b8 \uc0dd\uc131, \uc9c8\ubb38 \uc751\ub2f5 \uc2dc\uc2a4\ud15c, \uac10\uc815 \ubd84\uc11d \ub4f1\uc758 \ubd84\uc57c\uc5d0\uc11c \ud65c\uc6a9\ub418\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uadf8 \uc911\uc5d0\uc11c\ub3c4 \ud5c8\uae45\ud398\uc774\uc2a4(Hugging Face)<\/strong>\uc758 \ud2b8\ub79c\uc2a4\ud3ec\uba38(Transformers)<\/strong> \ub77c\uc774\ube0c\ub7ec\ub9ac\ub294 \ub525\ub7ec\ub2dd \uae30\ubc18\uc758 \uac15\ub825\ud55c NLP \ubaa8\ub378\ub4e4\uc744 \uc190\uc27d\uac8c \uc0ac\uc6a9\ud560 \uc218 \uc788\ub3c4\ub85d \ub3c4\uc640\uc90d\ub2c8\ub2e4.<\/p>\n

1. \ud5c8\uae45\ud398\uc774\uc2a4 \ud2b8\ub79c\uc2a4\ud3ec\uba38\ub780?<\/h2>\n

\ud5c8\uae45\ud398\uc774\uc2a4 \ud2b8\ub79c\uc2a4\ud3ec\uba38 \ub77c\uc774\ube0c\ub7ec\ub9ac\ub294 BERT, GPT-2, T5 \ub4f1\uc758 \ub2e4\uc591\ud55c \ud604\uc5c5\uc5d0\uc11c \ub9ce\uc774 \uc0ac\uc6a9\ub418\ub294 \uc0ac\uc804 \ud559\uc2b5(pre-trained) NLP \ubaa8\ub378\ub4e4\uc744 \uc81c\uacf5\ud569\ub2c8\ub2e4. \uc774 \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc0ac\uc6a9\ud558\uba74 \uba87 \uc904\uc758 \ucf54\ub4dc\ub85c \ubcf5\uc7a1\ud55c \ubaa8\ub378\uc744 \uc190\uc27d\uac8c \ubd88\ub7ec\uc624\uace0 \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n

2. DialoGPT \uc18c\uac1c<\/h2>\n

DialoGPT<\/strong>\ub294 OpenAI\uc758 GPT-2 \ubaa8\ub378\uc744 \uae30\ubc18\uc73c\ub85c \ud558\ub294 \ub300\ud654\ud615 \ubaa8\ub378\ub85c, \ud2b9\ud788 \ubb38\uc7a5 \uc0dd\uc131, \ub300\ud654 \uc0dd\uc131\uc5d0 \ud2b9\ud654\ub418\uc5b4 \uc788\uc2b5\ub2c8\ub2e4. \ub300\ud654 \ub370\uc774\ud130\ub97c \ud559\uc2b5\ud558\uc5ec \uc0ac\ub78c\uacfc \uc720\uc0ac\ud55c \uc790\uc5f0\uc2a4\ub7ec\uc6b4 \ub300\ud654\ub97c \uc0dd\uc131\ud560 \uc218 \uc788\ub294 \ub2a5\ub825\uc744 \uac00\uc9c0\uace0 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n

3. DialoGPT \uc124\uce58<\/h2>\n

\uba3c\uc800, DialoGPT \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uae30 \uc704\ud574 \ud544\uc694\ud55c \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc124\uce58\ud574\uc57c \ud569\ub2c8\ub2e4. \uc544\ub798\uc640 \uac19\uc740 \uba85\ub839\uc5b4\ub85c transformers<\/code> \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc124\uce58\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n

pip install transformers torch<\/code><\/pre>\n
4. \uac04\ub2e8\ud55c \uc608\uc81c: DialoGPT \ubaa8\ub378 \ub85c\ub4dc \ubc0f \ubb38\uc7a5 \uc0dd\uc131<\/h2>\n
\uc774\uc81c DialoGPT\ub97c \uc0ac\uc6a9\ud558\uc5ec \uac04\ub2e8\ud55c \ubb38\uc7a5\uc744 \uc0dd\uc131\ud574 \ubcf4\uaca0\uc2b5\ub2c8\ub2e4. \uc544\ub798\uc758 \ucf54\ub4dc\ub97c \ud1b5\ud574 \ubaa8\ub378\uc744 \ub85c\ub4dc\ud558\uace0, \uc0ac\uc6a9\uc790\ub85c\ubd80\ud130 \uc785\ub825\uc744 \ubc1b\uc544 \uadf8\uc5d0 \ub300\ud55c \uc751\ub2f5\uc744 \uc0dd\uc131\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
4.1 \ucf54\ub4dc \uc608\uc81c<\/h3>\n
\nfrom transformers import AutoModelForCausalLM, AutoTokenizer\nimport torch\n\n# \ubaa8\ub378\uacfc \ud1a0\ud06c\ub098\uc774\uc800 \ubd88\ub7ec\uc624\uae30\nmodel_name = \"microsoft\/DialoGPT-small\"\ntokenizer = AutoTokenizer.from_pretrained(model_name)\nmodel = AutoModelForCausalLM.from_pretrained(model_name)\n\n# \ub300\ud654 \ud788\uc2a4\ud1a0\ub9ac \ucd08\uae30\ud654\nchat_history_ids = None\n\nwhile True:\n    # \uc0ac\uc6a9\uc790 \uc785\ub825 \ubc1b\uae30\n    user_input = input(\"\uc0ac\uc6a9\uc790: \")\n    \n    # \uc785\ub825 \ud14d\uc2a4\ud2b8 \ud1a0\ud06c\ub098\uc774\uc988\n    new_input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors='pt')\n\n    # \ub300\ud654 \ud788\uc2a4\ud1a0\ub9ac \uc5c5\ub370\uc774\ud2b8\n    if chat_history_ids is not None:\n        bot_input_ids = torch.cat([chat_history_ids, new_input_ids], dim=-1)\n    else:\n        bot_input_ids = new_input_ids\n\n    # \uc751\ub2f5 \uc0dd\uc131\n    chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id)\n\n    # \uc751\ub2f5 \ub514\ucf54\ub4dc\n    bot_response = tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)\n    print(\"Bot: \", bot_response)\n    <\/code><\/pre>\n
4.2 \ucf54\ub4dc \uc124\uba85<\/h3>\n
\uc704 \ucf54\ub4dc\uc5d0\uc11c\ub294 DialoGPT-small \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \uac04\ub2e8\ud55c \ub300\ud654 \uc2dc\uc2a4\ud15c\uc744 \uad6c\ucd95\ud569\ub2c8\ub2e4. \uc8fc\uc694 \ubd80\ubd84\uc740 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:<\/p>\n