{"id":29614,"date":"2024-10-28T01:55:44","date_gmt":"2024-10-28T01:55:44","guid":{"rendered":"http:\/\/atmokpo.com\/w\/?p=29614"},"modified":"2024-11-26T06:52:04","modified_gmt":"2024-11-26T06:52:04","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-m2m100-%eb%b2%88%ec%97%ad-%ec%86%8c%ec%8a%a4%eb%a1%9c-%ec%a4%91","status":"publish","type":"post","link":"https:\/\/atmokpo.com\/w\/29614\/","title":{"rendered":"\ud5c8\uae45\ud398\uc774\uc2a4 \ud2b8\ub80c\uc2a4\ud3ec\uba38 \ud65c\uc6a9\uac15\uc88c, M2M100 \ubc88\uc5ed \uc18c\uc2a4\ub85c \uc911\uad6d\uc5b4 \ud14d\uc2a4\ud2b8 \uc900\ube44"},"content":{"rendered":"

<\/p>\n

\n
\n

\uc791\uc131\uc77c:

\n

1. \uc11c\ub860<\/h2>\n

\n \ucd5c\uadfc \uc778\uacf5\uc9c0\ub2a5 \ubd84\uc57c\uc5d0\uc11c \uc790\uc5f0\uc5b4 \ucc98\ub9ac(NLP)\ub294 \ub9ce\uc740 \uc8fc\ubaa9\uc744 \ubc1b\uace0 \uc788\uc2b5\ub2c8\ub2e4. \ud2b9\ud788, \ubc88\uc5ed \uae30\uc220\uc740 \uae00\ub85c\ubc8c \uc18c\ud1b5\uc744 \uac00\ub2a5\ud558\uac8c \ud558\ub294 \uc911\uc694\ud55c \uc694\uc18c\ub85c \uc790\ub9ac\uc7a1\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uc774 \uae00\uc5d0\uc11c\ub294 Hugging Face\uc758 Transformers \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \ud65c\uc6a9\ud558\uc5ec M2M100 \ubaa8\ub378\uc744 \ubc14\ud0d5\uc73c\ub85c \uc911\uad6d\uc5b4 \ud14d\uc2a4\ud2b8\ub97c \uc900\ube44\ud558\uace0 \ubc88\uc5ed\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \ub2e4\ub8f9\ub2c8\ub2e4. M2M100 \ubaa8\ub378\uc740 \ub2e4\uc591\ud55c \uc5b8\uc5b4 \uac04\uc758 \ubc88\uc5ed\uc744 \uc9c0\uc6d0\ud558\ub294 \ub2e4\uad6d\uc5b4 \ubc88\uc5ed \ubaa8\ub378\ub85c, \ud2b9\ud788 \uc911\uad6d\uc5b4\uc640 \uac19\uc740 \ubcf5\uc7a1\ud55c \uc5b8\uc5b4\ub97c \ucc98\ub9ac\ud558\ub294 \ub370 \uac15\uc810\uc744 \ubcf4\uc785\ub2c8\ub2e4.\n <\/p>\n<\/section>\n

\n

2. M2M100 \ubaa8\ub378 \uac1c\uc694<\/h2>\n

\n M2M100 \ubaa8\ub378\uc740 Facebook AI Research(Facebook AI)\uc5d0\uc11c \uac1c\ubc1c\ud55c \ub2e4\uad6d\uc5b4 \uae30\uacc4 \ubc88\uc5ed \ubaa8\ub378\uc785\ub2c8\ub2e4. \uc774 \ubaa8\ub378\uc740 100\uac1c \uc774\uc0c1\uc758 \uc5b8\uc5b4\ub97c \uc11c\ub85c \ubc88\uc5ed\ud560 \uc218 \uc788\ub3c4\ub85d \uc124\uacc4\ub418\uc5c8\uc73c\uba70, Transformer \uc544\ud0a4\ud14d\ucc98\uc5d0 \uae30\ubc18\ud569\ub2c8\ub2e4. \ud2b9\ubcc4\ud55c \uc810\uc740 M2M100\uc774 \ud2b9\uc815 \uc5b8\uc5b4 \uc30d\uc5d0 \uc758\uc874\ud558\uc9c0 \uc54a\uace0, \uc9c1\uc811 \ubc88\uc5ed\ud560 \uc218 \uc788\ub2e4\ub294 \uac83\uc785\ub2c8\ub2e4. \uc989, \uc601\uc5b4\ub97c \uc911\uc7ac\ud558\uc9c0 \uc54a\uace0\ub3c4 \uc911\uad6d\uc5b4\ub85c \uc9c1\uc811 \ubc88\uc5ed\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n

\n M2M100\uc740 \ub450 \uac00\uc9c0 \uc8fc\uc694 \uad6c\uc131 \uc694\uc18c\ub85c \uc774\ub8e8\uc5b4\uc838 \uc788\uc2b5\ub2c8\ub2e4: \uc778\ucf54\ub354\uc640 \ub514\ucf54\ub354. \uc778\ucf54\ub354\ub294 \uc785\ub825 \ubb38\uc7a5\uc744 \uc22b\uc790 \ubca1\ud130\ub85c \ubcc0\ud658\ud558\uace0, \ub514\ucf54\ub354\ub294 \uc774 \ubca1\ud130\ub97c \uae30\ubc18\uc73c\ub85c \ucd9c\ub825 \ubb38\uc7a5\uc744 \uc0dd\uc131\ud569\ub2c8\ub2e4. \uc774 \uacfc\uc815\uc740 Transformer \uc544\ud0a4\ud14d\ucc98\ub97c \ud1b5\ud574 \uc774\ub8e8\uc5b4\uc9c0\uba70, \uc778\ucf54\ub354-\ub514\ucf54\ub354 \uc544\ud0a4\ud14d\ucc98\ub294 \uae30\uacc4 \ubc88\uc5ed \uc2dc\uc2a4\ud15c\uc5d0\uc11c \uc8fc\uc694\ud55c \uc5ed\ud560\uc744 \ud569\ub2c8\ub2e4.\n <\/p>\n<\/section>\n

\n

3. \uc124\uce58 \ubc0f \uc900\ube44<\/h2>\n

\n \uc774 \uac15\uc88c\ub97c \uc9c4\ud589\ud558\uae30 \uc704\ud574\uc11c\ub294 Python\uacfc \uba87 \uac00\uc9c0 \ud544\uc218 \ub77c\uc774\ube0c\ub7ec\ub9ac\uac00 \uc124\uce58\ub418\uc5b4 \uc788\uc5b4\uc57c \ud569\ub2c8\ub2e4. Hugging Face\uc758 Transformers \ub77c\uc774\ube0c\ub7ec\ub9ac\uc640 PyTorch\ub97c \uc0ac\uc6a9\ud560 \uac83\uc785\ub2c8\ub2e4. \ub2e4\uc74c\uc740 \uc124\uce58 \ubc29\ubc95\uc785\ub2c8\ub2e4:\n <\/p>\n

\n                \n                    pip install transformers torch\n                <\/code>\n            <\/pre>\n
\n                \uc704 \uba85\ub839\uc5b4\ub97c \ud130\ubbf8\ub110\uc5d0 \uc785\ub825\ud558\uc5ec \ud544\uc694\ud55c \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc124\uce58\ud569\ub2c8\ub2e4.\n            <\/p>\n<\/section>\n
\n
4. \ub370\uc774\ud130\uc14b \uc900\ube44<\/h2>\n
\n                \ubc88\uc5ed \ubaa8\ub378\uc744 \ud559\uc2b5\ud558\uae30 \uc704\ud574\uc11c\ub294 \uc801\uc808\ud55c \ub370\uc774\ud130\uc14b\uc774 \ud544\uc694\ud569\ub2c8\ub2e4. \ubcf8 \ud504\ub85c\uc81d\ud2b8\uc5d0\uc11c\ub294 \uc0ac\uc6a9\ud560 \uc911\uad6d\uc5b4 \ubb38\uc7a5\uc744 \uc900\ube44\ud558\ub824\uace0 \ud569\ub2c8\ub2e4. \ub2e4\uc74c\uc740 \uac04\ub2e8\ud55c \uc911\uad6d\uc5b4 \ubb38\uc7a5\uc744 \ud3ec\ud568\ud558\ub294 \ub370\uc774\ud130\uc14b\uc744 \ub9cc\ub4dc\ub294 \ubc29\ubc95\uc785\ub2c8\ub2e4.\n            <\/p>\n
\n                \n                    # \uc911\uad6d\uc5b4 \ubb38\uc7a5 \ub9ac\uc2a4\ud2b8\n                    chinese_sentences = [\n                        \"\u4f60\u597d\uff0c\u4e16\u754c\uff01\",\n                        \"\u4eca\u5929\u7684\u5929\u6c14\u5f88\u597d\u3002\",\n                        \"\u6211\u559c\u6b22\u5b66\u4e60\u6df1\u5ea6\u5b66\u4e60\u3002\",\n                        \"\u4eba\u5de5\u667a\u80fd\u6b63\u5728\u6539\u53d8\u6211\u4eec\u7684\u751f\u6d3b\u3002\",\n                        \"\u4f60\u60f3\u5403\u4ec0\u4e48\uff1f\",\n                    ]\n                <\/code>\n            <\/pre>\n
\n                \uc704 \ucf54\ub4dc\uc5d0\uc11c\ub294 \uac04\ub2e8\ud55c \uc911\uad6d\uc5b4 \ubb38\uc7a5 5\uac1c\ub97c \ub9ac\uc2a4\ud2b8 \ud615\ud0dc\ub85c \uc815\uc758\ud558\uc600\uc2b5\ub2c8\ub2e4. \uc2e4\uc81c \ud504\ub85c\uc81d\ud2b8\uc5d0\uc11c\ub294 \ub354 \ub9ce\uc740 \ub370\uc774\ud130\uc14b\uc774 \ud544\uc694\ud569\ub2c8\ub2e4.\n            <\/p>\n<\/section>\n
\n
5. \ubaa8\ub378 \ub85c\ub4dc \ubc0f \ubc88\uc5ed<\/h2>\n
\n                \uc774\uc81c \uc900\ube44\ud55c \ub370\uc774\ud130\uc14b\uc744 \uc0ac\uc6a9\ud558\uc5ec M2M100 \ubaa8\ub378\uc744 \ud1b5\ud574 \ubc88\uc5ed \uc791\uc5c5\uc744 \uc218\ud589\ud574 \ubcf4\uaca0\uc2b5\ub2c8\ub2e4. Hugging Face\uc758 Transformers \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \ud1b5\ud574 \ubaa8\ub378\uc744 \uac04\ub2e8\ud788 \ub85c\ub4dc\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \ub2e4\uc74c\uc740 M2M100 \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc911\uad6d\uc5b4 \ubb38\uc7a5\uc744 \ubc88\uc5ed\ud558\ub294 \uc608\uc81c\uc785\ub2c8\ub2e4.\n            <\/p>\n
\n                \n                    from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer\n\n                    # \ubaa8\ub378\uacfc \ud1a0\ud06c\ub098\uc774\uc800 \ub85c\ub4dc\n                    model_name = \"facebook\/m2m100_418M\"\n                    model = M2M100ForConditionalGeneration.from_pretrained(model_name)\n                    tokenizer = M2M100Tokenizer.from_pretrained(model_name)\n\n                    def translate(text, target_lang=\"en\"):\n                        tokenizer.src_lang = \"zh\"  # \ucd9c\ubc1c \uc5b8\uc5b4\ub97c \uc911\uad6d\uc5b4\ub85c \uc124\uc815\n                        encoded = tokenizer(text, return_tensors=\"pt\")\n                        generated_tokens = model.generate(**encoded, forced_bos_token_id=tokenizer.get_lang_id(target_lang))\n                        return tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)\n\n                    # \ubc88\uc5ed \uc218\ud589\n                    for sentence in chinese_sentences:\n                        translated = translate(sentence)\n                        print(f\"\uc6d0\ubb38: {sentence} -> \ubc88\uc5ed: {translated[0]}\")\n                <\/code>\n            <\/pre>\n
\n                \uc704 \ucf54\ub4dc\uc5d0\uc11c \ubaa8\ub378\uacfc \ud1a0\ud06c\ub098\uc774\uc800\ub97c \ub85c\ub4dc\ud55c \ud6c4, \uac01\uac01\uc758 \ubb38\uc7a5\uc744 \ubc88\uc5ed\ud558\ub294 \ud568\uc218\ub97c \uc815\uc758\ud558\uc600\uc2b5\ub2c8\ub2e4. `translate` \ud568\uc218\uc5d0\uc11c\ub294 \ucd9c\ubc1c \uc5b8\uc5b4\ub97c \uc911\uad6d\uc5b4(`zh`)\ub85c \uc124\uc815\ud558\uace0, \uc6d0\ud558\ub294 \ubaa9\ud45c \uc5b8\uc5b4\ub85c \ubc88\uc5ed\ub41c \ubb38\uc7a5\uc744 \ucd9c\ub825\ud569\ub2c8\ub2e4.\n            <\/p>\n<\/section>\n
\n
6. \ucd9c\ub825 \uacb0\uacfc \ud655\uc778<\/h2>\n
\n                \uc704\uc758 \ucf54\ub4dc \uc2e4\ud589 \uacb0\uacfc\ub294 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:\n            <\/p>\n
\n                \n                    \uc6d0\ubb38: \u4f60\u597d\uff0c\u4e16\u754c\uff01 -> \ubc88\uc5ed: Hello, world!\n                    \uc6d0\ubb38: \u4eca\u5929\u7684\u5929\u6c14\u5f88\u597d\u3002 -> \ubc88\uc5ed: The weather is nice today.\n                    \uc6d0\ubb38: \u6211\u559c\u6b22\u5b66\u4e60\u6df1\u5ea6\u5b66\u4e60\u3002 -> \ubc88\uc5ed: I like to study deep learning.\n                    \uc6d0\ubb38: \u4eba\u5de5\u667a\u80fd\u6b63\u5728\u6539\u53d8\u6211\u4eec\u7684\u751f\u6d3b\u3002 -> \ubc88\uc5ed: Artificial intelligence is changing our lives.\n                    \uc6d0\ubb38: \u4f60\u60f3\u5403\u4ec0\u4e48\uff1f -> \ubc88\uc5ed: What do you want to eat?\n                <\/code>\n            <\/pre>\n
\n                \uc704\uc640 \uac19\uc774 \uac01\uac01\uc758 \ubb38\uc7a5\uc774 \uc131\uacf5\uc801\uc73c\ub85c \ubc88\uc5ed\ub418\uc5c8\uc2b5\ub2c8\ub2e4. \ubc88\uc5ed \uacb0\uacfc\ub294 \ubaa8\ub378\uc758 \uc131\ub2a5\uc5d0 \ub530\ub77c \ub2e4\ub97c \uc218 \uc788\uc73c\uba70, \uc774 \uc608\uc81c\uc5d0\uc11c\ub294 M2M100 \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc911\uad6d\uc5b4 \ubb38\uc7a5\uc744 \uc601\uc5b4\ub85c \ubc88\uc5ed\ud558\ub294 \uacfc\uc815\uc744 \ubcf4\uc5ec\uc8fc\uc5c8\uc2b5\ub2c8\ub2e4.\n            <\/p>\n<\/section>\n
\n
7. \uc2e4\uc2b5 \ubc0f \uc751\uc6a9<\/h2>\n
\n                \uc5b8\uc5b4 \ubc88\uc5ed\uc740 \ub2e4\uc591\ud55c \ubd84\uc57c\uc5d0 \uc751\uc6a9\ub420 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc608\ub97c \ub4e4\uc5b4, \ub2e4\uad6d\uc801 \uae30\uc5c5\uc758 \uc6f9\uc0ac\uc774\ud2b8 \ubc88\uc5ed, \uc5ec\ud589 \uac00\uc774\ub4dc \ubc88\uc5ed, \uace0\uac1d \uc9c0\uc6d0 \uc11c\ube44\uc2a4 \ub4f1\uc5d0\uc11c \ud65c\uc6a9\ub420 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \ub610\ud55c, M2M100 \ubaa8\ub378\uc740 \ub2e4\uc591\ud55c \uc5b8\uc5b4 \uc30d\uc5d0 \ub300\ud574 \uc9c0\uc6d0\ud558\ubbc0\ub85c, \uc911\uac04 \uc5b8\uc5b4 \uc5c6\uc774 \uc9c1\uc811 \ubc88\uc5ed\ud558\uc5ec \ub354 \uc790\uc5f0\uc2a4\ub7ec\uc6b4 \uacb0\uacfc\ub97c \uc5bb\uc744 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n            <\/p>\n
\n                \ucd94\uac00 \uc2e4\uc2b5\uc73c\ub85c, \ub2e4\ub978 \uc5b8\uc5b4(\uc608: \ud55c\uad6d\uc5b4, \uc77c\ubcf8\uc5b4)\ub85c \ubc88\uc5ed\ud558\ub294 \uc791\uc5c5\uc744 \ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc774\ub97c \uc704\ud574 `translate` \ud568\uc218\uc758 `target_lang` \ub9e4\uac1c\ubcc0\uc218\uc758 \uac12\uc744 \ubcc0\uacbd\ud558\uc5ec \uc6d0\ud558\ub294 \uc5b8\uc5b4\ub85c \ubc88\uc5ed\ud558\uba74 \ub429\ub2c8\ub2e4. \uc544\ub798 \ucf54\ub4dc\ub294 \ud55c\uad6d\uc5b4\ub85c \ubc88\uc5ed\ud558\ub294 \ubc29\ubc95\uc744 \ubcf4\uc5ec\uc90d\ub2c8\ub2e4.\n            <\/p>\n
\n                \n                    # \ud55c\uad6d\uc5b4\ub85c \ubc88\uc5ed\ud558\uae30\n                    for sentence in chinese_sentences:\n                        translated = translate(sentence, target_lang=\"ko\")\n                        print(f\"\uc6d0\ubb38: {sentence} -> \ubc88\uc5ed: {translated[0]}\")\n                <\/code>\n            <\/pre>\n<\/section>\n
\n
8. \ub9c8\ubb34\ub9ac<\/h2>\n
\n                \uc774\ubc88 \uac15\uc88c\uc5d0\uc11c\ub294 Hugging Face\uc758 Transformers \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \ud65c\uc6a9\ud558\uc5ec M2M100 \ubaa8\ub378\ub85c \uc911\uad6d\uc5b4 \ubb38\uc7a5\uc744 \uc900\ube44\ud558\uace0 \ubc88\uc5ed\ud558\ub294 \ubc29\ubc95\uc744 \uc54c\uc544\ubcf4\uc558\uc2b5\ub2c8\ub2e4. \ubc88\uc5ed \uae30\uc220\uc740 \uc55e\uc73c\ub85c\ub3c4 \ub9ce\uc740 \ubc1c\uc804\uc774 \uc608\uc0c1\ub418\uba70, \ub2e4\uc591\ud55c \ubaa8\ub378\uacfc \uc54c\uace0\ub9ac\uc998\uc774 \uc5f0\uad6c\ub418\uace0 \uac1c\ubc1c\ub420 \uac83\uc785\ub2c8\ub2e4. \uc774\uc640 \uac19\uc740 \ub525\ub7ec\ub2dd \ubaa8\ub378\uc744 \ud65c\uc6a9\ud558\uc5ec \ub2e4\uad6d\uc5b4 \ubc88\uc5ed\uc758 \ud6a8\uc728\uc131\uc744 \ub192\uc5ec\ubcf4\uc138\uc694.\n            <\/p>\n
\n                \ubcf8 \uae00\uc744 \ud1b5\ud574 \ub525\ub7ec\ub2dd\uacfc \uc790\uc5f0\uc5b4\ucc98\ub9ac\uc5d0 \ub300\ud55c \uc774\ud574\uac00 \ub354\uc6b1 \uae4a\uc5b4\uc9c0\uae30\ub97c \ubc14\ub77c\uba70, \uc5ec\ub7ec\ubd84\ub3c4 \uc9c1\uc811 \uc2e4\uc2b5\ud574\ubcf4\uc2dc\uae30\ub97c \uad8c\uc7a5\ud569\ub2c8\ub2e4. \ub2e4\uc74c \uac15\uc88c\uc5d0\uc11c\ub3c4 \ub354 \ub9ce\uc740 \ub525\ub7ec\ub2dd \uae30\ubc95\uacfc \uc2e4\uc2b5\uc744 \ub2e4\ub8f0 \uc608\uc815\uc774\ub2c8 \ub9ce\uc740 \uad00\uc2ec \ubd80\ud0c1\ub4dc\ub9bd\ub2c8\ub2e4.\n            <\/p>\n<\/section>\n