def init_neurons(self, input_shape):
"""Init layer neurons."""
from snntoolbox.bin.utils import get_log_keys, get_plot_keys
output_shape = self.compute_output_shape(input_shape)
self.v_thresh = k.variable(self._v_thresh, name='v_thresh')
self.mem = k.variable(self.init_membrane_potential(output_shape),
name='v_mem')
self.time = k.variable(self.dt, name='dt')
# To save memory and computations, allocate only where needed:
if self.tau_refrac > 0:
self.refrac_until = k.zeros(output_shape, name='refrac_until')
if any({
'spiketrains', 'spikerates', 'correlation', 'spikecounts',
'hist_spikerates_activations', 'operations',
'synaptic_operations_b_t', 'neuron_operations_b_t',
'spiketrains_n_b_l_t'
} & (get_plot_keys(self.config) | get_log_keys(self.config))):
self.spiketrain = k.zeros(output_shape, name='spiketrains')
if self.online_normalization:
self.spikecounts = k.zeros(output_shape, name='spikecounts')
self.max_spikerate = k.variable(0, name='max_spikerate')
if self.payloads:
self.payloads = k.zeros(output_shape, name='payloads')
self.payloads_sum = k.zeros(output_shape, name='payloads_sum')
if clamp_var:
self.spikerate = k.zeros(input_shape, name='spikerates')
self.var = k.zeros(input_shape, name='var')
if hasattr(self, 'clamp_idx'):
self.clamp_idx = self.get_clamp_idx()
def init_neurons(self, input_shape):
"""Init layer neurons."""
from snntoolbox.bin.utils import get_log_keys, get_plot_keys
output_shape = self.compute_output_shape(input_shape)
if self.v_thresh is None:
self.v_thresh = tf.Variable(self._v_thresh, name='v_thresh',
trainable=False)
if self.mem is None:
self.mem = tf.Variable(self.init_membrane_potential(output_shape),
name='v_mem', trainable=False)
if self.time is None:
self.time = tf.Variable(self.dt, name='dt', trainable=False)
# To save memory and computations, allocate only where needed:
if self.tau_refrac > 0 and self.refrac_until is None:
self.refrac_until = tf.Variable(
tf.zeros(output_shape), name='refrac_until', trainable=False)
if any({'spiketrains', 'spikerates', 'correlation', 'spikecounts',
'hist_spikerates_activations', 'operations',
'synaptic_operations_b_t', 'neuron_operations_b_t',
'spiketrains_n_b_l_t'} & (get_plot_keys(self.config) |
get_log_keys(self.config))) and self.spiketrain is None:
self.spiketrain = tf.Variable(tf.zeros(output_shape),
name='spiketrains', trainable=False)
if self.last_spiketimes is None:
self.last_spiketimes = tf.Variable(-tf.ones(output_shape),
name='last_spiketimes',
trainable=False)
def init_neurons(self, input_shape):
"""Init layer neurons."""
from snntoolbox.bin.utils import get_log_keys, get_plot_keys
output_shape = self.compute_output_shape(input_shape)
if self.v_thresh is None: # Need this check because of @tf.function.
self.v_thresh = tf.Variable(self._v_thresh,
name='v_thresh',
trainable=False)
if self.mem is None:
self.mem = tf.Variable(self.init_membrane_potential(output_shape),
name='v_mem',
trainable=False)
if self.time is None:
self.time = tf.Variable(self.dt, name='dt', trainable=False)
# To save memory and computations, allocate only where needed:
if self.tau_refrac > 0 and self.refrac_until is None:
self.refrac_until = tf.Variable(tf.zeros(output_shape),
name='refrac_until',
trainable=False)
if any({
'spiketrains', 'spikerates', 'correlation', 'spikecounts',
'hist_spikerates_activations', 'operations',
'synaptic_operations_b_t', 'neuron_operations_b_t',
'spiketrains_n_b_l_t'
} & (get_plot_keys(self.config)
| get_log_keys(self.config))) and self.spiketrain is None:
self.spiketrain = tf.Variable(tf.zeros(output_shape),
trainable=False,
name='spiketrains')
if self.online_normalization and self.spikecounts is None:
self.spikecounts = tf.Variable(tf.zeros(output_shape),
trainable=False,
name='spikecounts')
self.max_spikerate = tf.Variable(tf.zeros([1]),
trainable=False,
name='max_spikerate')
if self.config.getboolean('cell', 'payloads') \
and self.payloads is None:
self.payloads = tf.Variable(tf.zeros(output_shape),
trainable=False,
name='payloads')
self.payloads_sum = tf.Variable(tf.zeros(output_shape),
trainable=False,
name='payloads_sum')
if clamp_var and self.spikerate is None:
self.spikerate = tf.Variable(tf.zeros(input_shape),
trainable=False,
name='spikerates')
self.var = tf.Variable(tf.zeros(input_shape),
trainable=False,
name='var')
if hasattr(self, 'clamp_idx'):
self.clamp_idx = self.get_clamp_idx()
def init_neurons(self, input_shape):
"""Init layer neurons."""
from snntoolbox.bin.utils import get_log_keys, get_plot_keys
output_shape = self.compute_output_shape(input_shape)
self.mem = k.variable(self.init_membrane_potential(output_shape))
self.time = k.variable(self.dt)
if any({
'spiketrains', 'spikerates', 'correlation', 'spikecounts',
'hist_spikerates_activations', 'operations',
'synaptic_operations_b_t', 'neuron_operations_b_t',
'spiketrains_n_b_l_t'
} & (get_plot_keys(self.config) | get_log_keys(self.config))):
self.spiketrain = k.zeros(output_shape)
self.last_spiketimes = k.variable(-np.ones(output_shape))
def init_neurons(self, input_shape):
"""Init layer neurons."""
from snntoolbox.bin.utils import get_log_keys, get_plot_keys
output_shape = self.compute_output_shape(input_shape)
self.v_thresh = k.variable(self._v_thresh)
self.mem = k.variable(self.init_membrane_potential(output_shape))
self.time = k.variable(self.dt)
# To save memory and computations, allocate only where needed:
if self.tau_refrac > 0:
self.refrac_until = k.zeros(output_shape)
if any({'spiketrains', 'spikerates', 'correlation', 'spikecounts',
'hist_spikerates_activations', 'operations',
'synaptic_operations_b_t', 'neuron_operations_b_t',
'spiketrains_n_b_l_t'} & (get_plot_keys(self.config) |
get_log_keys(self.config))):
self.spiketrain = k.zeros(output_shape)
self.last_spiketimes = k.variable(-np.ones(output_shape))
