aihwkit.simulator.configs.utils module¶
Utility parameters for resistive processing units.
-
class
aihwkit.simulator.configs.utils.
BoundManagementType
(value)[source]¶ Bases:
enum.Enum
Bound management type.
In the case
Iterative
the MAC is iteratively recomputed with inputs iteratively halved, when the output bound was hit.Caution
Bound management is only available for the forward pass. It will be ignored when used for the backward pass.
-
ITERATIVE
= 'Iterative'¶ Iteratively recomputes input scale set to \(\alpha\leftarrow\alpha/2\).
It iteratively recomputes the bounds up to limit of passes (given by
max_bm_factor
ormax_bm_res
).
-
ITERATIVE_WORST_CASE
= 'IterativeWorstCase'¶ Worst case bound management.
Uses
AbsMax
noise management for the first pass and only when output bound is hit, theAbsMaxNPSum
for the second. Thus, at most 2 passes are computed.
-
NONE
= 'None'¶ No bound management.
-
SHIFT
= 'Shift'¶ Shift bound management.
Shifts the output by adding the difference
output_bound - max_output
to the analog output value. This is only useful to increase the dynamic range before the softmax, where the max can be safely.Note
Shifting needs hardware implementations.
-
-
class
aihwkit.simulator.configs.utils.
DriftParameter
(nu=0.0, t_0=1.0, reset_tol=1e-07, nu_dtod=0.0, nu_std=0.0, wg_ratio=1.0, g_offset=0.0, w_offset=0.0, nu_k=0.0, log_g0=0.0, w_noise_std=0.0)[source]¶ Bases:
aihwkit.simulator.configs.utils.SimpleDriftParameter
Parameter for a power law drift.
The drift is based on the model described by Oh et al (2019).
It computes: .. math:
w_{ij}*\left(\frac{t + \Delta t}{t_0}\right)^(-\nu^\text{actual}_{ij})
where the drift coefficient is drawn once at the beginning and might depend on device. It also can depend on the actual weight value.
The actual drift coefficient is computed as: .. math:
\nu_{ij}^\text{actual} = \nu_{ij} - \nu_k \log \frac{(w_{ij} - w_\text{off}) / r_\text{wg} + g_\text{off}}{G_0} + \nu\sigma_\nu\xi
here \(w_{ij}\) is the actual weight and nu_{ij} fixed for each device given by the mean \(\nu\) and the device-to-device variation: \(\nu_{ij} = \nu + \nu_dtod\nu\xi\) and are only drawn once at the beginning (tile instantiation). xi is Gaussian noise.
Note
If the weight has changed from the last drift call (determined by the
reset_tol
parameter), for instance due to update, decay or noise, then the drift time \(t\) will be reset and start from new, however, the drift coefficients \(\nu_{ij}\) are not changed. On the other hand, if the weights has not changed since last call, \(t\) will accumulate the time.Caution
Note that the drift coefficient does not depend on the initially programmed weight value at \(t=0\) in the current implementation (ie G0 is a constant for all devices), but instead on the actual weight. In some materials (e.g. phase changed materials), that might be not accurate.
-
g_offset
: float = 0.0¶ g_min
to convert to physical units.
-
log_g0
: float = 0.0¶ Log g0.
-
nu_dtod
: float = 0.0¶ Device-to-device variation of the \(\nu\) values.
-
nu_k
: float = 0.0¶ nu with \(W\).
That is \(\nu(R) = nu_0 - k \log(G/G_0)\). See Oh et al. for details.
- Type
Variation of math
-
nu_std
: float = 0.0¶ Cycle-to-cycle variation of \(\nu\).
A more realistic way to add noise of the drift might be using
w_noise_std
.
-
w_noise_std
: float = 0.0¶ Additional weight noise (Gaussian diffusion) added to the weights after the drift is applied.
-
w_offset
: float = 0.0¶ w(g_min)
, i.e. to what valueg_min
is mapped to in w-space.
-
wg_ratio
: float = 1.0¶ (w_max-w_min)/(g_max-g_min)
to convert to physical units.
-
-
class
aihwkit.simulator.configs.utils.
IOParameters
(bm_test_negative_bound=True, bound_management=<BoundManagementType.ITERATIVE: 'Iterative'>, inp_bound=1.0, inp_noise=0.0, inp_res=0.007936507936507936, inp_sto_round=False, is_perfect=False, max_bm_factor=1000, max_bm_res=0.25, nm_thres=0.0, noise_management=<NoiseManagementType.ABS_MAX: 'AbsMax'>, out_bound=12.0, out_noise=0.06, out_res=0.00196078431372549, out_scale=1.0, out_sto_round=False, w_noise=0.0, w_noise_type=<WeightNoiseType.NONE: 'None'>, ir_drop=0.0, ir_drop_g_ratio=571428.5714285715)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter that modify the IO behavior.
-
bm_test_negative_bound
: bool = True¶
-
bound_management
: aihwkit.simulator.configs.utils.BoundManagementType = 'Iterative'¶ Type of bound management, see
BoundManagementType
.Caution
Bound management is only available for the forward pass. It will be ignored when used for the backward pass.
-
inp_bound
: float = 1.0¶ Input bound and ranges for the digital-to-analog converter (DAC).
-
inp_noise
: float = 0.0¶ Std deviation of Gaussian input noise (\(\sigma_\text{inp}\)).
i.e. noisiness of the analog input (at the stage after DAC and before the multiplication).
-
inp_res
: float = 0.007936507936507936¶ Number of discretization steps for DAC (\(\le0\) means infinite steps) or resolution (1/steps).
-
inp_sto_round
: bool = False¶ Whether to enable stochastic rounding of DAC.
-
ir_drop
: float = 0.0¶ Scale of IR drop along the inputs (rows of the weight matrix).
The IR-drop is calculated assuming that the first input is farthest away from the output channel. The expected drop is approximating the steady-state voltage distributions and depends on the input current.
-
ir_drop_g_ratio
: float = 571428.5714285715¶ Physical ratio of wire conductance from one cell to the next to physical max conductance of a device.
Default is compute with 5mS maximal conductance set state and 0.35 Ohm wire resistance.
-
is_perfect
: bool = False¶ Short-cut to compute a perfect forward pass.
If
True
, it assumes an ideal forward pass (e.g. no bound, ADC etc…). Will disregard all other settings in this case.
-
max_bm_factor
: int = 1000¶ Maximal bound management factor.
If this factor is reached then the iterative process is stopped.
-
max_bm_res
: float = 0.25¶ Limit the maximal number of iterations of the bound management.
Another way to limit the maximal number of iterations of the bound management. The max effective resolution number of the inputs, e.g. use \(1/4\) for 2 bits.
-
nm_thres
: float = 0.0¶ Constant noise management value for
type
Constant
.In other cases, this is a upper threshold \(\theta\) above which the noise management factor is saturated. E.g. for AbsMax:
\begin{equation*} \alpha=\begin{cases}\max_i|x_i|, & \text{if} \max_i|x_i|<\theta \\ \theta, & \text{otherwise}\end{cases} \end{equation*}Caution
If
nm_thres
is set (and type is notConstant
), the noise management will clip some large input values, in favor of having a better SNR for smaller input values.
-
noise_management
: aihwkit.simulator.configs.utils.NoiseManagementType = 'AbsMax'¶ Type of noise management, see
NoiseManagementType
.
-
out_bound
: float = 12.0¶ Output bound and ranges for analog-to-digital converter (ADC).
-
out_noise
: float = 0.06¶ Std deviation of Gaussian output noise (\(\sigma_\text{out}\)).
i.e. noisiness of device summation at the output.
-
out_res
: float = 0.00196078431372549¶ Number of discretization steps for ADC or resolution.
Number of discretization steps for ADC (\(<=0\) means infinite steps) or resolution (1/steps).
-
out_scale
: float = 1.0¶ Additional fixed scalar factor.
-
out_sto_round
: bool = False¶ Whether to enable stochastic rounding of ADC.
-
w_noise
: float = 0.0¶ Scale of output referred weight noise (\(\sigma_w\)) for a given
w_noise_type
.
-
w_noise_type
: aihwkit.simulator.configs.utils.WeightNoiseType = 'None'¶ Type as specified in
OutputWeightNoiseType
.Note
This noise us applied each time anew as it is referred to the output. It will not change the conductance values of the weight matrix. For the latter one can apply
diffuse_weights()
.
-
-
class
aihwkit.simulator.configs.utils.
MappingParameter
(digital_bias=True, max_input_size=512, max_output_size=512)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter related to hardware design and the mapping of logical weight matrices to physical tiles.
Caution
Some of these parameters have only an affect for modules that support tile mappings.
-
digital_bias
: bool = True¶ Whether the bias term is handled by the analog tile or kept in digital.
Note
Default is having a digital bias so that bias values are not stored onto the analog crossbar. This needs to be supported by the chip design. Set to False if the analog bias is instead situated on the the crossbar itself (as an extra column)
Note
digital_bias
is supported by all analog modules.
-
max_input_size
: int = 512¶ Maximal input size (number of columns) of the weight matrix that is handled on a single analog tile.
If the logical weight matrix size exceeds this size it will be split and mapped onto multiple analog tiles.
Caution
Only relevant for
Mapped
modules such asaihwkit.nn.modules.linear_mapped.AnalogLinearMapped
.
-
max_output_size
: int = 512¶ Maximal output size (number of rows) of the weight matrix that is handled on a single analog tile.
If the logical weight matrix size exceeds this size it will be split and mapped onto multiple analog tiles.
Caution
Only relevant for
Mapped
modules such asaihwkit.nn.modules.linear_mapped.AnalogLinearMapped
.
-
-
class
aihwkit.simulator.configs.utils.
NoiseManagementType
(value)[source]¶ Bases:
enum.Enum
Noise management type.
Noise management determines a factor \(\alpha\) how the input is reduced:
\[\mathbf{y} = \alpha\;F_\text{analog-mac}\left(\mathbf{x}/\alpha\right)\]-
ABS_MAX
= 'AbsMax'¶ Use \(\alpha\equiv\max{|\mathbf{x}|}\).
-
ABS_MAX_NP_SUM
= 'AbsMaxNPSum'¶ Assume weight value is constant and given by
nm_assumed_wmax
.Takes a worst case scenario of the weight matrix to calculate the input scale to ensure that output is not clipping. Assumed weight value is constant and given by
nm_assumed_wmax
.
-
AVERAGE_ABS_MAX
= 'AverageAbsMax'¶ Moment-based scale input scale estimation.
Computes the average abs max over the mini-batch and applies
nm_decay
to update the value with the history.Note
nm_decay
is1-momentum
and always given in mini-batches. However, the CUDA implementation does not discount values within mini-batches, whereas the CPU implementation does.
-
CONSTANT
= 'Constant'¶ A constant value (given by parameter
nm_thres
).
-
MAX
= 'Max'¶ Use \(\alpha\equiv\max{\mathbf{x}}\).
-
NONE
= 'None'¶ No noise management.
-
-
class
aihwkit.simulator.configs.utils.
PulseType
(value)[source]¶ Bases:
enum.Enum
Pulse type.
-
DETERMINISTIC_IMPLICIT
= 'DeterministicImplicit'¶ Coincidences are computed in deterministic manner.
Coincidences are calculated by \(b_l x_q d_q\) where
BL
is the desired bit length (possibly subject to dynamic adjustments usingupdate_bl_management
) and \(x_q\) and \(d_q\) are the quantized input and error values, respectively, normalized to the range \(0,\ldots,1\). It can be shown that explicit bit lines exist that generate these coincidences.
-
MEAN_COUNT
= 'MeanCount'¶ Coincidence based in prob (\(p_a p_b\)).
-
NONE
= 'None'¶ Floating point update instead of pulses.
-
NONE_WITH_DEVICE
= 'NoneWithDevice'¶ Floating point like
None
, but with analog devices (e.g. weight clipping).
-
STOCHASTIC
= 'Stochastic'¶ Two passes for plus and minus (only CPU).
-
STOCHASTIC_COMPRESSED
= 'StochasticCompressed'¶ Generates actual stochastic bit lines.
Plus and minus pulses are taken in the same pass.
-
-
class
aihwkit.simulator.configs.utils.
SimpleDriftParameter
(nu=0.0, t_0=1.0, reset_tol=1e-07)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter for a simple power law drift.
The drift as a simple power law drift without device-to-device variation or conductance dependence.
It computes: .. math:
w_{ij}*\left(\frac{t + \Delta t}{t_0}\right)^(-\nu)
-
nu
: float = 0.0¶ Average drift \(\nu\) value.
Need to non-zero to actually use the drift.
-
reset_tol
: float = 1e-07¶ Reset tolerance.
This should a number smaller than the expected weight change as it is used to detect any changes in the weight from the last drift call. Every change to the weight above this tolerance will reset the drift time.
Caution
Any write noise or diffusion on the weight might thus interfere with the drift.
-
t_0
: float = 1.0¶ Time between write and first read.
Usually assumed in milliseconds, however, it really determines the time units of
time_since_last_call
when calling the drift.
-
-
class
aihwkit.simulator.configs.utils.
UpdateParameters
(desired_bl=31, fixed_bl=True, pulse_type=<PulseType.STOCHASTIC_COMPRESSED: 'StochasticCompressed'>, res=0, x_res_implicit=0, d_res_implicit=0, sto_round=False, update_bl_management=True, update_management=True)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter that modify the update behaviour of a pulsed device.
-
d_res_implicit
: float = 0¶ Resolution of each quantization step for the error
d
.Resolution (ie. bin width) of each quantization step for the error
d
in case of DeterministicImplicit pulse trains. SeePulseTypeMap
for details.
-
desired_bl
: int = 31¶ Desired length of the pulse trains.
For update BL management, it is the maximal pulse train length.
-
fixed_bl
: bool = True¶ Whether to fix the length of the pulse trains.
See also
update_bl_management
.In case of
True
(wheredw_min
is the mean minimal weight change step size) it is:BL = desired_BL A = B = sqrt(learning_rate / (dw_min * BL));
In case of
False
:if dw_min * desired_BL < learning_rate: A = B = 1; BL = ceil(learning_rate / dw_min; else: # same as for fixed_BL=True
-
pulse_type
: aihwkit.simulator.configs.utils.PulseType = 'StochasticCompressed'¶ Switching between different pulse types.
See also
PulseTypeMap
for details.Important
Pulsing can also be turned off in which case the update is done as if in floating point and all other update related parameter are ignored.
-
res
: float = 0¶ Resolution of the update probability for the stochastic bit line generation.
Resolution ie. bin width in
0..1
) of the update probability for the stochastic bit line generation. Use -1 for turning discretization off. Can be given as number of steps as well.
-
sto_round
: bool = False¶ Whether to enable stochastic rounding.
-
update_bl_management
: bool = True¶ Whether to enable dynamical adjustment of
A
,``B``,andBL
:BL = ceil(learning_rate * abs(x_j) * abs(d_i) / weight_granularity); BL = min(BL,desired_BL); A = B = sqrt(learning_rate / (weight_granularity * BL));
The
weight_granularity
is usually equal todw_min
.
-
update_management
: bool = True¶ Whether to apply additional scaling.
After the above setting an additional scaling (always on when using update_bl_management`) is applied to account for the different input strengths. If
\[\gamma \equiv \max_i |x_i| / \max_j |d_j|\]is the ratio between the two maximal inputs, then
A
is additionally scaled by \(\gamma\) andB
is scaled by \(1/\gamma\).
-
x_res_implicit
: float = 0¶ Resolution of each quantization step for the inputs
x
.Resolution (ie. bin width) of each quantization step for the inputs
x
in case ofDeterministicImplicit
pulse trains. SeePulseTypeMap
for details.
-
-
class
aihwkit.simulator.configs.utils.
VectorUnitCellUpdatePolicy
(value)[source]¶ Bases:
enum.Enum
Vector unit cell update policy.
-
ALL
= 'All'¶ All devices updated simultaneously.
-
SINGLE_FIXED
= 'SingleFixed'¶ Device index is not changed. Can be set initially and/or updated on the fly.
-
SINGLE_RANDOM
= 'SingleRandom'¶ A single device is selected by random choice each mini-batch.
-
SINGLE_SEQUENTIAL
= 'SingleSequential'¶ Each device one at a time in sequence.
-
-
class
aihwkit.simulator.configs.utils.
WeightClipParameter
(fixed_value=-1.0, sigma=2.5, type=<WeightClipType.NONE: 'None'>)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter that clip the weights during hardware-aware training.
-
fixed_value
: float = -1.0¶ Clipping value in case of
FixedValue
type.Caution
If
fixed_value > 0
it will be also applied during other clipping types.
-
sigma
: float = 2.5¶ Sigma value for clipping for the
LayerGaussian
type.
-
type
: aihwkit.simulator.configs.utils.WeightClipType = 'None'¶ Type of clipping.
-
-
class
aihwkit.simulator.configs.utils.
WeightClipType
(value)[source]¶ Bases:
enum.Enum
Weight clipper type.
-
AVERAGE_CHANNEL_MAX
= 'AverageChannelMax'¶ Calculates the abs max of each output channel (row of the weight matrix) and takes the average as clipping value for all.
-
FIXED_VALUE
= 'FixedValue'¶ Clip to fixed value give, symmetrical around zero.
-
LAYER_GAUSSIAN
= 'LayerGaussian'¶ Calculates the second moment of the whole weight matrix and clips at
sigma
times the result symmetrically around zero.
-
NONE
= 'None'¶ None.
-
-
class
aihwkit.simulator.configs.utils.
WeightModifierParameter
(std_dev=0.0, res=0.0, sto_round=False, dorefa_clip=0.6, pdrop=0.0, enable_during_test=False, rel_to_actual_wmax=True, assumed_wmax=1.0, copy_last_column=False, coeff0=0.0105392, coeff1=0.0768, coeff2=-0.046925, type=<WeightModifierType.COPY: 'Copy'>)[source]¶ Bases:
aihwkit.simulator.configs.helpers._PrintableMixin
Parameter that modify the forward/backward weights during hardware-aware training.
-
assumed_wmax
: float = 1.0¶ Assumed weight value that is mapped to the maximal conductance.
This is typically 1.0. This parameter will be ignored if
rel_to_actual_wmax
is set.
-
coeff0
: float = 0.0105392¶
-
coeff1
: float = 0.0768¶
-
coeff2
: float = -0.046925¶ Coefficients for the
POLY
weight modifier type.See
WeightModifierType
for details.
-
copy_last_column
: bool = False¶ Whether to not apply noise to the last column (which usually contains the bias values).
-
dorefa_clip
: float = 0.6¶ Parameter for DoReFa.
-
enable_during_test
: bool = False¶ Whether to use the last modified weight matrix during testing.
Caution
This will not remove drop connect or any other noise during evaluation, and thus should only used with care.
-
pdrop
: float = 0.0¶ Drop connect probability.
Drop connect sets weights to zero with the given probability. This implements drop connect.
Important
Drop connect can be used with any other modifier type in combination.
-
rel_to_actual_wmax
: bool = True¶ Whether to calculate the abs max of the weight and apply noise relative to this number.
If set to False,
assumed_wmax
is taken as relative units.
-
res
: float = 0.0¶ Resolution of the discretization.
The invert of
res
gives the number of equal sized steps in \(-a_\text{max}\ldots,a_\text{max}\) where the \(a_\text{max}\) is either given by the abs max (ifrel_to_actual_wmax
is set) orassumed_wmax
otherwise.res
is only used in the modifier typesDoReFa
,Discretize
, andDiscretizeAddNormal
.
-
std_dev
: float = 0.0¶ Standard deviation of the added noise to the weight matrix.
This parameter affects the modifier types
AddNormal
,MultNormal
andDiscretizeAddNormal
.Note
If the parameter
rel_to_actual_wmax
is set then thestd_dev
is computed in relative terms to the abs max of the given weight matrix, otherwise it in relative terms to the assumed max, which is set byassumed_wmax
.
-
sto_round
: bool = False¶ Whether the discretization is done with stochastic rounding enabled.
sto_round
is only used in the modifier typesDoReFa
,Discretize
, andDiscretizeAddNormal
.
-
type
: aihwkit.simulator.configs.utils.WeightModifierType = 'Copy'¶ Type of the weight modification.
-
-
class
aihwkit.simulator.configs.utils.
WeightModifierType
(value)[source]¶ Bases:
enum.Enum
Weight modifier type.
-
ADD_NORMAL
= 'AddNormal'¶ Additive Gaussian noise.
-
COPY
= 'Copy'¶ Just copy, however, could also drop.
-
DISCRETIZE
= 'Discretize'¶ Quantize the weights.
-
DISCRETIZE_ADD_NORMAL
= 'DiscretizeAddNormal'¶ First discretize and then additive Gaussian noise.
-
DOREFA
= 'DoReFa'¶ DoReFa discretization.
-
MULT_NORMAL
= 'MultNormal'¶ Multiplicative Gaussian noise.
-
POLY
= 'Poly'¶ 2nd order Polynomial noise model (in terms of the weight value).
In detail, for the duration of a mini-batch, each weight will be added a Gaussian random number with the standard deviation of \(\sigma_\text{wnoise} (c_0 + c_1 w_{ij}/\omega + c_2 w_{ij}^2/\omega^2\) where \(omega\) is either the actual max weight (if
rel_to_actual_wmax
is set) or the valueassumed_wmax
.
-
-
class
aihwkit.simulator.configs.utils.
WeightNoiseType
(value)[source]¶ Bases:
enum.Enum
Output weight noise type.
The weight noise is applied for each MAC computation, while not touching the actual weight matrix but referring it to the output.
\[y_i = \sum_j w_{ij}+\xi_{ij}\]-
ADDITIVE_CONSTANT
= 'AdditiveConstant'¶ The \(\xi\sim{\cal N}(0,\sigma)\) thus all are Gaussian distributed.
\(\sigma\) is determined by
w_noise
.
-
NONE
= 'None'¶ No weight noise.
-
PCM_READ
= 'PCMRead'¶ Output-referred PCM-like read noise.
Output-referred PCM-like read noise that scales with the amount of current generated for each output line and thus scales with both conductance values and input strength.
The same general for is taken as for PCM-like statistical model of the 1/f noise during inference, see
aihwkit.inference.noise.pcm.PCMLikeNoiseModel
.
-