3. hook在开源框架中的应用
3.1 keras在深度学习训练流程中,hook函数体现的淋漓尽致。一个训练过程(不包括数据准备),会轮询多次训练集,每次称为一个epoch,每个epoch又分为多个batch来训练。流程先后拆解成:开始训练
训练一个epoch前
训练一个batch前
训练一个batch后
训练一个epoch后
评估验证集
结束训练这些步骤是穿插在训练一个batch数据的过程中,这些可以理解成是钩子函数,我们可能需要在这些钩子函数中实现一些定制化的东西,比如在训练一个epoch后我们要保存下训练的模型,在结束训练时用最好的模型执行下测试集的效果等等。keras中是通过各种回调函数来实现钩子hook功能的。这里放一个callback的父类,定制时只要继承这个父类,实现你过关注的钩子就可以了。@keras_export("keras.callbacks.Callback")class Callback(object): """Abstract base class used to build new callbacks. Attributes: params: Dict. Training parameters (eg. verbosity, batch size, number of epochs...). model: Instance of `keras.models.Model`. Reference of the model being trained. The `logs` dictionary that callback methods take as argument will contain keys for quantities relevant to the current batch or epoch (see method-specific docstrings). """ def __init__(self): self.validation_data = None # pylint: disable=g-missing-from-attributes self.model = None # Whether this Callback should only run on the chief worker in a # Multi-Worker setting. # TODO(omalleyt): Make this attr public once solution is stable. self._chief_worker_only = None self._supports_tf_logs = False def set_params(self, params): self.params = params def set_model(self, model): self.model = model @doc_controls.for_subclass_implementers @generic_utils.default def on_batch_begin(self, batch, logs=None): """A backwards compatibility alias for `on_train_batch_begin`.""" @doc_controls.for_subclass_implementers @generic_utils.default def on_batch_end(self, batch, logs=None): """A backwards compatibility alias for `on_train_batch_end`.""" @doc_controls.for_subclass_implementers def on_epoch_begin(self, epoch, logs=None): """Called at the start of an epoch. Subclasses should override for any actions to run. This function should only be called during TRAIN mode. Arguments: epoch: Integer, index of epoch. logs: Dict. Currently no data is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_epoch_end(self, epoch, logs=None): """Called at the end of an epoch. Subclasses should override for any actions to run. This function should only be called during TRAIN mode. Arguments: epoch: Integer, index of epoch. logs: Dict, metric results for this training epoch, and for the validation epoch if validation is performed. Validation result keys are prefixed with `val_`. """ @doc_controls.for_subclass_implementers @generic_utils.default def on_train_batch_begin(self, batch, logs=None): """Called at the beginning of a training batch in `fit` methods. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict, contains the return value of `model.train_step`. Typically, the values of the `Model`"s metrics are returned. Example: `{"loss": 0.2, "accuracy": 0.7}`. """ # For backwards compatibility. self.on_batch_begin(batch, logs=logs) @doc_controls.for_subclass_implementers @generic_utils.default def on_train_batch_end(self, batch, logs=None): """Called at the end of a training batch in `fit` methods. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict. Aggregated metric results up until this batch. """ # For backwards compatibility. self.on_batch_end(batch, logs=logs) @doc_controls.for_subclass_implementers @generic_utils.default def on_test_batch_begin(self, batch, logs=None): """Called at the beginning of a batch in `evaluate` methods. Also called at the beginning of a validation batch in the `fit` methods, if validation data is provided. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict, contains the return value of `model.test_step`. Typically, the values of the `Model`"s metrics are returned. Example: `{"loss": 0.2, "accuracy": 0.7}`. """ @doc_controls.for_subclass_implementers @generic_utils.default def on_test_batch_end(self, batch, logs=None): """Called at the end of a batch in `evaluate` methods. Also called at the end of a validation batch in the `fit` methods, if validation data is provided. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict. Aggregated metric results up until this batch. """ @doc_controls.for_subclass_implementers @generic_utils.default def on_predict_batch_begin(self, batch, logs=None): """Called at the beginning of a batch in `predict` methods. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict, contains the return value of `model.predict_step`, it typically returns a dict with a key "outputs" containing the model"s outputs. """ @doc_controls.for_subclass_implementers @generic_utils.default def on_predict_batch_end(self, batch, logs=None): """Called at the end of a batch in `predict` methods. Subclasses should override for any actions to run. Arguments: batch: Integer, index of batch within the current epoch. logs: Dict. Aggregated metric results up until this batch. """ @doc_controls.for_subclass_implementers def on_train_begin(self, logs=None): """Called at the beginning of training. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently no data is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_train_end(self, logs=None): """Called at the end of training. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently the output of the last call to `on_epoch_end()` is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_test_begin(self, logs=None): """Called at the beginning of evaluation or validation. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently no data is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_test_end(self, logs=None): """Called at the end of evaluation or validation. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently the output of the last call to `on_test_batch_end()` is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_predict_begin(self, logs=None): """Called at the beginning of prediction. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently no data is passed to this argument for this method but that may change in the future. """ @doc_controls.for_subclass_implementers def on_predict_end(self, logs=None): """Called at the end of prediction. Subclasses should override for any actions to run. Arguments: logs: Dict. Currently no data is passed to this argument for this method but that may change in the future. """ def _implements_train_batch_hooks(self): """Determines if this Callback should be called for each train batch.""" return (not generic_utils.is_default(self.on_batch_begin) or not generic_utils.is_default(self.on_batch_end) or not generic_utils.is_default(self.on_train_batch_begin) or not generic_utils.is_default(self.on_train_batch_end))这些钩子的原始程序是在模型训练流程中的keras源码位置: tensorflow\python\keras\engine\training.py部分摘录如下(## I am hook):# Container that configures and calls `tf.keras.Callback`s. if not isinstance(callbacks, callbacks_module.CallbackList): callbacks = callbacks_module.CallbackList( callbacks, add_history=True, add_progbar=verbose != 0, model=self, verbose=verbose, epochs=epochs, steps=data_handler.inferred_steps) ## I am hook callbacks.on_train_begin() training_logs = None # Handle fault-tolerance for multi-worker. # TODO(omalleyt): Fix the ordering issues that mean this has to # happen after `callbacks.on_train_begin`. data_handler._initial_epoch = ( # pylint: disable=protected-access self._maybe_load_initial_epoch_from_ckpt(initial_epoch)) for epoch, iterator in data_handler.enumerate_epochs(): self.reset_metrics() callbacks.on_epoch_begin(epoch) with data_handler.catch_stop_iteration(): for step in data_handler.steps(): with trace.Trace( "TraceContext", graph_type="train", epoch_num=epoch, step_num=step, batch_size=batch_size): ## I am hook callbacks.on_train_batch_begin(step) tmp_logs = train_function(iterator) if data_handler.should_sync: context.async_wait() logs = tmp_logs # No error, now safe to assign to logs. end_step = step + data_handler.step_increment callbacks.on_train_batch_end(end_step, logs) epoch_logs = copy.copy(logs) # Run validation. ## I am hook callbacks.on_epoch_end(epoch, epoch_logs)
3.2 mmdetectionmmdetection是一个目标检测的开源框架,集成了许多不同的目标检测深度学习算法(pytorch版),如faster-rcnn, fpn, retianet等。里面也大量使用了hook,暴露给应用实现流程中具体部分。详见https://github.com/open-mmlab/mmdetection这里看一个训练的调用例子(摘录)(https://github.com/open-mmlab/mmdetection/blob/5d592154cca589c5113e8aadc8798bbc73630d98/mmdet/apis/train.py)def train_detector(model, dataset, cfg, distributed=False, validate=False, timestamp=None, meta=None): logger = get_root_logger(cfg.log_level) # prepare data loaders # put model on gpus # build runner optimizer = build_optimizer(model, cfg.optimizer) runner = EpochBasedRunner( model, optimizer=optimizer, work_dir=cfg.work_dir, logger=logger, meta=meta) # an ugly workaround to make .log and .log.json filenames the same runner.timestamp = timestamp # fp16 setting # register hooks runner.register_training_hooks(cfg.lr_config, optimizer_config, cfg.checkpoint_config, cfg.log_config, cfg.get("momentum_config", None)) if distributed: runner.register_hook(DistSamplerSeedHook()) # register eval hooks if validate: # Support batch_size > 1 in validation eval_cfg = cfg.get("evaluation", {}) eval_hook = DistEvalHook if distributed else EvalHook runner.register_hook(eval_hook(val_dataloader, **eval_cfg)) # user-defined hooks if cfg.get("custom_hooks", None): custom_hooks = cfg.custom_hooks assert isinstance(custom_hooks, list), \ f"custom_hooks expect list type, but got {type(custom_hooks)}" for hook_cfg in cfg.custom_hooks: assert isinstance(hook_cfg, dict), \ "Each item in custom_hooks expects dict type, but got " \ f"{type(hook_cfg)}" hook_cfg = hook_cfg.copy() priority = hook_cfg.pop("priority", "NORMAL") hook = build_from_cfg(hook_cfg, HOOKS) runner.register_hook(hook, priority=priority)
4. 总结本文介绍了hook的概念和应用,并给出了python的实现细则。希望对比有帮助。总结如下:hook函数是流程中预定义好的一个步骤,没有实现
挂载或者注册时, 流程执行就会执行这个钩子函数
回调函数和hook函数功能上是一致的
hook设计方式带来灵活性,如果流程中有一个步骤,你想让调用方来实现,你可以用hook函数