def create_analog_network(input_size, hidden_sizes, output_size):
"""Create the neural network using analog and digital layers.
Args:
input_size (int): size of the Tensor at the input.
hidden_sizes (list): list of sizes of the hidden layers (2 layers).
output_size (int): size of the Tensor at the output.
Returns:
nn.Module: created analog model
"""
model = AnalogSequential(
AnalogLinear(input_size,
hidden_sizes[0],
True,
rpu_config=InferenceRPUConfig()), nn.Sigmoid(),
AnalogLinear(hidden_sizes[0],
hidden_sizes[1],
True,
rpu_config=InferenceRPUConfig()), nn.Sigmoid(),
AnalogLinearMapped(hidden_sizes[1],
output_size,
True,
rpu_config=InferenceRPUConfig()),
nn.LogSoftmax(dim=1))
return model
def __init__(self,
out_size: int,
in_size: int,
rpu_config: Optional[InferenceRPUConfig] = None,
bias: bool = False,
in_trans: bool = False,
out_trans: bool = False):
rpu_config = rpu_config or InferenceRPUConfig()
# Noise model.
self.noise_model = deepcopy(rpu_config.noise_model)
# Drift compensation.
self.drift_compensation = None
if rpu_config.drift_compensation:
self.drift_compensation = deepcopy(rpu_config.drift_compensation)
self.drift_baseline = None
self.drift_readout_tensor = None # type: Optional[Tensor]
self.alpha = ones((1, ))
# Helpers.
self.reference_combined_weights = None # type: Optional[Tensor]
self.programmed_weights = None # type: Optional[Tensor]
self.nu_drift_list = None # type: Optional[List[Tensor]]
super().__init__(out_size, in_size, rpu_config, bias, in_trans,
out_trans)
def __init__(self,
out_size: int,
in_size: int,
rpu_config: Optional['InferenceRPUConfig'] = None,
bias: bool = False,
in_trans: bool = False,
out_trans: bool = False):
if not rpu_config:
# Import `InferenceRPUConfig` dynamically to avoid import cycles.
# pylint: disable=import-outside-toplevel
from aihwkit.simulator.configs import InferenceRPUConfig
rpu_config = InferenceRPUConfig()
# Noise model.
self.noise_model = deepcopy(rpu_config.noise_model)
# Drift compensation.
self.drift_compensation = None
if rpu_config.drift_compensation:
self.drift_compensation = deepcopy(rpu_config.drift_compensation)
self.drift_baseline = None
self.drift_readout_tensor = None # type: Optional[Tensor]
self.alpha = ones((1, ))
# Helpers.
self.reference_combined_weights = None # type: Optional[Tensor]
self.programmed_weights = None # type: Optional[Tensor]
self.nu_drift_list = None # type: Optional[List[Tensor]]
super().__init__(out_size, in_size, rpu_config, bias, in_trans,
out_trans)
def __init__(
self,
input_size: int,
hidden_size: int,
bias: bool,
rpu_config: Optional[RPUConfigAlias] = None,
realistic_read_write: bool = False,
):
super().__init__()
# Default to InferenceRPUConfig
if not rpu_config:
rpu_config = InferenceRPUConfig()
self.input_size = input_size
self.hidden_size = hidden_size
self.weight_ih = AnalogLinear(
input_size,
hidden_size,
bias=bias,
rpu_config=rpu_config,
realistic_read_write=realistic_read_write)
self.weight_hh = AnalogLinear(
hidden_size,
hidden_size,
bias=bias,
rpu_config=rpu_config,
realistic_read_write=realistic_read_write)
def get_rpu_config(self):
return InferenceRPUConfig()
def get_rpu_config(self):
rpu_config = InferenceRPUConfig()
rpu_config.forward.is_perfect = True
return rpu_config
TRAIN_DATASET = os.path.join('data', 'TRAIN_DATASET')
TEST_DATASET = os.path.join('data', 'TEST_DATASET')
# Path to store results
RESULTS = os.path.join('results', 'LENET5')
# Training parameters
SEED = 1
N_EPOCHS = 30
BATCH_SIZE = 8
LEARNING_RATE = 0.01
N_CLASSES = 10
# Define the properties of the neural network in terms of noise simulated during
# the inference/training pass
RPU_CONFIG = InferenceRPUConfig()
RPU_CONFIG.forward.out_res = -1. # Turn off (output) ADC discretization.
RPU_CONFIG.forward.w_noise_type = WeightNoiseType.ADDITIVE_CONSTANT
RPU_CONFIG.forward.w_noise = 0.02
RPU_CONFIG.noise_model = PCMLikeNoiseModel(g_max=25.0)
def load_images():
"""Load images for train from torchvision datasets."""
transform = transforms.Compose([transforms.ToTensor()])
train_set = datasets.MNIST(TRAIN_DATASET, download=True, train=True, transform=transform)
val_set = datasets.MNIST(TEST_DATASET, download=True, train=False, transform=transform)
train_data = torch.utils.data.DataLoader(train_set, batch_size=BATCH_SIZE, shuffle=True)
validation_data = torch.utils.data.DataLoader(val_set, batch_size=BATCH_SIZE, shuffle=False)
return train_data, validation_data
EPOCHS = 50
BATCH_SIZE = 5
SEQ_LEN = 501
RNN_CELL = AnalogGRUCell # type of RNN cell
WITH_EMBEDDING = True # RNN with embedding
WITH_BIDIR = False
USE_ANALOG_TRAINING = False # or hardware-aware training
DEVICE = torch.device('cuda') if cuda.is_compiled() else torch.device('cpu')
if USE_ANALOG_TRAINING:
# Define a RPU configuration for analog training
rpu_config = GokmenVlasovPreset()
else:
# Define an RPU configuration using inference/hardware-aware training tile
rpu_config = InferenceRPUConfig()
rpu_config.forward.out_res = -1. # Turn off (output) ADC discretization.
rpu_config.forward.w_noise_type = WeightNoiseType.ADDITIVE_CONSTANT
rpu_config.forward.w_noise = 0.02 # Short-term w-noise.
rpu_config.clip.type = WeightClipType.FIXED_VALUE
rpu_config.clip.fixed_value = 1.0
rpu_config.modifier.pdrop = 0.03 # Drop connect.
rpu_config.modifier.type = WeightModifierType.ADD_NORMAL # Fwd/bwd weight noise.
rpu_config.modifier.std_dev = 0.1
rpu_config.modifier.rel_to_actual_wmax = True
# Inference noise model.
rpu_config.noise_model = PCMLikeNoiseModel(g_max=25.0)
# drift compensation
# Imports from aihwkit.
from aihwkit.nn import AnalogLinear
from aihwkit.optim import AnalogSGD
from aihwkit.simulator.configs import InferenceRPUConfig
from aihwkit.simulator.configs.utils import (WeightNoiseType, WeightClipType,
WeightModifierType)
from aihwkit.inference import PCMLikeNoiseModel, GlobalDriftCompensation
from aihwkit.simulator.rpu_base import cuda
# Prepare the datasets (input and expected output).
x = Tensor([[0.1, 0.2, 0.4, 0.3], [0.2, 0.1, 0.1, 0.3]])
y = Tensor([[1.0, 0.5], [0.7, 0.3]])
# Define a single-layer network, using inference/hardware-aware training tile
rpu_config = InferenceRPUConfig()
rpu_config.forward.out_res = -1. # Turn off (output) ADC discretization.
rpu_config.forward.w_noise_type = WeightNoiseType.ADDITIVE_CONSTANT
rpu_config.forward.w_noise = 0.02 # Short-term w-noise.
rpu_config.clip.type = WeightClipType.FIXED_VALUE
rpu_config.clip.fixed_value = 1.0
rpu_config.modifier.pdrop = 0.03 # Drop connect.
rpu_config.modifier.type = WeightModifierType.ADD_NORMAL # Fwd/bwd weight noise.
rpu_config.modifier.std_dev = 0.1
rpu_config.modifier.rel_to_actual_wmax = True
# Inference noise model.
rpu_config.noise_model = PCMLikeNoiseModel(g_max=25.0)
# drift compensation