Files
transformers/docs/source/ko/accelerate.md
陈赣 06f1fd69a6
Some checks failed
Self-hosted runner (nightly-past-ci-caller) / Get number (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.11 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.10 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.9 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.8 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.7 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.6 (push) Has been cancelled
Self-hosted runner (nightly-past-ci-caller) / TensorFlow 2.5 (push) Has been cancelled
Self-hosted runner (benchmark) / Benchmark (aws-g5-4xlarge-cache) (push) Has been cancelled
Build documentation / build (push) Has been cancelled
Build documentation / build_other_lang (push) Has been cancelled
CodeQL Security Analysis / CodeQL Analysis (push) Has been cancelled
New model PR merged notification / Notify new model (push) Has been cancelled
PR CI / pr-ci (push) Has been cancelled
Slow tests on important models (on Push - A10) / Get all modified files (push) Has been cancelled
Secret Leaks / trufflehog (push) Has been cancelled
Update Transformers metadata / build_and_package (push) Has been cancelled
Slow tests on important models (on Push - A10) / Model CI (push) Has been cancelled
Check Tiny Models / Check tiny models (push) Has been cancelled
Self-hosted runner (Intel Gaudi3 scheduled CI caller) / Model CI (push) Has been cancelled
Self-hosted runner (Intel Gaudi3 scheduled CI caller) / Pipeline CI (push) Has been cancelled
Self-hosted runner (Intel Gaudi3 scheduled CI caller) / Example CI (push) Has been cancelled
Self-hosted runner (Intel Gaudi3 scheduled CI caller) / DeepSpeed CI (push) Has been cancelled
Self-hosted runner (Intel Gaudi3 scheduled CI caller) / Trainer/FSDP CI (push) Has been cancelled
Nvidia CI - Flash Attn / Setup (push) Has been cancelled
Nvidia CI - Flash Attn / Model CI (push) Has been cancelled
Nvidia CI / Setup (push) Has been cancelled
Nvidia CI / Model CI (push) Has been cancelled
Nvidia CI / Torch pipeline CI (push) Has been cancelled
Nvidia CI / Example CI (push) Has been cancelled
Nvidia CI / Trainer/FSDP CI (push) Has been cancelled
Nvidia CI / DeepSpeed CI (push) Has been cancelled
Nvidia CI / Quantization CI (push) Has been cancelled
Nvidia CI / Kernels CI (push) Has been cancelled
Doctests / Setup (push) Has been cancelled
Doctests / Call doctest jobs (push) Has been cancelled
Doctests / Send results to webhook (push) Has been cancelled
Extras Smoke Test / Get supported Python versions (push) Has been cancelled
Extras Smoke Test / Test extras on Python ${{ matrix.python-version }} (push) Has been cancelled
Extras Smoke Test / Check Slack token availability (push) Has been cancelled
Extras Smoke Test / Notify failures to Slack (push) Has been cancelled
Self-hosted runner (AMD scheduled CI caller) / Trigger Scheduled AMD CI (push) Has been cancelled
Stale Bot / Close Stale Issues (push) Has been cancelled
first commit
2026-06-05 16:53:03 +08:00

5.2 KiB

🤗 Accelerate를 활용한 분산 학습distributed-training-with-accelerate

모델이 커지면서 병렬 처리는 제한된 하드웨어에서 더 큰 모델을 훈련하고 훈련 속도를 몇 배로 가속화하기 위한 전략으로 등장했습니다. Hugging Face에서는 사용자가 하나의 머신에 여러 개의 GPU를 사용하든 여러 머신에 여러 개의 GPU를 사용하든 모든 유형의 분산 설정에서 🤗 Transformers 모델을 쉽게 훈련할 수 있도록 돕기 위해 🤗 Accelerate 라이브러리를 만들었습니다. 이 튜토리얼에서는 분산 환경에서 훈련할 수 있도록 기본 PyTorch 훈련 루프를 커스터마이즈하는 방법을 알아봅시다.

설정setup

🤗 Accelerate 설치 시작하기:

pip install accelerate

그 다음, [~accelerate.Accelerator] 객체를 불러오고 생성합니다. [~accelerate.Accelerator]는 자동으로 분산 설정 유형을 감지하고 훈련에 필요한 모든 구성 요소를 초기화합니다. 장치에 모델을 명시적으로 배치할 필요는 없습니다.

>>> from accelerate import Accelerator

>>> accelerator = Accelerator()

가속화를 위한 준비prepare-to-accelerate

다음 단계는 관련된 모든 훈련 객체를 [~accelerate.Accelerator.prepare] 메소드에 전달하는 것입니다. 여기에는 훈련 및 평가 데이터로더, 모델 및 옵티마이저가 포함됩니다:

>>> train_dataloader, eval_dataloader, model, optimizer = accelerator.prepare(
...     train_dataloader, eval_dataloader, model, optimizer
... )

백워드(Backward)backward

마지막으로 훈련 루프의 일반적인 loss.backward()🤗 Accelerate의 [~accelerate.Accelerator.backward] 메소드로 대체하기만 하면 됩니다:

>>> for epoch in range(num_epochs):
...     for batch in train_dataloader:
...         outputs = model(**batch)
...         loss = outputs.loss
...         accelerator.backward(loss)

...         optimizer.step()
...         lr_scheduler.step()
...         optimizer.zero_grad()
...         progress_bar.update(1)

다음 코드에서 볼 수 있듯이, 훈련 루프에 코드 네 줄만 추가하면 분산 학습을 활성화할 수 있습니다!

+ from accelerate import Accelerator
  from transformers import AdamW, AutoModelForSequenceClassification, get_scheduler

+ accelerator = Accelerator()

  model = AutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=2)
  optimizer = AdamW(model.parameters(), lr=3e-5)

- device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
- model.to(device)

+ train_dataloader, eval_dataloader, model, optimizer = accelerator.prepare(
+     train_dataloader, eval_dataloader, model, optimizer
+ )

  num_epochs = 3
  num_training_steps = num_epochs * len(train_dataloader)
  lr_scheduler = get_scheduler(
      "linear",
      optimizer=optimizer,
      num_warmup_steps=0,
      num_training_steps=num_training_steps
  )

  progress_bar = tqdm(range(num_training_steps))

  model.train()
  for epoch in range(num_epochs):
      for batch in train_dataloader:
-         batch = {k: v.to(device) for k, v in batch.items()}
          outputs = model(**batch)
          loss = outputs.loss
-         loss.backward()
+         accelerator.backward(loss)

          optimizer.step()
          lr_scheduler.step()
          optimizer.zero_grad()
          progress_bar.update(1)

학습train

관련 코드를 추가한 후에는 스크립트나 Colaboratory와 같은 노트북에서 훈련을 시작하세요.

스크립트로 학습하기train-with-a-script

스크립트에서 훈련을 실행하는 경우, 다음 명령을 실행하여 구성 파일을 생성하고 저장합니다:

accelerate config

Then launch your training with:

accelerate launch train.py

노트북으로 학습하기train-with-a-notebook

Collaboratory의 TPU를 사용하려는 경우, 노트북에서도 🤗 Accelerate를 실행할 수 있습니다. 훈련을 담당하는 모든 코드를 함수로 감싸서 [~accelerate.notebook_launcher]에 전달하세요:

>>> from accelerate import notebook_launcher

>>> notebook_launcher(training_function)

🤗 Accelerate 및 다양한 기능에 대한 자세한 내용은 documentation를 참조하세요.