{"id":11491,"date":"2026-04-14T22:55:10","date_gmt":"2026-04-14T13:55:10","guid":{"rendered":"https:\/\/www.stechstar.com\/user\/wordpress\/?p=11491"},"modified":"2026-08-27T18:13:35","modified_gmt":"2026-08-27T09:13:35","slug":"%ec%9d%b8%ea%b3%b5%ec%a7%80%eb%8a%a5-%ea%b0%9c%eb%b0%9c%ec%9e%90-llm%ec%9d%84-%ec%84%9c%eb%b9%99%ed%95%98%eb%8a%94-%ed%94%84%eb%a0%88%ec%9e%84%ec%9b%8c%ed%81%ac-vllm-%ec%82%ac%ec%9a%a9%eb%b2%95","status":"publish","type":"post","link":"https:\/\/www.stechstar.com\/user\/wordpress\/%ec%9d%b8%ea%b3%b5%ec%a7%80%eb%8a%a5-%ea%b0%9c%eb%b0%9c%ec%9e%90-llm%ec%9d%84-%ec%84%9c%eb%b9%99%ed%95%98%eb%8a%94-%ed%94%84%eb%a0%88%ec%9e%84%ec%9b%8c%ed%81%ac-vllm-%ec%82%ac%ec%9a%a9%eb%b2%95\/","title":{"rendered":"[\uc778\uacf5\uc9c0\ub2a5 \uac1c\ubc1c\uc790] LLM\uc744 \uc11c\ube59\ud558\ub294 \ud504\ub808\uc784\uc6cc\ud06c, vLLM \uc0ac\uc6a9\ubc95"},"content":{"rendered":"<h1>[\uc778\uacf5\uc9c0\ub2a5 \uac1c\ubc1c\uc790] LLM\uc744 \uc11c\ube59\ud558\ub294 \ud504\ub808\uc784\uc6cc\ud06c, vLLM \uc0ac\uc6a9\ubc95<\/h1>\n<h2 id=\"%EA%B0%9C%EC%9A%94-1\" data-ke-size=\"size26\"><a href=\"https:\/\/ariz1623.tistory.com\/375#%EA%B0%9C%EC%9A%94-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#%EA%B0%9C%EC%9A%94-1\">\uac1c\uc694<\/a><\/h2>\n<p data-ke-size=\"size16\">\ucd5c\uadfc&nbsp;\ub300\ud615&nbsp;\uc5b8\uc5b4&nbsp;\ubaa8\ub378(LLM)\uc744&nbsp;\uc2e4\uc81c&nbsp;\uc11c\ube44\uc2a4\uc5d0&nbsp;\uc801\uc6a9\ud558\ub824\ub294&nbsp;\uc218\uc694\uac00&nbsp;\uc99d\uac00\ud558\uba74\uc11c,&nbsp;LLM\uc744&nbsp;\ud6a8\uc728\uc801\uc73c\ub85c&nbsp;\uc11c\ube59\ud558\ub294&nbsp;\uae30\uc220\uc758&nbsp;\uc911\uc694\uc131\ub3c4&nbsp;\ucee4\uc9c0\uace0&nbsp;\uc788\uc2b5\ub2c8\ub2e4.&nbsp;\ub300\ud45c\uc801\uc778&nbsp;\uc11c\ube59&nbsp;\ud504\ub808\uc784\uc6cc\ud06c\ub85c\ub294&nbsp;SGLang,&nbsp;TensorRT-LLM,&nbsp;vLLM&nbsp;\ub4f1\uc774&nbsp;\uc788\uc73c\uba70,&nbsp;\uc774\ub4e4&nbsp;\uac01\uac01\uc740&nbsp;\uc131\ub2a5&nbsp;\ud2b9\uc131\uc5d0&nbsp;\ucc28\uc774\ub97c&nbsp;\ubcf4\uc785\ub2c8\ub2e4.<\/p>\n<p>\uc774&nbsp;\uc911&nbsp;vLLM\uc740&nbsp;\ucd08\uae30\ubd80\ud130&nbsp;\uc18d\ub3c4\uc640&nbsp;\uba54\ubaa8\ub9ac&nbsp;\ud6a8\uc728\uc131&nbsp;\uba74\uc5d0\uc11c&nbsp;\uc548\uc815\uc801\uc778&nbsp;\uad6c\uc870\ub97c&nbsp;\uac16\ucd98&nbsp;\ud504\ub808\uc784\uc6cc\ud06c\ub85c&nbsp;\uc8fc\ubaa9\uc744&nbsp;\ubc1b\uc558\uace0,&nbsp;\ud604\uc7ac&nbsp;\uac00\uc7a5&nbsp;\ub110\ub9ac&nbsp;\ud65c\uc6a9\ub418\uace0&nbsp;\uc788\ub294&nbsp;LLM&nbsp;\uc11c\ube59&nbsp;\uc194\ub8e8\uc158&nbsp;\uc911&nbsp;\ud558\ub098\uc785\ub2c8\ub2e4<\/p>\n<h4 data-ke-size=\"size20\">vLLM\uc758 \ud2b9\uc9d5<\/h4>\n<ul data-ke-list-type=\"disc\">\n<li>PagedAttention: \uae30\uc874 \uc2dc\ud000\uc2a4 \uae30\ubc18 Attention \ubc29\uc2dd \ub300\uc2e0 \ud398\uc774\uc9c0 \uae30\ubc18\uc758 \uc720\uc5f0\ud55c \uba54\ubaa8\ub9ac \ud560\ub2f9 \uc804\ub7b5\uc744 \ub3c4\uc785\ud558\uc5ec&nbsp;\ubc30\uce58 \ud6a8\uc728 \uadf9\ub300\ud654<\/li>\n<li>\ube44\ub3d9\uae30 \uc5d4\uc9c4 \uad6c\uc870: \ucd94\ub860 \uc694\uccad\uc744 \ud6a8\uc728\uc801\uc73c\ub85c \ucc98\ub9ac\ud558\uae30 \uc704\ud55c&nbsp;\ube44\ub3d9\uae30 \ucc98\ub9ac&nbsp;\uad6c\uc870\ub85c \ub192\uc740 \ucc98\ub9ac\ub7c9 \uc720\uc9c0<\/li>\n<li>OpenAI \ud638\ud658 API \uc9c0\uc6d0: \uae30\uc874 OpenAI API \ubc29\uc2dd\uc73c\ub85c\ub3c4 \uc27d\uac8c \uc11c\ube59 \uac00\ub2a5<\/li>\n<\/ul>\n<h4 data-ke-size=\"size20\">&nbsp;<\/h4>\n<h4 data-ke-size=\"size20\">\uc18d\ub3c4 \ube44\uad50<\/h4>\n<figure class=\"imageblock alignCenter\" data-ke-mobilestyle=\"widthOrigin\" data-origin-height=\"867\" data-origin-width=\"1280\"><span data-phocus=\"https:\/\/blog.kakaocdn.net\/dna\/L6y0P\/btsNMe9109B\/AAAAAAAAAAAAAAAAAAAAAHC1lU9lTByZ5htWK6lzu6V4lnvahlqrlpkbdmmBjG1-\/img.png?credential=yqXZFxpELC7KVnFOS48ylbz2pIh7yKj8&amp;expires=1777561199&amp;allow_ip=&amp;allow_referer=&amp;signature=%2BEvkj%2BLQU7sUQP28OM%2B3wamPDlo%3D\" data-url=\"https:\/\/blog.kakaocdn.net\/dna\/L6y0P\/btsNMe9109B\/AAAAAAAAAAAAAAAAAAAAAHC1lU9lTByZ5htWK6lzu6V4lnvahlqrlpkbdmmBjG1-\/img.png?credential=yqXZFxpELC7KVnFOS48ylbz2pIh7yKj8&amp;expires=1777561199&amp;allow_ip=&amp;allow_referer=&amp;signature=%2BEvkj%2BLQU7sUQP28OM%2B3wamPDlo%3D\"><img loading=\"lazy\" decoding=\"async\" class=\"lazyautosizes lazyloaded\" src=\"https:\/\/www.stechstar.com\/user\/zbxe\/files\/attach\/images\/112516\/334\/113\/56faca6386cdf24e28ad03a45cbf6e2f.png\" sizes=\"auto, 671px\" srcset=\"https:\/\/img1.daumcdn.net\/thumb\/R1280x0\/?scode=mtistory2&amp;fname=https%3A%2F%2Fblog.kakaocdn.net%2Fdna%2FL6y0P%2FbtsNMe9109B%2FAAAAAAAAAAAAAAAAAAAAAHC1lU9lTByZ5htWK6lzu6V4lnvahlqrlpkbdmmBjG1-%2Fimg.png%3Fcredential%3DyqXZFxpELC7KVnFOS48ylbz2pIh7yKj8%26expires%3D1777561199%26allow_ip%3D%26allow_referer%3D%26signature%3D%252BEvkj%252BLQU7sUQP28OM%252B3wamPDlo%253D\" width=\"671\" height=\"455\" data-autoattach=\"success\" data-evernote-lazy-src=\".\/files\/attach\/images\/112516\/334\/113\/56faca6386cdf24e28ad03a45cbf6e2f.png\" data-origin-height=\"867\" data-origin-width=\"1280\" data-sizes=\"auto\" data-src=\".\/files\/attach\/images\/112516\/334\/113\/56faca6386cdf24e28ad03a45cbf6e2f.png\" data-srcset=\"https:\/\/img1.daumcdn.net\/thumb\/R1280x0\/?scode=mtistory2&amp;fname=https%3A%2F%2Fblog.kakaocdn.net%2Fdna%2FL6y0P%2FbtsNMe9109B%2FAAAAAAAAAAAAAAAAAAAAAHC1lU9lTByZ5htWK6lzu6V4lnvahlqrlpkbdmmBjG1-%2Fimg.png%3Fcredential%3DyqXZFxpELC7KVnFOS48ylbz2pIh7yKj8%26expires%3D1777561199%26allow_ip%3D%26allow_referer%3D%26signature%3D%252BEvkj%252BLQU7sUQP28OM%252B3wamPDlo%253D\" data-cke-saved-src=\".\/files\/attach\/images\/112516\/334\/113\/56faca6386cdf24e28ad03a45cbf6e2f.png\"><\/span><\/figure>\n<hr contenteditable=\"false\" data-ke-style=\"style6\" data-ke-type=\"horizontalRule\">\n<h2 id=\"vLLM%20Quick%20Start%C2%A0-1\" data-ke-size=\"size26\"><a href=\"https:\/\/ariz1623.tistory.com\/375#vLLM%20Quick%20Start%C2%A0-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#vLLM%20Quick%20Start%C2%A0-1\">vLLM Quick Start&nbsp;<\/a><\/h2>\n<h3 id=\"Offline%20Batched%20Inference-1\" data-ke-size=\"size23\"><a href=\"https:\/\/ariz1623.tistory.com\/375#Offline%20Batched%20Inference-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#Offline%20Batched%20Inference-1\">Offline Batched Inference<\/a><\/h3>\n<h4 data-ke-size=\"size20\">\uc124\uce58<\/h4>\n<pre class=\"cmake\"><code class=\"hljs language-cmake\">pip <span class=\"hljs-keyword\">install<\/span> vllm<\/code><\/pre>\n<h4 data-ke-size=\"size20\">&nbsp;<\/h4>\n<h4 data-ke-size=\"size20\">\ubaa8\ub378 \ub85c\ub529 \ubc0f \ucd94\ub860<\/h4>\n<pre class=\"python\" data-ke-language=\"python\"><code class=\"hljs language-python\"><span class=\"hljs-keyword\">from<\/span> vllm <span class=\"hljs-keyword\">import<\/span> LLM, SamplingParams\n\nmodel_name =  <span class=\"hljs-string\">\"kakaocorp\/kanana-nano-2.1b-base\"<\/span> <span class=\"hljs-comment\"># \uc6d0\ud558\ub294 \ubaa8\ub378 \uc774\ub984<\/span>\n\nllm = LLM(model=model_name) \n\nprompts = [\n    <span class=\"hljs-string\">\"\ud55c\uad6d\uc758 \uc218\ub3c4\ub294 \uc5b4\ub514\uc778\uac00\uc694?\"<\/span>,\n    <span class=\"hljs-string\">\"\uc11c\uc6b8 \uc5ec\ud589\uc9c0 3\uacf3\uc744 \uc18c\uac1c\ud574\uc8fc\uc138\uc694.\"<\/span>\n]\n\nsampling_params = SamplingParams(temperature=<span class=\"hljs-number\">0.7<\/span>, top_p=<span class=\"hljs-number\">0.9<\/span>, max_tokens=<span class=\"hljs-number\">100<\/span>)\n\n<span class=\"hljs-comment\"># \ub2f5\ubcc0 \uc0dd\uc131<\/span>\noutputs = llm.generate(prompts, sampling_params)\n\n<span class=\"hljs-comment\"># \ub2f5\ubcc0 \ucd9c\ub825<\/span>\n<span class=\"hljs-keyword\">for<\/span> output <span class=\"hljs-keyword\">in<\/span> outputs:\n    prompt = output.prompt\n    generated_text = output.outputs[<span class=\"hljs-number\">0<\/span>].text\n    <span class=\"hljs-built_in\">print<\/span>(<span class=\"hljs-string\">f\"Prompt: <span class=\"hljs-subst\">{prompt!r}<\/span>, Generated text: <span class=\"hljs-subst\">{generated_text!r}<\/span>\"<\/span>)<\/code><\/pre>\n<p data-ke-size=\"size16\">\ub85c\uceec\uc5d0\uc11c&nbsp;\ubaa8\ub378\uc744&nbsp;\uc9c1\uc811&nbsp;\ub85c\ub4dc\ud574&nbsp;\uc0ac\uc6a9\ud560&nbsp;\uc218&nbsp;\uc788\uc73c\uba70,&nbsp;SamplingParams\ub85c&nbsp;\ub514\ucf54\ub529&nbsp;\uc804\ub7b5\uc744&nbsp;\uc870\uc808\ud560&nbsp;\uc218&nbsp;\uc788\uc2b5\ub2c8\ub2e4.<\/p>\n<h3 id=\"OpenAI%C2%A0Completions%C2%A0API%C2%A0with%C2%A0vLLM-1\" data-ke-size=\"size23\"><a href=\"https:\/\/ariz1623.tistory.com\/375#OpenAI%C2%A0Completions%C2%A0API%C2%A0with%C2%A0vLLM-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#OpenAI%C2%A0Completions%C2%A0API%C2%A0with%C2%A0vLLM-1\">OpenAI&nbsp;Completions&nbsp;API&nbsp;with&nbsp;vLLM<\/a><\/h3>\n<p data-ke-size=\"size16\">vLLM\uc740 OpenAI API \uc2a4\ud0c0\uc77c\uc758 \uc11c\ubc84\ub3c4 \uc81c\uacf5\ud569\ub2c8\ub2e4. \uc774\ub97c \ud1b5\ud574 \uae30\uc874 \ucf54\ub4dc \ubcc0\uacbd \uc5c6\uc774\ub3c4 \uc27d\uac8c \uc11c\ube59 \uc778\ud504\ub77c\uc5d0 \uc5f0\uacb0\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n<h4 data-ke-size=\"size20\">\uc11c\ubc84 \uc2e4\ud589<\/h4>\n<pre class=\"python\" data-ke-language=\"python\"><code class=\"hljs language-python\">vllm serve Qwen\/Qwen3-<span class=\"hljs-number\">1.7<\/span>B-Instruct<\/code><\/pre>\n<h4 data-ke-size=\"size20\">&nbsp;<\/h4>\n<h4 data-ke-size=\"size20\">\ud074\ub77c\uc774\uc5b8\ud2b8 \uc608\uc81c(curl)<\/h4>\n<pre id=\"code_1746488059986\" class=\"bash\" data-ke-language=\"bash\" data-ke-type=\"codeblock\"><code class=\"hljs language-bash\">curl http:\/\/localhost:8000\/v1\/completions \\\n    -H <span class=\"hljs-string\">\"Content-Type: application\/json\"<\/span> \\\n    -d <span class=\"hljs-string\">'{\n        \"model\": \"Qwen\/Qwen3-1.7B-Instruct\",\n        \"prompt\": \"\ud55c\uad6d\uc758 \uc218\ub3c4\ub294 \uc5b4\ub514\uc778\uac00\uc694?\",\n        \"max_tokens\": 7,\n        \"temperature\": 0\n    }'<\/span><\/code><\/pre>\n<h4 data-ke-size=\"size20\">&nbsp;<\/h4>\n<h4 data-ke-size=\"size20\">\ud074\ub77c\uc774\uc5b8\ud2b8 \uc608\uc81c (python)<\/h4>\n<pre class=\"python\" data-ke-language=\"python\"><code class=\"hljs language-python\"><span class=\"hljs-keyword\">import<\/span> openai\n\nopenai.api_base = <span class=\"hljs-string\">\"http:\/\/localhost:8000\/v1\"<\/span>\n\nresponse = openai.ChatCompletion.create(\n    model=<span class=\"hljs-string\">\"Qwen\/Qwen3-1.7B-Instruct\"<\/span>,\n    messages=[\n        {<span class=\"hljs-string\">\"role\"<\/span>: <span class=\"hljs-string\">\"user\"<\/span>, <span class=\"hljs-string\">\"content\"<\/span>: <span class=\"hljs-string\">\"Tell me a joke about AI.\"<\/span>}\n    ],\n    temperature=<span class=\"hljs-number\">0.7<\/span>,\n    max_tokens=<span class=\"hljs-number\">100<\/span>\n)\n\n<span class=\"hljs-built_in\">print<\/span>(response[<span class=\"hljs-string\">'choices'<\/span>][<span class=\"hljs-number\">0<\/span>][<span class=\"hljs-string\">'message'<\/span>][<span class=\"hljs-string\">'content'<\/span>])<\/code><\/pre>\n<p data-ke-size=\"size16\">\uc2e4\uc81c&nbsp;OpenAI\uc640&nbsp;\uac70\uc758&nbsp;\ub3d9\uc77c\ud55c&nbsp;\uc778\ud130\ud398\uc774\uc2a4\ub97c&nbsp;\uc81c\uacf5\ud558\ubbc0\ub85c,&nbsp;API&nbsp;\uc804\ud658\uc774&nbsp;\uac04\ud3b8\ud569\ub2c8\ub2e4.<\/p>\n<h2 id=\"vLLM%20%EC%A3%BC%EC%9A%94%20%ED%8C%8C%EB%9D%BC%EB%AF%B8%ED%84%B0-1\" data-ke-size=\"size26\"><a href=\"https:\/\/ariz1623.tistory.com\/375#vLLM%20%EC%A3%BC%EC%9A%94%20%ED%8C%8C%EB%9D%BC%EB%AF%B8%ED%84%B0-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#vLLM%20%EC%A3%BC%EC%9A%94%20%ED%8C%8C%EB%9D%BC%EB%AF%B8%ED%84%B0-1\">vLLM \uc8fc\uc694 \ud30c\ub77c\ubbf8\ud130<\/a><\/h2>\n<h4 data-ke-size=\"size20\">\uc5d4\uc9c4 \ucd08\uae30\ud654 \ud30c\ub77c\ubbf8\ud130<\/h4>\n<table class=\" cke_show_border\" data-ke-align=\"alignLeft\">\n<thead>\n<tr>\n<th>\ud30c\ub77c\ubbf8\ud130<\/th>\n<th>\uc124\uba85<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>gpu_memory_utilization<\/code><\/td>\n<td>GPU \uba54\ubaa8\ub9ac \uc0ac\uc6a9 \ube44\uc728. \uae30\ubcf8\uac12\uc740 0.9\ub85c, GPU \uba54\ubaa8\ub9ac 90%\uae4c\uc9c0\ub9cc \uc0ac\uc6a9\ud558\uc5ec \uc548\uc815\uc131 \ud655\ubcf4<\/td>\n<\/tr>\n<tr>\n<td><code>max_model_len<\/code><\/td>\n<td>\ud558\ub098\uc758 \uc785\ub825 \uc2dc\ud000\uc2a4\uc5d0\uc11c \ucc98\ub9ac\ud560 \uc218 \uc788\ub294 \ucd5c\ub300 \ud1a0\ud070 \uae38\uc774. (ex: 2048, 4096 \ub4f1)<\/td>\n<\/tr>\n<tr>\n<td><code>tensor_parallel_size<\/code><\/td>\n<td>\ubaa8\ub378\uc744 \uc5ec\ub7ec GPU\uc5d0 \ub098\ub220\uc11c \ubcd1\ub82c\ud654\ud560 \ub54c \uc0ac\uc6a9. ex) 2\ub85c \uc124\uc815\ud558\uba74 2\uac1c GPU \uc0ac\uc6a9<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4 data-ke-size=\"size20\">\ucd94\ub860 \ud30c\ub77c\ubbf8\ud130<\/h4>\n<table class=\" cke_show_border\" data-ke-align=\"alignLeft\">\n<thead>\n<tr>\n<th>\ud30c\ub77c\ubbf8\ud130<\/th>\n<th>\uc124\uba85<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>temperature<\/code><\/td>\n<td>\uc0d8\ud50c\ub9c1\uc758 \ubb34\uc791\uc704\uc131 \uc815\ub3c4. \ub0ae\uc744\uc218\ub85d \uc77c\uc815\ud558\uace0, \ub192\uc744\uc218\ub85d \ubb34\uc791\uc704\uc801(\ucc3d\uc758\ub825\uc774 \ub192\uc544\uc9d0)<\/td>\n<\/tr>\n<tr>\n<td><code>top_p<\/code><\/td>\n<td>\uc0c1\uc704 p%\uc758 \ud1a0\ud070\ub9cc \uace0\ub824\ud558\uc5ec \uc0d8\ud50c\ub9c1<\/td>\n<\/tr>\n<tr>\n<td><code>top_k<\/code><\/td>\n<td>\ud655\ub960 \uc0c1\uc704 k\uac1c\uc758 \ud6c4\ubcf4\ub9cc \uace0\ub824\ud558\uc5ec \uc0d8\ud50c\ub9c1<\/td>\n<\/tr>\n<tr>\n<td><code>max_tokens<\/code><\/td>\n<td>\uc0dd\uc131\ud560 \ucd5c\ub300 \ud1a0\ud070 \uc218<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2 id=\"%EB%A7%88%EB%AC%B4%EB%A6%AC-1\" data-ke-size=\"size26\"><a href=\"https:\/\/ariz1623.tistory.com\/375#%EB%A7%88%EB%AC%B4%EB%A6%AC-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#%EB%A7%88%EB%AC%B4%EB%A6%AC-1\">\ub9c8\ubb34\ub9ac<\/a><\/h2>\n<p data-ke-size=\"size16\">vLLM\uc740 \uac04\ub2e8\ud55c \uc124\uce58\uc640 \ub192\uc740 \uc131\ub2a5, \uadf8\ub9ac\uace0 OpenAI API\uc640\uc758 \ud638\ud658\uc131 \ub355\ubd84\uc5d0 LLM \uc11c\ube59\uc758 \uc9c4\uc785 \uc7a5\ubcbd\uc744 \ud06c\uac8c \ub0ae\ucdb0\uc8fc\ub294 \ub3c4\uad6c\uc785\ub2c8\ub2e4. \ud2b9\ud788&nbsp;\uc11c\ube59 \ud6a8\uc728\uacfc&nbsp;\uba54\ubaa8\ub9ac \ucd5c\uc801\ud654\uac00 \uc911\uc694\ud55c \uc2e4\uc2dc\uac04 \uc560\ud50c\ub9ac\ucf00\uc774\uc158\uc5d0\uc11c \ud070 \uac15\uc810\uc744 \ubc1c\ud718\ud569\ub2c8\ub2e4.<\/p>\n<h2 id=\"Reference-1\" data-ke-size=\"size26\"><a href=\"https:\/\/ariz1623.tistory.com\/375#Reference-1\" data-cke-saved-href=\"https:\/\/ariz1623.tistory.com\/375#Reference-1\">Reference<\/a><\/h2>\n<ul data-ke-list-type=\"disc\">\n<li><a href=\"https:\/\/docs.vllm.ai\/en\/v0.7.1\/getting_started\/quickstart.html\" target=\"_blank\" rel=\"noopener&nbsp;noreferrer\" data-cke-saved-href=\"https:\/\/docs.vllm.ai\/en\/v0.7.1\/getting_started\/quickstart.html\">https:\/\/docs.vllm.ai\/en\/v0.7.1\/getting_started\/quickstart.html<\/a><\/li>\n<li><a href=\"https:\/\/blog.vllm.ai\/2024\/09\/05\/perf-update.html?utm_source=chatgpt.com#appendix\" target=\"_blank\" rel=\"noopener&nbsp;noreferrer\" data-cke-saved-href=\"https:\/\/blog.vllm.ai\/2024\/09\/05\/perf-update.html?utm_source=chatgpt.com#appendix\">https:\/\/blog.vllm.ai\/2024\/09\/05\/perf-update.html?utm_source=chatgpt.com#appendix<\/a><\/li>\n<\/ul>\n<p>[\ucd9c\ucc98]&nbsp;https:\/\/ariz1623.tistory.com\/375<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_11491\" class=\"pvc_stats all  \" data-element-id=\"11491\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg aria-hidden=\"true\" focusable=\"false\" data-prefix=\"far\" data-icon=\"chart-bar\" role=\"img\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" viewBox=\"0 0 512 512\" class=\"svg-inline--fa fa-chart-bar fa-w-16 fa-2x\"><path fill=\"currentColor\" d=\"M396.8 352h22.4c6.4 0 12.8-6.4 12.8-12.8V108.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v230.4c0 6.4 6.4 12.8 12.8 12.8zm-192 0h22.4c6.4 0 12.8-6.4 12.8-12.8V140.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v198.4c0 6.4 6.4 12.8 12.8 12.8zm96 0h22.4c6.4 0 12.8-6.4 12.8-12.8V204.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v134.4c0 6.4 6.4 12.8 12.8 12.8zM496 400H48V80c0-8.84-7.16-16-16-16H16C7.16 64 0 71.16 0 80v336c0 17.67 14.33 32 32 32h464c8.84 0 16-7.16 16-16v-16c0-8.84-7.16-16-16-16zm-387.2-48h22.4c6.4 0 12.8-6.4 12.8-12.8v-70.4c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v70.4c0 6.4 6.4 12.8 12.8 12.8z\" class=\"\"><\/path><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/www.stechstar.com\/user\/wordpress\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>[\uc778\uacf5\uc9c0\ub2a5 \uac1c\ubc1c\uc790] LLM\uc744 \uc11c\ube59\ud558\ub294 \ud504\ub808\uc784\uc6cc\ud06c, vLLM \uc0ac\uc6a9\ubc95 \uac1c\uc694 \ucd5c\uadfc&nbsp;\ub300\ud615&nbsp;\uc5b8\uc5b4&nbsp;\ubaa8\ub378(LLM)\uc744&nbsp;\uc2e4\uc81c&nbsp;\uc11c\ube44\uc2a4\uc5d0&nbsp;\uc801\uc6a9\ud558\ub824\ub294&nbsp;\uc218\uc694\uac00&nbsp;\uc99d\uac00\ud558\uba74\uc11c,&nbsp;LLM\uc744&nbsp;\ud6a8\uc728\uc801\uc73c\ub85c&nbsp;\uc11c\ube59\ud558\ub294&nbsp;\uae30\uc220\uc758&nbsp;\uc911\uc694\uc131\ub3c4&nbsp;\ucee4\uc9c0\uace0&nbsp;\uc788\uc2b5\ub2c8\ub2e4.&nbsp;\ub300\ud45c\uc801\uc778&nbsp;\uc11c\ube59&nbsp;\ud504\ub808\uc784\uc6cc\ud06c\ub85c\ub294&nbsp;SGLang,&nbsp;TensorRT-LLM,&nbsp;vLLM&nbsp;\ub4f1\uc774&nbsp;\uc788\uc73c\uba70,&nbsp;\uc774\ub4e4&nbsp;\uac01\uac01\uc740&nbsp;\uc131\ub2a5&nbsp;\ud2b9\uc131\uc5d0&nbsp;\ucc28\uc774\ub97c&nbsp;\ubcf4\uc785\ub2c8\ub2e4. \uc774&nbsp;\uc911&nbsp;vLLM\uc740&nbsp;\ucd08\uae30\ubd80\ud130&nbsp;\uc18d\ub3c4\uc640&nbsp;\uba54\ubaa8\ub9ac&nbsp;\ud6a8\uc728\uc131&nbsp;\uba74\uc5d0\uc11c&nbsp;\uc548\uc815\uc801\uc778&nbsp;\uad6c\uc870\ub97c&nbsp;\uac16\ucd98&nbsp;\ud504\ub808\uc784\uc6cc\ud06c\ub85c&nbsp;\uc8fc\ubaa9\uc744&nbsp;\ubc1b\uc558\uace0,&nbsp;\ud604\uc7ac&nbsp;\uac00\uc7a5&nbsp;\ub110\ub9ac&nbsp;\ud65c\uc6a9\ub418\uace0&nbsp;\uc788\ub294&nbsp;LLM&nbsp;\uc11c\ube59&nbsp;\uc194\ub8e8\uc158&nbsp;\uc911&nbsp;\ud558\ub098\uc785\ub2c8\ub2e4 vLLM\uc758 \ud2b9\uc9d5 PagedAttention: \uae30\uc874 \uc2dc\ud000\uc2a4 \uae30\ubc18 Attention \ubc29\uc2dd \ub300\uc2e0 \ud398\uc774\uc9c0 \uae30\ubc18\uc758 \uc720\uc5f0\ud55c \uba54\ubaa8\ub9ac \ud560\ub2f9 \uc804\ub7b5\uc744 \ub3c4\uc785\ud558\uc5ec&nbsp;\ubc30\uce58 \ud6a8\uc728 \uadf9\ub300\ud654 \ube44\ub3d9\uae30 \uc5d4\uc9c4 \uad6c\uc870: \ucd94\ub860 \uc694\uccad\uc744 \ud6a8\uc728\uc801\uc73c\ub85c \ucc98\ub9ac\ud558\uae30 \uc704\ud55c&nbsp;\ube44\ub3d9\uae30 \ucc98\ub9ac&nbsp;\uad6c\uc870\ub85c \ub192\uc740 \ucc98\ub9ac\ub7c9 \uc720\uc9c0 OpenAI \ud638\ud658 API \uc9c0\uc6d0: \uae30\uc874 OpenAI API \ubc29\uc2dd\uc73c\ub85c\ub3c4 \uc27d\uac8c \uc11c\ube59 \uac00\ub2a5 &nbsp; \uc18d\ub3c4 \ube44\uad50 vLLM [&hellip;]<\/p>\n<div class=\"pvc_clear\"><\/div>\n<p id=\"pvc_stats_11491\" class=\"pvc_stats all  \" data-element-id=\"11491\" style=\"\"><i class=\"pvc-stats-icon medium\" aria-hidden=\"true\"><svg aria-hidden=\"true\" focusable=\"false\" data-prefix=\"far\" data-icon=\"chart-bar\" role=\"img\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" viewBox=\"0 0 512 512\" class=\"svg-inline--fa fa-chart-bar fa-w-16 fa-2x\"><path fill=\"currentColor\" d=\"M396.8 352h22.4c6.4 0 12.8-6.4 12.8-12.8V108.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v230.4c0 6.4 6.4 12.8 12.8 12.8zm-192 0h22.4c6.4 0 12.8-6.4 12.8-12.8V140.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v198.4c0 6.4 6.4 12.8 12.8 12.8zm96 0h22.4c6.4 0 12.8-6.4 12.8-12.8V204.8c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v134.4c0 6.4 6.4 12.8 12.8 12.8zM496 400H48V80c0-8.84-7.16-16-16-16H16C7.16 64 0 71.16 0 80v336c0 17.67 14.33 32 32 32h464c8.84 0 16-7.16 16-16v-16c0-8.84-7.16-16-16-16zm-387.2-48h22.4c6.4 0 12.8-6.4 12.8-12.8v-70.4c0-6.4-6.4-12.8-12.8-12.8h-22.4c-6.4 0-12.8 6.4-12.8 12.8v70.4c0 6.4 6.4 12.8 12.8 12.8z\" class=\"\"><\/path><\/svg><\/i> <img loading=\"lazy\" decoding=\"async\" width=\"16\" height=\"16\" alt=\"Loading\" src=\"https:\/\/www.stechstar.com\/user\/wordpress\/wp-content\/plugins\/page-views-count\/ajax-loader-2x.gif\" border=0 \/><\/p>\n<div class=\"pvc_clear\"><\/div>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ngg_post_thumbnail":0,"footnotes":""},"categories":[21,24,20],"tags":[],"class_list":["post-11491","post","type-post","status-publish","format-standard","hentry","category-21","category-24","category-20"],"a3_pvc":{"activated":true,"total_views":13,"today_views":0},"_links":{"self":[{"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/posts\/11491","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/comments?post=11491"}],"version-history":[{"count":4,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/posts\/11491\/revisions"}],"predecessor-version":[{"id":11996,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/posts\/11491\/revisions\/11996"}],"wp:attachment":[{"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/media?parent=11491"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/categories?post=11491"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.stechstar.com\/user\/wordpress\/wp-json\/wp\/v2\/tags?post=11491"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}