neural_compressor.compression.callbacks

This is a module for Component class.

The Component class will be inherited by the class ‘QuantizationAwareTrainingCallbacks’, ‘PruningCallbacks’ and ‘DistillationCallbacks’.

Classes

BaseCallbacks	This is base class of Neural Compressor Callbacks.
QuantizationAwareTrainingCallbacks	This is the class for callbacks of quantization aware training.
PruningCallbacks	This is the class for callbacks of pruning object.
DistillationCallbacks	Distillation class derived from Component class.

Module Contents

class neural_compressor.compression.callbacks.BaseCallbacks(conf=None, model=None)

This is base class of Neural Compressor Callbacks.

This class will be inherited by the class ‘QuantizationAwareTrainingCallbacks’, ‘PruningCallbacks’ and ‘DistillationCallbacks’. This design is mainly for pruning/distillation/quantization-aware training. In this class will apply all hooks for ‘Quantization’, ‘Pruning’ and ‘Distillation’.

class neural_compressor.compression.callbacks.QuantizationAwareTrainingCallbacks(conf=None, model=None, adaptor=None)

This is the class for callbacks of quantization aware training.

This design is mainly for Quantization-Aware Training. In this class will apply all hooks for Quantization-Aware Training.

class neural_compressor.compression.callbacks.PruningCallbacks(conf=None, model=None)

This is the class for callbacks of pruning object.

In this class will apply all hooks for Pruning.

class neural_compressor.compression.callbacks.DistillationCallbacks(conf=None, model=None)

Distillation class derived from Component class.

Distillation class abstracted the pipeline of knowledge distillation, transfer the knowledge of the teacher model to the student model.

Parameters:

• conf – Distillation_Conf containing teacher model, distillation criterion etc.

• model – Student model. It should be neural compressor model.

_epoch_ran

A integer indicating how much epochs ran.

eval_frequency

The frequency for doing evaluation of the student model in terms of epoch.

best_score

The best metric of the student model in the training.

best_model

The best student model found in the training.