Files
transformers/docs/source/ko/hpo_train.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

5.3 KiB

Trainer API를 사용한 하이퍼파라미터 탐색 hyperparameter-search-using-trainer-api

🤗 Transformers에서는 🤗 Transformers 모델을 학습시키는데 최적화된 [Trainer] 클래스를 제공하기 때문에, 사용자는 직접 훈련 루프를 작성할 필요 없이 더욱 간편하게 학습을 시킬 수 있습니다. 또한, [Trainer]는 하이퍼파라미터 탐색을 위한 API를 제공합니다. 이 문서에서 이 API를 활용하는 방법을 예시와 함께 보여드리겠습니다.

하이퍼파라미터 탐색 백엔드 hyperparameter-search-backend

[Trainer]는 현재 아래 4가지 하이퍼파라미터 탐색 백엔드를 지원합니다: optuna와, raytune, wandb 입니다.

하이퍼파라미터 탐색 백엔드로 사용하기 전에 아래의 명령어를 사용하여 라이브러리들을 설치하세요.

pip install optuna/wandb/ray[tune]

예제에서 하이퍼파라미터 탐색을 활성화하는 방법 how-to-enable-hyperparameter-search-in-example

하이퍼파라미터 탐색 공간을 정의하세요. 하이퍼파라미터 탐색 백엔드마다 서로 다른 형식이 필요합니다.

optuna의 경우, 해당 object_parameter 문서를 참조하여 아래와 같이 작성하세요:

>>> def optuna_hp_space(trial):
...     return {
...         "learning_rate": trial.suggest_float("learning_rate", 1e-6, 1e-4, log=True),
...         "per_device_train_batch_size": trial.suggest_categorical("per_device_train_batch_size", [16, 32, 64, 128]),
...     }

raytune의 경우, 해당 object_parameter 문서를 참조하여 아래와 같이 작성하세요:

>>> def ray_hp_space(trial):
...     return {
...         "learning_rate": tune.loguniform(1e-6, 1e-4),
...         "per_device_train_batch_size": tune.choice([16, 32, 64, 128]),
...     }

wandb의 경우, 해당 object_parameter 문서를 참조하여 아래와 같이 작성하세요:

>>> def wandb_hp_space(trial):
...     return {
...         "method": "random",
...         "metric": {"name": "objective", "goal": "minimize"},
...         "parameters": {
...             "learning_rate": {"distribution": "uniform", "min": 1e-6, "max": 1e-4},
...             "per_device_train_batch_size": {"values": [16, 32, 64, 128]},
...         },
...     }

model_init 함수를 정의하고 이를 [Trainer]에 전달하세요. 아래는 그 예시입니다.

>>> def model_init(trial):
...     return AutoModelForSequenceClassification.from_pretrained(
...         model_args.model_name_or_path,
...         from_tf=bool(".ckpt" in model_args.model_name_or_path),
...         config=config,
...         cache_dir=model_args.cache_dir,
...         revision=model_args.model_revision,
...     )

아래와 같이 model_init 함수, 훈련 인수, 훈련 및 테스트 데이터셋, 그리고 평가 함수를 사용하여 [Trainer]를 생성하세요:

>>> trainer = Trainer(
...     model=None,
...     args=training_args,
...     train_dataset=small_train_dataset,
...     eval_dataset=small_eval_dataset,
...     compute_metrics=compute_metrics,
...     processing_class=tokenizer,
...     model_init=model_init,
...     data_collator=data_collator,
... )

하이퍼파라미터 탐색을 호출하고, 최적의 시험 매개변수를 가져오세요. 백엔드는 "optuna"/"wandb"/"ray" 중에서 선택할 수 있습니다. 방향은 "minimize" 또는 "maximize" 중 선택하며, 목표를 최소화할 것인지 최대화할 것인지를 결정합니다.

자신만의 compute_objective 함수를 정의할 수 있습니다. 만약 이 함수를 정의하지 않으면, 기본 compute_objective가 호출되고, f1과 같은 평가 지표의 합이 목푯값으로 반환됩니다.

>>> best_trial = trainer.hyperparameter_search(
...     direction="maximize",
...     backend="optuna",
...     hp_space=optuna_hp_space,
...     n_trials=20,
...     compute_objective=compute_objective,
... )

DDP 미세 조정을 위한 하이퍼파라미터 탐색 hyperparameter-search-for-ddp-finetune

현재, DDP(Distributed Data Parallelism; 분산 데이터 병렬처리)를 위한 하이퍼파라미터 탐색은 optuna 가능합니다. 최상위 프로세스가 하이퍼파라미터 탐색 과정을 시작하고 그 결과를 다른 프로세스에 전달합니다.