aihwkit.simulator.parameters.pre_post module

Pre-post processing related parameters for resistive processing units.

class aihwkit.simulator.parameters.pre_post.InputRangeParameter(enable=<factory>, learn_input_range=True, init_value=3.0, init_from_data=100, init_std_alpha=3.0, decay=0.001, input_min_percentage=0.95, manage_output_clipping=False, output_min_percentage=0.95, gradient_scale=1.0, gradient_relative=True, calibration_info=None)[source]

Bases: _PrintableMixin

Parameter related to input range learning

Parameters:

  • enable (bool) –

  • learn_input_range (bool) –

  • init_value (float) –

  • init_from_data (int) –

  • init_std_alpha (float) –

  • decay (float) –

  • input_min_percentage (float) –

  • manage_output_clipping (bool) –

  • output_min_percentage (float) –

  • gradient_scale (float) –

  • gradient_relative (bool) –

  • calibration_info (str | None) –

calibration_info: str | None = None

Information field for potential post-training calibrations.

decay: float = 0.001

Decay rate for input range learning.

enable: bool

Whether to enable to learn the input range. Note that if enable is False then no clip is applied.

Note

The input bound (forward.inp_bound) is assumed to be 1 if enabled as the input range already scales the input into to the range \((-1, 1)\) by dividing the input to the type by itself and multiplying the output accordingly.

Typically, noise and bound management should be set to NONE for the input range learning as it replaces the dynamic managements with a static but learned input bound. However, in some exceptional experimental cases one might want to enable the management techniques on top of the input range learning, so that no error is raised if they are not set to NONE.

gradient_relative: bool = True

Whether to make the gradient of the input range learning relative to the current range value.

gradient_scale: float = 1.0

Scale of the gradient magnitude (learning rate) for the input range learning.

init_from_data: int = 100

Number of batches to use for initialization from data. Set 0 to turn off.

init_std_alpha: float = 3.0

Standard deviation multiplier for initialization from data.

init_value: float = 3.0

Initial setting of the input range in case of input range learning.

input_min_percentage: float = 0.95

Decay is only applied if percentage of non-clipped values is above this value.

Note

The added gradient is (in case of non-clipped input percentage percentage > input_min_percentage):

grad += decay * input_range
learn_input_range: bool = True

Whether to learn the input range when enabled.

Note

If not learned, the input range should in general be set with some calibration method before training the DNN.

manage_output_clipping: bool = False

Whether to increase the input range when output clipping occurs.

Caution

The output bound is taken from the forward.out_bound value, which has to exist. Noise and bound management have to be set to NONE if this feature is enabled otherwise a ConfigError is raised.

output_min_percentage: float = 0.95

Increase of the input range is only applied if percentage of non-clipped output values is below this value.

Note

The gradient subtracted from the input range is (in case of output_percentage < output_min_percentage):

grad -= (1.0 - output_percentage) * input_range
supports_manage_output_clipping(rpu_config)[source]

Checks whether rpu_config supported manage_output_clipping.

Parameters:

rpu_config (Any) – RPUConfig to check

Returns:

True if supported otherwise False

Return type:

bool

class aihwkit.simulator.parameters.pre_post.PeripheryQuantizationParameter(n_bits=0, symmetric=True, learn_quant_params=False, init_learning_after=100)[source]

Bases: _PrintableMixin

Defines parameters used for the quantization of the periphery operations in the QuantizedTorchInferenceTile. These parameters will be applied both to the affine scales (e.g., per-column weight scaling) but also on the bias that is added, whether that happens at the tile level or at the array level (see QuantizedTileModuleArray).

Parameters:

  • n_bits (int) –

  • symmetric (bool) –

  • learn_quant_params (bool) –

  • init_learning_after (int) –

init_learning_after: int = 100

If learn_quant_params is True, the quantization parameters will be estimated (minmax) up to init_learning_after batches before switching to being learnt. Has to be atleast 1 to initialize the params, by default 100.

learn_quant_params: bool = False

If True, the quantization scale (and offset if asymmetrical) will be learned during training (or was learned, if loading a checkpoint). If False, the scales will be estimated during training or used as trained. By default False.

n_bits: int = 0

The number of bits for the quantization of the periphery operations. If <= 0 is selected, no quantization will be applied. By default 0 (no quantization).

symmetric: bool = True

If True, the quantization in the periphery will be symmetrical. If False, asymmetric quantization will be used. This option is valid only if n_bits > 0.

class aihwkit.simulator.parameters.pre_post.PrePostProcessingParameter(input_range=<factory>)[source]

Bases: _PrintableMixin

Parameter related to digital input and output processing, such as input clip learning.

Parameters:

input_range (InputRangeParameter) –

input_range: InputRangeParameter
class aihwkit.simulator.parameters.pre_post.PrePostProcessingParameterQuant(input_range=<factory>, periph_quant=<factory>)[source]

Bases: PrePostProcessingParameter

Parameter related to digital input and output processing, such as input clip learning and periphery quantization.

Parameters:
periph_quant: PeripheryQuantizationParameter
class aihwkit.simulator.parameters.pre_post.PrePostProcessingRPU(tile_class, runtime=<factory>, pre_post=<factory>)[source]

Bases: RPUConfigBase, _PrintableMixin

Defines the pre-post parameters and utility factories

Parameters:
pre_post: PrePostProcessingParameter

Parameter related digital pre and post processing.