# 콜백 [[callbacks]] 콜백은 PyTorch [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer)의 반복 학습 동작을 사용자 정의할 수 있는 객체입니다 (이 기능은 TensorFlow에서는 아직 구현되지 않았습니다). 콜백은 반복 학습의 상태를 검사하여 (진행 상황 보고, TensorBoard 또는 기타 머신 러닝 플랫폼에 로그 남기기 등) 결정(예: 조기 종료)을 내릴 수 있습니다. 콜백은 [TrainerControl](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerControl) 객체를 반환하는 것 외에는 반복 학습에서 어떤 것도 변경할 수 없는 "읽기 전용" 코드 조각입니다. 반복 학습에 변경이 필요한 사용자 정의 작업이 필요한 경우, [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer)를 서브클래스로 만들어 필요한 메소드들을 오버라이드해야 합니다 (예시는 [trainer](trainer)를 참조하세요). 기본적으로 `TrainingArguments.report_to`는 `"all"`로 설정되어 있으므로, [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer)는 다음 콜백을 사용합니다. - [DefaultFlowCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.DefaultFlowCallback)는 로그, 저장, 평가에 대한 기본 동작을 처리합니다. - [PrinterCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.PrinterCallback) 또는 [ProgressCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.ProgressCallback)는 진행 상황을 표시하고 로그를 출력합니다 ([TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)를 통해 tqdm을 비활성화하면 첫 번째 콜백이 사용되고, 그렇지 않으면 두 번째가 사용됩니다). - [TensorBoardCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.TensorBoardCallback)는 TensorBoard가 (PyTorch >= 1.4 또는 tensorboardX를 통해) 접근 가능하면 사용됩니다. - [WandbCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.WandbCallback)는 [wandb](https://www.wandb.com/)가 설치되어 있으면 사용됩니다. - [CometCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.CometCallback)는 [comet_ml](https://www.comet.com/site/)이 설치되어 있으면 사용됩니다. - [MLflowCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.MLflowCallback)는 [mlflow](https://www.mlflow.org/)가 설치되어 있으면 사용됩니다. - [NeptuneCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.NeptuneCallback)는 [neptune](https://neptune.ai/)이 설치되어 있으면 사용됩니다. - [AzureMLCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.AzureMLCallback)는 [azureml-sdk](https://pypi.org/project/azureml-sdk/)가 설치되어 있으면 사용됩니다. - [CodeCarbonCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.CodeCarbonCallback)는 [codecarbon](https://pypi.org/project/codecarbon/)이 설치되어 있으면 사용됩니다. - [ClearMLCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.ClearMLCallback)는 [clearml](https://github.com/allegroai/clearml)이 설치되어 있으면 사용됩니다. - [DagsHubCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.DagsHubCallback)는 [dagshub](https://dagshub.com/)이 설치되어 있으면 사용됩니다. - [FlyteCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.FlyteCallback)는 [flyte](https://flyte.org/)가 설치되어 있으면 사용됩니다. - [DVCLiveCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.DVCLiveCallback)는 [dvclive](https://dvc.org/doc/dvclive)가 설치되어 있으면 사용됩니다. - [SwanLabCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.integrations.SwanLabCallback)는 [swanlab](https://swanlab.cn)가 설치되어 있으면 사용됩니다. 패키지가 설치되어 있지만 해당 통합 기능을 사용하고 싶지 않다면, `TrainingArguments.report_to`를 사용하고자 하는 통합 기능 목록으로 변경할 수 있습니다 (예: `["azure_ml", "wandb"]`). 콜백을 구현하는 주요 클래스는 [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback)입니다. 이 클래스는 [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer)를 인스턴스화하는 데 사용된 [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)를 가져오고, 해당 Trainer의 내부 상태를 [TrainerState](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerState)를 통해 접근할 수 있으며, [TrainerControl](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerControl)을 통해 반복 학습에서 일부 작업을 수행할 수 있습니다. ## 사용 가능한 콜백 [[available-callbacks]][[transformers.integrations.CometCallback]] 라이브러리에서 사용 가능한 [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) 목록은 다음과 같습니다:
class transformers.integrations.CometCallbacktransformers.integrations.CometCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1198[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [Comet ML](https://www.comet.com/site/).
setuptransformers.integrations.CometCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1212[{"name": "args", "val": ""}, {"name": "state", "val": ""}, {"name": "model", "val": ""}] Setup the optional Comet integration. Environment: - **COMET_MODE** (`str`, *optional*, default to `get_or_create`): Control whether to create and log to a new Comet experiment or append to an existing experiment. It accepts the following values: * `get_or_create`: Decides automatically depending if `COMET_EXPERIMENT_KEY` is set and whether an Experiment with that key already exists or not. * `create`: Always create a new Comet Experiment. * `get`: Always try to append to an Existing Comet Experiment. Requires `COMET_EXPERIMENT_KEY` to be set. * `ONLINE`: **deprecated**, used to create an online Experiment. Use `COMET_START_ONLINE=1` instead. * `OFFLINE`: **deprecated**, used to created an offline Experiment. Use `COMET_START_ONLINE=0` instead. * `DISABLED`: **deprecated**, used to disable Comet logging. Use the `--report_to` flag to control the integrations used for logging result instead. - **COMET_PROJECT_NAME** (`str`, *optional*): Comet project name for experiments. - **COMET_LOG_ASSETS** (`str`, *optional*, defaults to `TRUE`): Whether or not to log training assets (tf event logs, checkpoints, etc), to Comet. Can be `TRUE`, or `FALSE`. For a number of configurable items in the environment, see [here](https://www.comet.com/docs/v2/guides/experiment-management/configure-sdk/#explore-comet-configuration-options).
class transformers.DefaultFlowCallbacktransformers.DefaultFlowCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L574[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that handles the default flow of the training loop for logs, evaluation and checkpoints.
class transformers.PrinterCallbacktransformers.PrinterCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L701[] A bare [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that just prints the logs.
class transformers.ProgressCallbacktransformers.ProgressCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L627[{"name": "max_str_len", "val": ": int = 100"}] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that displays the progress of training or evaluation. You can modify `max_str_len` to control how long strings are truncated when logging.
class transformers.EarlyStoppingCallbacktransformers.EarlyStoppingCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L712[{"name": "early_stopping_patience", "val": ": int = 1"}, {"name": "early_stopping_threshold", "val": ": typing.Optional[float] = 0.0"}]- **early_stopping_patience** (`int`) -- Use with `metric_for_best_model` to stop training when the specified metric worsens for `early_stopping_patience` evaluation calls. - **early_stopping_threshold(`float`,** *optional*) -- Use with TrainingArguments `metric_for_best_model` and `early_stopping_patience` to denote how much the specified metric must improve to satisfy early stopping conditions. `0 A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that handles early stopping. This callback depends on [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments) argument *load_best_model_at_end* functionality to set best_metric in [TrainerState](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerState). Note that if the [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments) argument *save_steps* differs from *eval_steps*, the early stopping will not occur until the next save step.
class transformers.integrations.TensorBoardCallbacktransformers.integrations.TensorBoardCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L669[{"name": "tb_writer", "val": " = None"}]- **tb_writer** (`SummaryWriter`, *optional*) -- The writer to use. Will instantiate one if not set.0 A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [TensorBoard](https://www.tensorflow.org/tensorboard).
class transformers.integrations.WandbCallbacktransformers.integrations.WandbCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L802[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that logs metrics, media, model checkpoints to [Weight and Biases](https://www.wandb.com/).
setuptransformers.integrations.WandbCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L831[{"name": "args", "val": ""}, {"name": "state", "val": ""}, {"name": "model", "val": ""}, {"name": "**kwargs", "val": ""}] Setup the optional Weights & Biases (*wandb*) integration. One can subclass and override this method to customize the setup if needed. Find more information [here](https://docs.wandb.ai/guides/integrations/huggingface). You can also override the following environment variables: Environment: - **WANDB_LOG_MODEL** (`str`, *optional*, defaults to `"false"`): Whether to log model and checkpoints during training. Can be `"end"`, `"checkpoint"` or `"false"`. If set to `"end"`, the model will be uploaded at the end of training. If set to `"checkpoint"`, the checkpoint will be uploaded every `args.save_steps` . If set to `"false"`, the model will not be uploaded. Use along with `load_best_model_at_end()` to upload best model. Setting `WANDB_LOG_MODEL` as `bool` will be deprecated in version 5 of 🤗 Transformers. - **WANDB_WATCH** (`str`, *optional* defaults to `"false"`): Can be `"gradients"`, `"all"`, `"parameters"`, or `"false"`. Set to `"all"` to log gradients and parameters. - **WANDB_PROJECT** (`str`, *optional*, defaults to `"huggingface"`): Set this to a custom string to store results in a different project. - **WANDB_DISABLED** (`bool`, *optional*, defaults to `False`): Whether to disable wandb entirely. Set `WANDB_DISABLED=true` to disable.
class transformers.integrations.MLflowCallbacktransformers.integrations.MLflowCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1361[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [MLflow](https://www.mlflow.org/). Can be disabled by setting environment variable `DISABLE_MLFLOW_INTEGRATION = TRUE`.
setuptransformers.integrations.MLflowCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1380[{"name": "args", "val": ""}, {"name": "state", "val": ""}, {"name": "model", "val": ""}] Setup the optional MLflow integration. Environment: - **HF_MLFLOW_LOG_ARTIFACTS** (`str`, *optional*): Whether to use MLflow `.log_artifact()` facility to log artifacts. This only makes sense if logging to a remote server, e.g. s3 or GCS. If set to `True` or *1*, will copy each saved checkpoint on each save in [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)'s `output_dir` to the local or remote artifact storage. Using it without a remote storage will just copy the files to your artifact location. - **MLFLOW_TRACKING_URI** (`str`, *optional*): Whether to store runs at a specific path or remote server. Unset by default, which skips setting the tracking URI entirely. - **MLFLOW_EXPERIMENT_NAME** (`str`, *optional*, defaults to `None`): Whether to use an MLflow experiment_name under which to launch the run. Default to `None` which will point to the `Default` experiment in MLflow. Otherwise, it is a case sensitive name of the experiment to be activated. If an experiment with this name does not exist, a new experiment with this name is created. - **MLFLOW_TAGS** (`str`, *optional*): A string dump of a dictionary of key/value pair to be added to the MLflow run as tags. Example: `os.environ['MLFLOW_TAGS']='{"release.candidate": "RC1", "release.version": "2.2.0"}'`. - **MLFLOW_NESTED_RUN** (`str`, *optional*): Whether to use MLflow nested runs. If set to `True` or *1*, will create a nested run inside the current run. - **MLFLOW_RUN_ID** (`str`, *optional*): Allow to reattach to an existing run which can be useful when resuming training from a checkpoint. When `MLFLOW_RUN_ID` environment variable is set, `start_run` attempts to resume a run with the specified run ID and other parameters are ignored. - **MLFLOW_FLATTEN_PARAMS** (`str`, *optional*, defaults to `False`): Whether to flatten the parameters dictionary before logging. - **MLFLOW_MAX_LOG_PARAMS** (`int`, *optional*): Set the maximum number of parameters to log in the run.
class transformers.integrations.AzureMLCallbacktransformers.integrations.AzureMLCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1338[{"name": "azureml_run", "val": " = None"}] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [AzureML](https://pypi.org/project/azureml-sdk/).
class transformers.integrations.CodeCarbonCallbacktransformers.integrations.CodeCarbonCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1875[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that tracks the CO2 emission of training.
class transformers.integrations.NeptuneCallbacktransformers.integrations.NeptuneCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1602[{"name": "api_token", "val": ": typing.Optional[str] = None"}, {"name": "project", "val": ": typing.Optional[str] = None"}, {"name": "name", "val": ": typing.Optional[str] = None"}, {"name": "base_namespace", "val": ": str = 'finetuning'"}, {"name": "run", "val": " = None"}, {"name": "log_parameters", "val": ": bool = True"}, {"name": "log_checkpoints", "val": ": typing.Optional[str] = None"}, {"name": "**neptune_run_kwargs", "val": ""}]- **api_token** (`str`, *optional*) -- Neptune API token obtained upon registration. You can leave this argument out if you have saved your token to the `NEPTUNE_API_TOKEN` environment variable (strongly recommended). See full setup instructions in the [docs](https://docs.neptune.ai/setup/installation). - **project** (`str`, *optional*) -- Name of an existing Neptune project, in the form "workspace-name/project-name". You can find and copy the name in Neptune from the project settings -> Properties. If None (default), the value of the `NEPTUNE_PROJECT` environment variable is used. - **name** (`str`, *optional*) -- Custom name for the run. - **base_namespace** (`str`, *optional*, defaults to "finetuning") -- In the Neptune run, the root namespace that will contain all of the metadata logged by the callback. - **log_parameters** (`bool`, *optional*, defaults to `True`) -- If True, logs all Trainer arguments and model parameters provided by the Trainer. - **log_checkpoints** (`str`, *optional*) -- If "same", uploads checkpoints whenever they are saved by the Trainer. If "last", uploads only the most recently saved checkpoint. If "best", uploads the best checkpoint (among the ones saved by the Trainer). If `None`, does not upload checkpoints. - **run** (`Run`, *optional*) -- Pass a Neptune run object if you want to continue logging to an existing run. Read more about resuming runs in the [docs](https://docs.neptune.ai/logging/to_existing_object). - ****neptune_run_kwargs** (*optional*) -- Additional keyword arguments to be passed directly to the [`neptune.init_run()`](https://docs.neptune.ai/api/neptune#init_run) function when a new run is created.0 TrainerCallback that sends the logs to [Neptune](https://app.neptune.ai). For instructions and examples, see the [Transformers integration guide](https://docs.neptune.ai/integrations/transformers) in the Neptune documentation.
class transformers.integrations.ClearMLCallbacktransformers.integrations.ClearMLCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1909[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [ClearML](https://clear.ml/). Environment: - **CLEARML_PROJECT** (`str`, *optional*, defaults to `HuggingFace Transformers`): ClearML project name. - **CLEARML_TASK** (`str`, *optional*, defaults to `Trainer`): ClearML task name. - **CLEARML_LOG_MODEL** (`bool`, *optional*, defaults to `False`): Whether to log models as artifacts during training.
class transformers.integrations.DagsHubCallbacktransformers.integrations.DagsHubCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1542[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that logs to [DagsHub](https://dagshub.com/). Extends `MLflowCallback`
setuptransformers.integrations.DagsHubCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L1556[{"name": "*args", "val": ""}, {"name": "**kwargs", "val": ""}] Setup the DagsHub's Logging integration. Environment: - **HF_DAGSHUB_LOG_ARTIFACTS** (`str`, *optional*): Whether to save the data and model artifacts for the experiment. Default to `False`.
class transformers.integrations.FlyteCallbacktransformers.integrations.FlyteCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L2160[{"name": "save_log_history", "val": ": bool = True"}, {"name": "sync_checkpoints", "val": ": bool = True"}]- **save_log_history** (`bool`, *optional*, defaults to `True`) -- When set to True, the training logs are saved as a Flyte Deck. - **sync_checkpoints** (`bool`, *optional*, defaults to `True`) -- When set to True, checkpoints are synced with Flyte and can be used to resume training in the case of an interruption.0 A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [Flyte](https://flyte.org/). NOTE: This callback only works within a Flyte task. Example: ```python # Note: This example skips over some setup steps for brevity. from flytekit import current_context, task @task def train_hf_transformer(): cp = current_context().checkpoint trainer = Trainer(..., callbacks=[FlyteCallback()]) output = trainer.train(resume_from_checkpoint=cp.restore()) ```
class transformers.integrations.DVCLiveCallbacktransformers.integrations.DVCLiveCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L2223[{"name": "live", "val": ": typing.Optional[typing.Any] = None"}, {"name": "log_model", "val": ": typing.Union[typing.Literal['all'], bool, NoneType] = None"}, {"name": "**kwargs", "val": ""}]- **live** (`dvclive.Live`, *optional*, defaults to `None`) -- Optional Live instance. If None, a new instance will be created using **kwargs. - **log_model** (Union[Literal["all"], bool], *optional*, defaults to `None`) -- Whether to use `dvclive.Live.log_artifact()` to log checkpoints created by [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer). If set to `True`, the final checkpoint is logged at the end of training. If set to `"all"`, the entire [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)'s `output_dir` is logged at each checkpoint.0 A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that sends the logs to [DVCLive](https://www.dvc.org/doc/dvclive). Use the environment variables below in `setup` to configure the integration. To customize this callback beyond those environment variables, see [here](https://dvc.org/doc/dvclive/ml-frameworks/huggingface).
setuptransformers.integrations.DVCLiveCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L2264[{"name": "args", "val": ""}, {"name": "state", "val": ""}, {"name": "model", "val": ""}] Setup the optional DVCLive integration. To customize this callback beyond the environment variables below, see [here](https://dvc.org/doc/dvclive/ml-frameworks/huggingface). Environment: - **HF_DVCLIVE_LOG_MODEL** (`str`, *optional*): Whether to use `dvclive.Live.log_artifact()` to log checkpoints created by [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer). If set to `True` or *1*, the final checkpoint is logged at the end of training. If set to `all`, the entire [TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)'s `output_dir` is logged at each checkpoint.
class transformers.integrations.SwanLabCallbacktransformers.integrations.SwanLabCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L2328[] A [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) that logs metrics, media, model checkpoints to [SwanLab](https://swanlab.cn/).
setuptransformers.integrations.SwanLabCallback.setuphttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/integrations/integration_utils.py#L2342[{"name": "args", "val": ""}, {"name": "state", "val": ""}, {"name": "model", "val": ""}, {"name": "**kwargs", "val": ""}] Setup the optional SwanLab (*swanlab*) integration. One can subclass and override this method to customize the setup if needed. Find more information [here](https://docs.swanlab.cn/guide_cloud/integration/integration-huggingface-transformers.html). You can also override the following environment variables. Find more information about environment variables [here](https://docs.swanlab.cn/en/api/environment-variable.html#environment-variables) Environment: - **SWANLAB_API_KEY** (`str`, *optional*, defaults to `None`): Cloud API Key. During login, this environment variable is checked first. If it doesn't exist, the system checks if the user is already logged in. If not, the login process is initiated. - If a string is passed to the login interface, this environment variable is ignored. - If the user is already logged in, this environment variable takes precedence over locally stored login information. - **SWANLAB_PROJECT** (`str`, *optional*, defaults to `None`): Set this to a custom string to store results in a different project. If not specified, the name of the current running directory is used. - **SWANLAB_LOG_DIR** (`str`, *optional*, defaults to `swanlog`): This environment variable specifies the storage path for log files when running in local mode. By default, logs are saved in a folder named swanlog under the working directory. - **SWANLAB_MODE** (`Literal["local", "cloud", "disabled"]`, *optional*, defaults to `cloud`): SwanLab's parsing mode, which involves callbacks registered by the operator. Currently, there are three modes: local, cloud, and disabled. Note: Case-sensitive. Find more information [here](https://docs.swanlab.cn/en/api/py-init.html#swanlab-init) - **SWANLAB_LOG_MODEL** (`str`, *optional*, defaults to `None`): SwanLab does not currently support the save mode functionality.This feature will be available in a future release - **SWANLAB_WEB_HOST** (`str`, *optional*, defaults to `None`): Web address for the SwanLab cloud environment for private version (its free) - **SWANLAB_API_HOST** (`str`, *optional*, defaults to `None`): API address for the SwanLab cloud environment for private version (its free)
## TrainerCallback [[trainercallback]][[transformers.TrainerCallback]]
class transformers.TrainerCallbacktransformers.TrainerCallbackhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L297[]- **args** ([TrainingArguments](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.TrainingArguments)) -- The training arguments used to instantiate the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer). - **state** ([TrainerState](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerState)) -- The current state of the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer). - **control** ([TrainerControl](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerControl)) -- The object that is returned to the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer) and can be used to make some decisions. - **model** ([PreTrainedModel](/docs/transformers/v4.57.0/ko/main_classes/model#transformers.PreTrainedModel) or `torch.nn.Module`) -- The model being trained. - **tokenizer** ([PreTrainedTokenizer](/docs/transformers/v4.57.0/ko/main_classes/tokenizer#transformers.PreTrainedTokenizer)) -- The tokenizer used for encoding the data. This is deprecated in favour of `processing_class`. - **processing_class** ([`PreTrainedTokenizer` or `BaseImageProcessor` or `ProcessorMixin` or `FeatureExtractionMixin`]) -- The processing class used for encoding the data. Can be a tokenizer, a processor, an image processor or a feature extractor. - **optimizer** (`torch.optim.Optimizer`) -- The optimizer used for the training steps. - **lr_scheduler** (`torch.optim.lr_scheduler.LambdaLR`) -- The scheduler used for setting the learning rate. - **train_dataloader** (`torch.utils.data.DataLoader`, *optional*) -- The current dataloader used for training. - **eval_dataloader** (`torch.utils.data.DataLoader`, *optional*) -- The current dataloader used for evaluation. - **metrics** (`dict[str, float]`) -- The metrics computed by the last evaluation phase. Those are only accessible in the event `on_evaluate`. - **logs** (`dict[str, float]`) -- The values to log. Those are only accessible in the event `on_log`.0 A class for objects that will inspect the state of the training loop at some events and take some decisions. At each of those events the following arguments are available: The `control` object is the only one that can be changed by the callback, in which case the event that changes it should return the modified version. The argument `args`, `state` and `control` are positionals for all events, all the others are grouped in `kwargs`. You can unpack the ones you need in the signature of the event using them. As an example, see the code of the simple [PrinterCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.PrinterCallback). Example: ```python class PrinterCallback(TrainerCallback): def on_log(self, args, state, control, logs=None, **kwargs): _ = logs.pop("total_flos", None) if state.is_local_process_zero: print(logs) ```
on_epoch_begintransformers.TrainerCallback.on_epoch_beginhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L368[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the beginning of an epoch.
on_epoch_endtransformers.TrainerCallback.on_epoch_endhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L374[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the end of an epoch.
on_evaluatetransformers.TrainerCallback.on_evaluatehttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L412[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called after an evaluation phase.
on_init_endtransformers.TrainerCallback.on_init_endhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L350[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the end of the initialization of the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer).
on_logtransformers.TrainerCallback.on_loghttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L430[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called after logging the last logs.
on_optimizer_steptransformers.TrainerCallback.on_optimizer_stephttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L393[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called after the optimizer step but before gradients are zeroed out. Useful for monitoring gradients.
on_pre_optimizer_steptransformers.TrainerCallback.on_pre_optimizer_stephttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L387[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called before the optimizer step but after gradient clipping. Useful for monitoring gradients.
on_predicttransformers.TrainerCallback.on_predicthttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L418[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "metrics", "val": ""}, {"name": "**kwargs", "val": ""}] Event called after a successful prediction.
on_prediction_steptransformers.TrainerCallback.on_prediction_stephttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L436[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called after a prediction step.
on_savetransformers.TrainerCallback.on_savehttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L424[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called after a checkpoint save.
on_step_begintransformers.TrainerCallback.on_step_beginhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L380[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the beginning of a training step. If using gradient accumulation, one training step might take several inputs.
on_step_endtransformers.TrainerCallback.on_step_endhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L405[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the end of a training step. If using gradient accumulation, one training step might take several inputs.
on_substep_endtransformers.TrainerCallback.on_substep_endhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L399[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the end of an substep during gradient accumulation.
on_train_begintransformers.TrainerCallback.on_train_beginhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L356[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the beginning of training.
on_train_endtransformers.TrainerCallback.on_train_endhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L362[{"name": "args", "val": ": TrainingArguments"}, {"name": "state", "val": ": TrainerState"}, {"name": "control", "val": ": TrainerControl"}, {"name": "**kwargs", "val": ""}] Event called at the end of training.
여기 PyTorch [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer)와 함께 사용자 정의 콜백을 등록하는 예시가 있습니다: ```python class MyCallback(TrainerCallback): "A callback that prints a message at the beginning of training" def on_train_begin(self, args, state, control, **kwargs): print("Starting training") trainer = Trainer( model, args, train_dataset=train_dataset, eval_dataset=eval_dataset, callbacks=[MyCallback], # 우리는 콜백 클래스를 이 방식으로 전달하거나 그것의 인스턴스(MyCallback())를 전달할 수 있습니다 ) ``` 또 다른 콜백을 등록하는 방법은 `trainer.add_callback()`을 호출하는 것입니다: ```python trainer = Trainer(...) trainer.add_callback(MyCallback) # 다른 방법으로는 콜백 클래스의 인스턴스를 전달할 수 있습니다 trainer.add_callback(MyCallback()) ``` ## TrainerState [[trainerstate]][[transformers.TrainerState]]
class transformers.TrainerStatetransformers.TrainerStatehttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L36[{"name": "epoch", "val": ": typing.Optional[float] = None"}, {"name": "global_step", "val": ": int = 0"}, {"name": "max_steps", "val": ": int = 0"}, {"name": "logging_steps", "val": ": int = 500"}, {"name": "eval_steps", "val": ": int = 500"}, {"name": "save_steps", "val": ": int = 500"}, {"name": "train_batch_size", "val": ": typing.Optional[int] = None"}, {"name": "num_train_epochs", "val": ": int = 0"}, {"name": "num_input_tokens_seen", "val": ": int = 0"}, {"name": "total_flos", "val": ": float = 0"}, {"name": "log_history", "val": ": list = None"}, {"name": "best_metric", "val": ": typing.Optional[float] = None"}, {"name": "best_global_step", "val": ": typing.Optional[int] = None"}, {"name": "best_model_checkpoint", "val": ": typing.Optional[str] = None"}, {"name": "is_local_process_zero", "val": ": bool = True"}, {"name": "is_world_process_zero", "val": ": bool = True"}, {"name": "is_hyper_param_search", "val": ": bool = False"}, {"name": "trial_name", "val": ": typing.Optional[str] = None"}, {"name": "trial_params", "val": ": typing.Optional[dict[str, typing.Union[str, float, int, bool]]] = None"}, {"name": "stateful_callbacks", "val": ": typing.Optional[list['TrainerCallback']] = None"}]- **epoch** (`float`, *optional*) -- Only set during training, will represent the epoch the training is at (the decimal part being the percentage of the current epoch completed). - **global_step** (`int`, *optional*, defaults to 0) -- During training, represents the number of update steps completed. - **max_steps** (`int`, *optional*, defaults to 0) -- The number of update steps to do during the current training. - **logging_steps** (`int`, *optional*, defaults to 500) -- Log every X updates steps - **eval_steps** (`int`, *optional*) -- Run an evaluation every X steps. - **save_steps** (`int`, *optional*, defaults to 500) -- Save checkpoint every X updates steps. - **train_batch_size** (`int`, *optional*) -- The batch size for the training dataloader. Only needed when `auto_find_batch_size` has been used. - **num_input_tokens_seen** (`int`, *optional*, defaults to 0) -- When tracking the inputs tokens, the number of tokens seen during training (number of input tokens, not the number of prediction tokens). - **total_flos** (`float`, *optional*, defaults to 0) -- The total number of floating operations done by the model since the beginning of training (stored as floats to avoid overflow). - **log_history** (`list[dict[str, float]]`, *optional*) -- The list of logs done since the beginning of training. - **best_metric** (`float`, *optional*) -- When tracking the best model, the value of the best metric encountered so far. - **best_global_step** (`int`, *optional*) -- When tracking the best model, the step at which the best metric was encountered. Used for setting `best_model_checkpoint`. - **best_model_checkpoint** (`str`, *optional*) -- When tracking the best model, the value of the name of the checkpoint for the best model encountered so far. - **is_local_process_zero** (`bool`, *optional*, defaults to `True`) -- Whether or not this process is the local (e.g., on one machine if training in a distributed fashion on several machines) main process. - **is_world_process_zero** (`bool`, *optional*, defaults to `True`) -- Whether or not this process is the global main process (when training in a distributed fashion on several machines, this is only going to be `True` for one process). - **is_hyper_param_search** (`bool`, *optional*, defaults to `False`) -- Whether we are in the process of a hyper parameter search using Trainer.hyperparameter_search. This will impact the way data will be logged in TensorBoard. - **stateful_callbacks** (`list[StatefulTrainerCallback]`, *optional*) -- Callbacks attached to the `Trainer` that should have their states be saved or restored. Relevant callbacks should implement a `state` and `from_state` function.0 A class containing the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer) inner state that will be saved along the model and optimizer when checkpointing and passed to the [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback). In all this class, one step is to be understood as one update step. When using gradient accumulation, one update step may require several forward and backward passes: if you use `gradient_accumulation_steps=n`, then one update step requires going through *n* batches.
compute_stepstransformers.TrainerState.compute_stepshttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L157[{"name": "args", "val": ""}, {"name": "max_steps", "val": ""}] Calculates and stores the absolute value for logging, eval, and save steps based on if it was a proportion or not.
init_training_referencestransformers.TrainerState.init_training_referenceshttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L170[{"name": "trainer", "val": ""}, {"name": "max_steps", "val": ""}, {"name": "num_train_epochs", "val": ""}, {"name": "trial", "val": ""}] Stores the initial training references needed in `self`
load_from_jsontransformers.TrainerState.load_from_jsonhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L150[{"name": "json_path", "val": ": str"}] Create an instance from the content of `json_path`.
save_to_jsontransformers.TrainerState.save_to_jsonhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L144[{"name": "json_path", "val": ": str"}] Save the content of this instance in JSON format inside `json_path`.
## TrainerControl [[trainercontrol]][[transformers.TrainerControl]]
class transformers.TrainerControltransformers.TrainerControlhttps://github.com/huggingface/transformers/blob/v4.57.0/src/transformers/trainer_callback.py#L236[{"name": "should_training_stop", "val": ": bool = False"}, {"name": "should_epoch_stop", "val": ": bool = False"}, {"name": "should_save", "val": ": bool = False"}, {"name": "should_evaluate", "val": ": bool = False"}, {"name": "should_log", "val": ": bool = False"}]- **should_training_stop** (`bool`, *optional*, defaults to `False`) -- Whether or not the training should be interrupted. If `True`, this variable will not be set back to `False`. The training will just stop. - **should_epoch_stop** (`bool`, *optional*, defaults to `False`) -- Whether or not the current epoch should be interrupted. If `True`, this variable will be set back to `False` at the beginning of the next epoch. - **should_save** (`bool`, *optional*, defaults to `False`) -- Whether or not the model should be saved at this step. If `True`, this variable will be set back to `False` at the beginning of the next step. - **should_evaluate** (`bool`, *optional*, defaults to `False`) -- Whether or not the model should be evaluated at this step. If `True`, this variable will be set back to `False` at the beginning of the next step. - **should_log** (`bool`, *optional*, defaults to `False`) -- Whether or not the logs should be reported at this step. If `True`, this variable will be set back to `False` at the beginning of the next step.0 A class that handles the [Trainer](/docs/transformers/v4.57.0/ko/main_classes/trainer#transformers.Trainer) control flow. This class is used by the [TrainerCallback](/docs/transformers/v4.57.0/ko/main_classes/callback#transformers.TrainerCallback) to activate some switches in the training loop.