diffusers/pr_13881/ko/_app/immutable/nodes/25.DNKpw8Oh.js

import"../chunks/DsnmJJEf.js";import{i as b,h as g,C as U,H as e,a as l,E as T,s as Z}from"../chunks/CFM6C53a.js";import{p as _,o as G,s,f as I,a as y,b as k,c as w,n as z}from"../chunks/CNc7KuUZ.js";const E='{"title":"여러 GPU를 사용한 분산 추론","local":"여러-gpu를-사용한-분산-추론","sections":[{"title":"🤗 Accelerate","local":"-accelerate","sections":[],"depth":2},{"title":"Pytoerch 분산","local":"pytoerch-분산","sections":[],"depth":2}],"depth":1}';var X=w('<meta name="hf:doc:metadata"/>'),R=w('<p></p> <!> <!> <p>분산 설정에서는 여러 개의 프롬프트를 동시에 생성할 때 유용한 🤗 <a href="https://huggingface.co/docs/accelerate/index" rel="nofollow">Accelerate</a> 또는 <a href="https://pytorch.org/tutorials/beginner/dist_overview.html" rel="nofollow">PyTorch Distributed</a>를 사용하여 여러 GPU에서 추론을 실행할 수 있습니다.</p> <p>이 가이드에서는 분산 추론을 위해 🤗 Accelerate와 PyTorch Distributed를 사용하는 방법을 보여드립니다.</p> <!> <p>🤗 <a href="https://huggingface.co/docs/accelerate/index" rel="nofollow">Accelerate</a>는 분산 설정에서 추론을 쉽게 훈련하거나 실행할 수 있도록 설계된 라이브러리입니다. 분산 환경 설정 프로세스를 간소화하여 PyTorch 코드에 집중할 수 있도록 해줍니다.</p> <p>시작하려면 Python 파일을 생성하고 <code>accelerate.PartialState</code>를 초기화하여 분산 환경을 생성하면, 설정이 자동으로 감지되므로 <code>rank</code> 또는 <code>world_size</code>를 명시적으로 정의할 필요가 없습니다. [‘DiffusionPipeline`]을 <code>distributed_state.device</code>로 이동하여 각 프로세스에 GPU를 할당합니다.</p> <p>이제 컨텍스트 관리자로 <code>split_between_processes</code> 유틸리티를 사용하여 프로세스 수에 따라 프롬프트를 자동으로 분배합니다.</p> <!> <p>Use the <code>--num_processes</code> argument to specify the number of GPUs to use, and call <code>accelerate launch</code> to run the script:</p> <!> <blockquote class="tip"><p>자세한 내용은 <a href="https://huggingface.co/docs/accelerate/en/usage_guides/distributed_inference#distributed-inference-with-accelerate" rel="nofollow">🤗 Accelerate를 사용한 분산 추론</a> 가이드를 참조하세요.</p></blockquote> <!> <p>PyTorch는 데이터 병렬 처리를 가능하게 하는 <a href="https://pytorch.org/docs/stable/generated/torch.nn.parallel.DistributedDataParallel.html" rel="nofollow"><code>DistributedDataParallel</code></a>을 지원합니다.</p> <p>시작하려면 Python 파일을 생성하고 <code>torch.distributed</code> 및 <code>torch.multiprocessing</code>을 임포트하여 분산 프로세스 그룹을 설정하고 각 GPU에서 추론용 프로세스를 생성합니다. 그리고 <code>DiffusionPipeline</code>도 초기화해야 합니다:</p> <p>확산 파이프라인을 <code>rank</code>로 이동하고 <code>get_rank</code>를 사용하여 각 프로세스에 GPU를 할당하면 각 프로세스가 다른 프롬프트를 처리합니다:</p> <!> <p>사용할 백엔드 유형, 현재 프로세스의 <code>rank</code>, <code>world_size</code> 또는 참여하는 프로세스 수로 분산 환경 생성을 처리하는 함수<code>init_process_group</code>를 만들어 추론을 실행해야 합니다.</p> <p>2개의 GPU에서 추론을 병렬로 실행하는 경우 <code>world_size</code>는 2입니다.</p> <!> <p>분산 추론을 실행하려면 <a href="https://pytorch.org/docs/stable/multiprocessing.html#torch.multiprocessing.spawn" rel="nofollow"><code>mp.spawn</code></a>을 호출하여 <code>world_size</code>에 정의된 GPU 수에 대해 <code>run_inference</code> 함수를 실행합니다:</p> <!> <p>추론 스크립트를 완료했으면 <code>--nproc_per_node</code> 인수를 사용하여 사용할 GPU 수를 지정하고 <code>torchrun</code>을 호출하여 스크립트를 실행합니다:</p> <!> <!> <p></p>',1);function A(J,j){_(j,!1),G(()=>{new URLSearchParams(window.location.search).get("fw")}),b();var a=R();g("el5ux9",u=>{var m=X();Z(m,"content",E),y(u,m)});var t=s(I(a),2);U(t,{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"});var o=s(t,2);e(o,{title:"여러 GPU를 사용한 분산 추론",local:"여러-gpu를-사용한-분산-추론",headingTag:"h1"});var c=s(o,6);e(c,{title:"🤗 Accelerate",local:"-accelerate",headingTag:"h2"});var n=s(c,8);l(n,{code:"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",highlighted:`<span class="hljs-keyword">from</span> accelerate <span class="hljs-keyword">import</span> PartialState
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline

pipeline = DiffusionPipeline.from_pretrained(<span class="hljs-string">&quot;stable-diffusion-v1-5/stable-diffusion-v1-5&quot;</span>, torch_dtype=torch.float16)
distributed_state = PartialState()
pipeline.to(distributed_state.device)

<span class="hljs-keyword">with</span> distributed_state.split_between_processes([<span class="hljs-string">&quot;a dog&quot;</span>, <span class="hljs-string">&quot;a cat&quot;</span>]) <span class="hljs-keyword">as</span> prompt:
    result = pipeline(prompt).images[<span class="hljs-number">0</span>]
    result.save(<span class="hljs-string">f&quot;result_<span class="hljs-subst">{distributed_state.process_index}</span>.png&quot;</span>)`,lang:"py",wrap:!1});var r=s(n,4);l(r,{code:"YWNjZWxlcmF0ZSUyMGxhdW5jaCUyMHJ1bl9kaXN0cmlidXRlZC5weSUyMC0tbnVtX3Byb2Nlc3NlcyUzRDI=",highlighted:"accelerate launch run_distributed.py --num_processes=2",lang:"bash",wrap:!1});var p=s(r,4);e(p,{title:"Pytoerch 분산",local:"pytoerch-분산",headingTag:"h2"});var i=s(p,8);l(i,{code:"aW1wb3J0JTIwdG9yY2glMEFpbXBvcnQlMjB0b3JjaC5kaXN0cmlidXRlZCUyMGFzJTIwZGlzdCUwQWltcG9ydCUyMHRvcmNoLm11bHRpcHJvY2Vzc2luZyUyMGFzJTIwbXAlMEElMEFmcm9tJTIwZGlmZnVzZXJzJTIwaW1wb3J0JTIwRGlmZnVzaW9uUGlwZWxpbmUlMEElMEFzZCUyMCUzRCUyMERpZmZ1c2lvblBpcGVsaW5lLmZyb21fcHJldHJhaW5lZCglMjJzdGFibGUtZGlmZnVzaW9uLXYxLTUlMkZzdGFibGUtZGlmZnVzaW9uLXYxLTUlMjIlMkMlMjB0b3JjaF9kdHlwZSUzRHRvcmNoLmZsb2F0MTYp",highlighted:`<span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">import</span> torch.distributed <span class="hljs-keyword">as</span> dist
<span class="hljs-keyword">import</span> torch.multiprocessing <span class="hljs-keyword">as</span> mp

<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline

sd = DiffusionPipeline.from_pretrained(<span class="hljs-string">&quot;stable-diffusion-v1-5/stable-diffusion-v1-5&quot;</span>, torch_dtype=torch.float16)`,lang:"py",wrap:!1});var d=s(i,6);l(d,{code:"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",highlighted:`<span class="hljs-keyword">def</span> <span class="hljs-title function_">run_inference</span>(<span class="hljs-params">rank, world_size</span>):
    dist.init_process_group(<span class="hljs-string">&quot;nccl&quot;</span>, rank=rank, world_size=world_size)

    sd.to(rank)

    <span class="hljs-keyword">if</span> torch.distributed.get_rank() == <span class="hljs-number">0</span>:
        prompt = <span class="hljs-string">&quot;a dog&quot;</span>
    <span class="hljs-keyword">elif</span> torch.distributed.get_rank() == <span class="hljs-number">1</span>:
        prompt = <span class="hljs-string">&quot;a cat&quot;</span>

    image = sd(prompt).images[<span class="hljs-number">0</span>]
    image.save(<span class="hljs-string">f&quot;./<span class="hljs-subst">{<span class="hljs-string">&#x27;_&#x27;</span>.join(prompt)}</span>.png&quot;</span>)`,lang:"py",wrap:!1});var h=s(d,4);l(h,{code:"ZGVmJTIwbWFpbigpJTNBJTBBJTIwJTIwJTIwJTIwd29ybGRfc2l6ZSUyMCUzRCUyMDIlMEElMjAlMjAlMjAlMjBtcC5zcGF3bihydW5faW5mZXJlbmNlJTJDJTIwYXJncyUzRCh3b3JsZF9zaXplJTJDKSUyQyUyMG5wcm9jcyUzRHdvcmxkX3NpemUlMkMlMjBqb2luJTNEVHJ1ZSklMEElMEElMEFpZiUyMF9fbmFtZV9fJTIwJTNEJTNEJTIwJTIyX19tYWluX18lMjIlM0ElMEElMjAlMjAlMjAlMjBtYWluKCk=",highlighted:`<span class="hljs-keyword">def</span> <span class="hljs-title function_">main</span>():
    world_size = <span class="hljs-number">2</span>
    mp.spawn(run_inference, args=(world_size,), nprocs=world_size, join=<span class="hljs-literal">True</span>)


<span class="hljs-keyword">if</span> __name__ == <span class="hljs-string">&quot;__main__&quot;</span>:
    main()`,lang:"py",wrap:!1});var M=s(h,4);l(M,{code:"dG9yY2hydW4lMjBydW5fZGlzdHJpYnV0ZWQucHklMjAtLW5wcm9jX3Blcl9ub2RlJTNEMg==",highlighted:"torchrun run_distributed.py --nproc_per_node=2",lang:"bash",wrap:!1});var f=s(M,2);T(f,{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/ko/training/distributed_inference.md"}),z(2),y(J,a),k()}export{A as component};