Pruning¶
Runners¶
-
class
compressors.pruning.runners.PruneRunner(num_sessions: int = 5, input_key: Union[str, List[str]] = 'features', output_key: Union[str, List[str]] = 'logits', *runner_args, **runner_kwargs)¶ -
handle_batch(batch: Mapping[str, Any]) → None¶ Inner method to handle specified data batch. Used to make a train/valid/infer stage during Experiment run.
- Parameters
batch (Mapping[str, Any]) – dictionary with data batches from DataLoader.
-
predict_batch(batch: Mapping[str, Any], **kwargs) → Mapping[str, Any]¶ Run model inference on specified data batch.
- Parameters
batch – dictionary with data batches from DataLoader.
**kwargs – additional kwargs to pass to the model
- Returns
model output dictionary
- Return type
Mapping
- Raises
NotImplementedError – if not implemented yet
-
property
stages¶ Experiment’s stage names (array with one value).
-
Callbacks¶
-
class
compressors.pruning.callbacks.LotteryTicketCallback(initial_state_dict: Dict[str, Any])¶ -
on_stage_start(runner: catalyst.core.runner.IRunner) → None¶ Event handler.
- Parameters
runner – experiment runner
-
-
class
compressors.pruning.callbacks.PrepareForFinePruningCallback(train_loader_key: str = 'train', *logits_dataset_args, **logits_dataset_kwargs)¶ -
on_experiment_start(runner: catalyst.core.runner.IRunner) → None¶ Event handler for experiment start.
- Parameters
runner – IRunner instance.
Note
This event work only on IRunner.
-