avalanche.training.PackNet

class avalanche.training.PackNet(*, model: ~avalanche.models.packnet.PackNetModule | ~avalanche.models.packnet.PackNetModel, optimizer: ~torch.optim.optimizer.Optimizer, post_prune_epochs: int, prune_proportion: float, criterion=CrossEntropyLoss(), train_mb_size: int = 1, train_epochs: int = 1, eval_mb_size: int | None = None, device: str | ~torch.device = 'cpu', plugins: ~typing.List[~avalanche.core.SupervisedPlugin] | None = None, evaluator: ~avalanche.training.plugins.evaluation.EvaluationPlugin | ~typing.Callable[[], ~avalanche.training.plugins.evaluation.EvaluationPlugin] = <function default_evaluator>, eval_every=-1, **base_kwargs)

Task-incremental fixed-network parameter isolation with PackNet.

The strategy packs multiple tasks into a single network by pruning parameters that are not important for the current task. This is done by pruning the network after each task. The network is then finetuned for a few epochs to recover the performance. This process is repeated for each task.

The supplied model must be wrapped in a PackNetModel or implement PackNetModule. These wrappers are designed to automatically upgrade most models to support PackNet.

Mallya, A., & Lazebnik, S. (2018). PackNet: Adding Multiple Tasks to a

Single Network by Iterative Pruning. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7765-7773. https://doi.org/10.1109/CVPR.2018.00810

__init__(*, model: ~avalanche.models.packnet.PackNetModule | ~avalanche.models.packnet.PackNetModel, optimizer: ~torch.optim.optimizer.Optimizer, post_prune_epochs: int, prune_proportion: float, criterion=CrossEntropyLoss(), train_mb_size: int = 1, train_epochs: int = 1, eval_mb_size: int | None = None, device: str | ~torch.device = 'cpu', plugins: ~typing.List[~avalanche.core.SupervisedPlugin] | None = None, evaluator: ~avalanche.training.plugins.evaluation.EvaluationPlugin | ~typing.Callable[[], ~avalanche.training.plugins.evaluation.EvaluationPlugin] = <function default_evaluator>, eval_every=-1, **base_kwargs)

Creates an instance of the Naive strategy.

Parameters:

• model – The model. You can use many modules wrapped in a avalanche.models.PackNetModel or your own implementation of avalanche.models.PackNetModule.

• optimizer – The optimizer to use.

• post_prune_epochs – The number of epochs to finetune the model after pruning the parameters. Must be less than the number of training epochs.

• prune_proportion – The proportion of parameters to prune each durring each task. Must be between 0 and 1.

• criterion – The loss criterion to use.

• train_mb_size – The train minibatch size. Defaults to 1.

• train_epochs – The number of training epochs. Defaults to 1.

• eval_mb_size – The eval minibatch size. Defaults to 1.

• device – The device to use. Defaults to None (cpu).

• plugins – Plugins to be added. Defaults to None.

• evaluator – (optional) instance of EvaluationPlugin for logging and metric computations.

• eval_every – the frequency of the calls to eval inside the training loop. -1 disables the evaluation. 0 means eval is called only at the end of the learning experience. Values >0 mean that eval is called every eval_every epochs and at the end of the learning experience.

• base_kwargs – any additional BaseTemplate constructor arguments.

Methods

__init__(*, model, optimizer, ...[, ...])	Creates an instance of the Naive strategy.
backward()	Run the backward pass.
check_model_and_optimizer([...])
criterion()	Loss function for supervised problems.
eval(exp_list, **kwargs)	Evaluate the current model on a series of experiences and returns the last recorded value for each metric.
eval_dataset_adaptation(**kwargs)	Initialize self.adapted_dataset.
eval_epoch(**kwargs)	Evaluation loop over the current self.dataloader.
forward()	Compute the model's output given the current mini-batch.
make_eval_dataloader([num_workers, shuffle, ...])	Initializes the eval data loader. :param num_workers: How many subprocesses to use for data loading. 0 means that the data will be loaded in the main process. (default: 0). :param pin_memory: If True, the data loader will copy Tensors into CUDA pinned memory before returning them. Defaults to True. :param kwargs: :return:.
make_optimizer([reset_optimizer_state, ...])	Optimizer initialization.
make_train_dataloader([num_workers, ...])	Data loader initialization.
model_adaptation([model])	Adapts the model to the current data.
optimizer_step()	Execute the optimizer step (weights update).
stop_training()	Signals to stop training at the next iteration.
train(experiences[, eval_streams])	Training loop.
train_dataset_adaptation(**kwargs)	Initialize self.adapted_dataset.
training_epoch(**kwargs)	Training epoch.

Attributes

is_eval	True if the strategy is in evaluation mode.
mb_task_id	Current mini-batch task labels.
mb_x	Current mini-batch input.
mb_y	Current mini-batch target.
mbatch	Current mini-batch.
mb_output	Model's output computed on the current mini-batch.
dataloader	Dataloader.
optimizer	PyTorch optimizer.
loss	Loss of the current mini-batch.
train_epochs	Number of training epochs.
train_mb_size	Training mini-batch size.
eval_mb_size	Eval mini-batch size.
retain_graph	Retain graph when calling loss.backward().
evaluator	EvaluationPlugin used for logging and metric computations.
clock	Incremental counters for strategy events.
adapted_dataset	Data used to train.
model	PyTorch model.
device	PyTorch device where the model will be allocated.
plugins	List of `SupervisedPlugin`s. .
experience	Current experience.
is_training	True if the strategy is in training mode.
current_eval_stream	Current evaluation stream.