(Masse, et al., 2019): RNN with STP for Working Memory
Re-implementation of the paper with BrainPy:
Masse, Nicolas Y., Guangyu R. Yang, H. Francis Song, Xiao-Jing Wang, and David J. Freedman. “Circuit mechanisms for the maintenance and manipulation of information in working memory.” Nature neuroscience 22, no. 7 (2019): 1159-1167.
Thanks the corresponding GitHub code: https://github.com/nmasse/Short-term-plasticity-RNN
The code for the implmentation of Task please refer to the Masse_2019_STP_RNN_tasks.py.
The analysis methods please refer to the original repository: https://github.com/nmasse/Short-term-plasticity-RNN/blob/master/analysis.py
[1]:
import brainpy as bp
import brainpy.math as bm
bp.math.set_platform('cpu')
[2]:
import os
import math
import pickle
import numpy as np
from Masse_2019_STP_RNN_tasks import Task
[3]:
# Time parameters
dt = 100 # ms
dt_sec = dt / 1000
time_constant = 100 # ms
alpha = dt / time_constant
[4]:
# Loss parameters
spike_regularization = 'L2' # 'L1' or 'L2'
spike_cost = 2e-2
weight_cost = 0.
clip_max_grad_val = 0.1
[5]:
# Training specs
batch_size = 1024
learning_rate = 2e-2
[6]:
def initialize(shape, prob, size):
w = bm.random.gamma(shape, size=size)
w *= (bm.random.random(size) < prob)
return bm.asarray(w, dtype=bm.float32)
Model
[7]:
class Model(bp.layers.Module):
def __init__(self, task, num_hidden=100, name=None):
super(Model, self).__init__(name=name)
assert isinstance(task, Task)
self.task = task
# Network configuration
self.exc_inh_prop = 0.8 # excitatory/inhibitory ratio
self.conn_prob = 0.2
# Network shape
self.num_output = task.num_output
self.num_hidden = num_hidden
self.num_input = task.num_input
# EI
self.num_exc = int(self.num_hidden * self.exc_inh_prop)
self.num_inh = self.num_hidden - self.num_exc
self.EI_list = bm.ones(self.num_hidden)
self.EI_list[self.num_exc:] = -1.
self.EI_matrix = bm.diag(self.EI_list)
self.inh_index = bm.arange(self.num_exc, self.num_hidden)
# Input and noise
self.noise_rnn = math.sqrt(2 * alpha) * 0.5
# Synaptic plasticity specs
self.tau_fast = 200 # ms
self.tau_slow = 1500 # ms
self.U_stf = 0.15
self.U_std = 0.45
# Initial hidden values
self.init_h = bm.TrainVar(bm.ones((batch_size, self.num_hidden)) * 0.1)
self.h = bm.Variable(bm.ones((batch_size, self.num_hidden)) * 0.1)
# Input/recurrent/output weights
# 1. w_ir (input => recurrent)
prob = self.conn_prob * task.num_receptive_fields
self.w_ir = bm.TrainVar(initialize(0.2, prob, (self.num_input, self.num_hidden)))
self.w_ir_mask = bm.ones((self.num_input, self.num_hidden))
if task.trial_type == 'location_DMS':
self.w_ir_mask *= 0.
target_ind = [range(0, self.num_hidden, 3), range(1, self.num_hidden, 3), range(2, self.num_hidden, 3)]
for n in range(self.num_input):
u = int(n // (self.num_input / 3))
self.w_ir_mask[n, target_ind[u]] = 1.
self.w_ir *= self.w_ir_mask # only preserve
# 2. w_rr (recurrent => recurrent)
self.w_rr = bm.TrainVar(initialize(0.1, self.conn_prob, (self.num_hidden, self.num_hidden)))
self.w_rr[:, self.num_exc:] = initialize(0.2, self.conn_prob, (self.num_hidden, self.num_inh))
self.w_rr[self.num_exc:, :] = initialize(0.2, self.conn_prob, (self.num_inh, self.num_hidden))
self.w_rr_mask = bm.ones((self.num_hidden, self.num_hidden)) - bm.eye(self.num_hidden)
self.w_rr *= self.w_rr_mask # remove self-connections
self.b_rr = bm.TrainVar(bm.zeros((1, self.num_hidden)))
# 3. w_ro (input => recurrent)
self.w_ro = bm.TrainVar(initialize(0.1, self.conn_prob, (self.num_hidden, self.num_output)))
self.w_ro_mask = bm.ones((self.num_hidden, self.num_output))
self.w_ro_mask[self.num_exc:, :] = 0.
self.w_ro *= self.w_ro_mask # remove inhibitory-to-output connections
# 4. b_ro (bias)
self.b_ro = bm.TrainVar(bm.zeros((1, self.num_output)))
# Synaptic variables
# - The first row (first half neurons) are facilitating synapses
# - The second row (last half neurons) are depressing synapses
alpha_stf = bm.ones((2, int(self.num_hidden / 2)))
alpha_stf[0] = dt / self.tau_slow
alpha_stf[1] = dt / self.tau_fast
alpha_std = bm.ones((2, int(self.num_hidden / 2)))
alpha_std[0] = dt / self.tau_fast
alpha_std[1] = dt / self.tau_slow
U = bm.ones((2, int(self.num_hidden / 2)))
U[0] = 0.15
U[1] = 0.45
u = bm.ones((batch_size, 2, int(self.num_hidden / 2))) * 0.3
u[:, 0] = 0.15
u[:, 1] = 0.45
# - final
self.alpha_stf = alpha_stf.reshape((1, -1))
self.alpha_std = alpha_std.reshape((1, -1))
self.U = U.reshape((1, -1))
self.u = bm.Variable(u.reshape((batch_size, -1)))
self.x = bm.Variable(bm.ones((batch_size, self.num_hidden)))
self.y = bm.Variable(bm.ones((batch_size, self.num_output)))
self.y_hist = bm.Variable(bm.zeros((task.num_steps, batch_size, task.num_output)))
# Loss
self.loss = bm.Variable(bm.zeros(1))
self.perf_loss = bm.Variable(bm.zeros(1))
self.spike_loss = bm.Variable(bm.zeros(1))
self.weight_loss = bm.Variable(bm.zeros(1))
def reset(self):
u = bm.ones((batch_size, 2, int(self.num_hidden / 2))) * 0.3
u[:, 0] = 0.15
u[:, 1] = 0.45
self.u.value = u.reshape((batch_size, -1))
self.x.value = bm.ones((batch_size, self.num_hidden))
self.loss[:] = 0.
self.perf_loss[:] = 0.
self.spike_loss[:] = 0.
self.weight_loss[:] = 0.
def update(self, input, **kwargs):
# update STP variables
self.x += (self.alpha_std * (1 - self.x) - dt_sec * self.u * self.x * self.h)
self.u += (self.alpha_stf * (self.U - self.u) + dt_sec * self.U * (1 - self.u) * self.h)
self.x.value = bm.minimum(1., bm.relu(self.x))
self.u.value = bm.minimum(1., bm.relu(self.u))
h_post = self.u * self.x * self.h
# Update the hidden state. Only use excitatory projections from input layer to RNN
# All input and RNN activity will be non-negative
state = alpha * (input @ bm.relu(self.w_ir) + h_post @ self.w_rr + self.b_rr)
state += bm.random.normal(0, self.noise_rnn, self.h.shape)
self.h.value = bm.relu(state) + self.h * (1 - alpha)
self.y.value = self.h @ bm.relu(self.w_ro) + self.b_ro
def predict(self, inputs):
self.h[:] = self.init_h
scan = bm.make_loop(body_fun=self.update,
dyn_vars=[self.x, self.u, self.h, self.y],
out_vars=[self.y, self.h])
logits, hist_h = scan(inputs)
self.y_hist[:] = logits
return logits, hist_h
def loss_func(self, inputs, targets, mask):
logits, hist_h = self.predict(inputs)
# Calculate the performance loss
perf_loss = bp.losses.cross_entropy_loss(logits, targets, reduction='none') * mask
self.perf_loss[:] = bm.mean(perf_loss)
# L1/L2 penalty term on hidden state activity to encourage low spike rate solutions
n = 2 if spike_regularization == 'L2' else 1
self.spike_loss[:] = bm.mean(hist_h ** n)
self.weight_loss[:] = bm.mean(bm.relu(self.w_rr) ** n)
# final loss
self.loss[:] = self.perf_loss + spike_cost * self.spike_loss + weight_cost * self.weight_loss
return self.loss.mean()
Analysis
[8]:
def get_perf(target, output, mask):
"""Calculate task accuracy by comparing the actual network output to the desired output
only examine time points when test stimulus is on, e.g. when y[:,:,0] = 0 """
target = target.numpy()
output = output.numpy()
mask = mask.numpy()
mask_full = mask > 0
mask_test = mask_full * (target[:, :, 0] == 0)
mask_non_match = mask_full * (target[:, :, 1] == 1)
mask_match = mask_full * (target[:, :, 2] == 1)
target_max = np.argmax(target, axis=2)
output_max = np.argmax(output, axis=2)
match = target_max == output_max
accuracy = np.sum(match * mask_test) / np.sum(mask_test)
acc_non_match = np.sum(match * np.squeeze(mask_non_match)) / np.sum(mask_non_match)
acc_match = np.sum(match * np.squeeze(mask_match)) / np.sum(mask_match)
return accuracy, acc_non_match, acc_match
Training
[9]:
def trial(task_name, save_fn=None, num_iterations=2000, iter_between_outputs=5):
task = Task(task_name, dt=dt, tau=time_constant, batch_size=batch_size)
# trial_info = task.generate_trial(set_rule=None)
# task.plot_neural_input(trial_info)
model = Model(task)
opt = bp.optimizers.Adam(learning_rate, train_vars=model.train_vars())
grad_f = bm.grad(model.loss_func,
dyn_vars=model.vars(),
grad_vars=model.train_vars(),
return_value=True)
@bm.jit
@bp.math.function(nodes=(model, opt))
def train_op(x, y, mask):
grads, _ = grad_f(x, y, mask)
capped_gs = dict()
for key, grad in grads.items():
if 'w_rr' in key: grad *= model.w_rr_mask
elif 'w_ro' in key: grad *= model.w_ro_mask
elif 'w_ri' in key: grad *= model.w_ir_mask
capped_gs[key] = bm.clip_by_norm(grad, clip_max_grad_val)
opt.update(grads=capped_gs)
# keep track of the model performance across training
model_performance = {'accuracy': [], 'loss': [], 'perf_loss': [],
'spike_loss': [], 'weight_loss': [], 'iteration': []}
for i in range(num_iterations):
model.reset()
# generate batch of batch_train_size
trial_info = task.generate_trial(set_rule=None)
inputs = bm.array(trial_info['neural_input'], dtype=bm.float32)
targets = bm.array(trial_info['desired_output'], dtype=bm.float32)
mask = bm.array(trial_info['train_mask'], dtype=bm.float32)
# Run the model
train_op(inputs, targets, mask)
# get metrics
accuracy, _, _ = get_perf(targets, model.y_hist, mask)
model_performance['accuracy'].append(accuracy)
model_performance['loss'].append(model.loss)
model_performance['perf_loss'].append(model.perf_loss)
model_performance['spike_loss'].append(model.spike_loss)
model_performance['weight_loss'].append(model.weight_loss)
model_performance['iteration'].append(i)
# Save the network model and output model performance to screen
if i % iter_between_outputs == 0:
print(task_name +
f' Iter {i:4d}' +
f' | Accuracy {accuracy:0.4f}' +
f' | Perf loss {model.perf_loss[0]:0.4f}' +
f' | Spike loss {model.spike_loss[0]:0.4f}' +
f' | Weight loss {model.weight_loss[0]:0.4f}' +
f' | Mean activity {bm.mean(model.h):0.4f}')
if save_fn:
if not os.path.exists(os.path.dirname(save_fn)):
os.makedirs(os.path.dirname(save_fn))
# Save model and results
weights = model.train_vars().unique().dict()
results = {'weights': weights, 'parameters': {}}
for k, v in model_performance.items():
results[k] = v
pickle.dump(results, open(save_fn, 'wb'))
[10]:
trial('DMS')
DMS Iter 0 | Accuracy 0.0729 | Perf loss 2.0206 | Spike loss 6.2892 | Weight loss 0.0248 | Mean activity 3.0050
DMS Iter 5 | Accuracy 0.1117 | Perf loss 1.2712 | Spike loss 1.8198 | Weight loss 0.0225 | Mean activity 0.9206
DMS Iter 10 | Accuracy 0.2549 | Perf loss 0.6618 | Spike loss 2.2975 | Weight loss 0.0238 | Mean activity 1.0835
DMS Iter 15 | Accuracy 0.3957 | Perf loss 0.3958 | Spike loss 3.6774 | Weight loss 0.0249 | Mean activity 1.2306
DMS Iter 20 | Accuracy 0.4393 | Perf loss 0.3054 | Spike loss 3.3282 | Weight loss 0.0251 | Mean activity 1.0884
DMS Iter 25 | Accuracy 0.4510 | Perf loss 0.2711 | Spike loss 2.3361 | Weight loss 0.0250 | Mean activity 0.9458
DMS Iter 30 | Accuracy 0.4596 | Perf loss 0.2318 | Spike loss 2.2611 | Weight loss 0.0254 | Mean activity 0.8223
DMS Iter 35 | Accuracy 0.4898 | Perf loss 0.2024 | Spike loss 2.2651 | Weight loss 0.0262 | Mean activity 0.7581
DMS Iter 40 | Accuracy 0.5250 | Perf loss 0.1881 | Spike loss 2.0889 | Weight loss 0.0268 | Mean activity 0.7269
DMS Iter 45 | Accuracy 0.5355 | Perf loss 0.1775 | Spike loss 1.8978 | Weight loss 0.0272 | Mean activity 0.6891
DMS Iter 50 | Accuracy 0.5391 | Perf loss 0.1787 | Spike loss 1.7190 | Weight loss 0.0279 | Mean activity 0.6473
DMS Iter 55 | Accuracy 0.5699 | Perf loss 0.1655 | Spike loss 1.6436 | Weight loss 0.0284 | Mean activity 0.6288
DMS Iter 60 | Accuracy 0.5637 | Perf loss 0.1631 | Spike loss 1.5490 | Weight loss 0.0291 | Mean activity 0.6208
DMS Iter 65 | Accuracy 0.6098 | Perf loss 0.1489 | Spike loss 1.5589 | Weight loss 0.0300 | Mean activity 0.6087
DMS Iter 70 | Accuracy 0.6285 | Perf loss 0.1440 | Spike loss 1.5051 | Weight loss 0.0312 | Mean activity 0.6047
DMS Iter 75 | Accuracy 0.6564 | Perf loss 0.1361 | Spike loss 1.5515 | Weight loss 0.0324 | Mean activity 0.6102
DMS Iter 80 | Accuracy 0.6707 | Perf loss 0.1302 | Spike loss 1.5195 | Weight loss 0.0334 | Mean activity 0.6126
DMS Iter 85 | Accuracy 0.6896 | Perf loss 0.1299 | Spike loss 1.4762 | Weight loss 0.0340 | Mean activity 0.6181
DMS Iter 90 | Accuracy 0.7006 | Perf loss 0.1200 | Spike loss 1.4774 | Weight loss 0.0349 | Mean activity 0.6169
DMS Iter 95 | Accuracy 0.7154 | Perf loss 0.1138 | Spike loss 1.4967 | Weight loss 0.0360 | Mean activity 0.6233
DMS Iter 100 | Accuracy 0.7258 | Perf loss 0.1136 | Spike loss 1.4727 | Weight loss 0.0369 | Mean activity 0.6211
DMS Iter 105 | Accuracy 0.7297 | Perf loss 0.1133 | Spike loss 1.4581 | Weight loss 0.0377 | Mean activity 0.6160
DMS Iter 110 | Accuracy 0.7355 | Perf loss 0.1098 | Spike loss 1.5052 | Weight loss 0.0386 | Mean activity 0.6125
DMS Iter 115 | Accuracy 0.7588 | Perf loss 0.1003 | Spike loss 1.4705 | Weight loss 0.0394 | Mean activity 0.6215
DMS Iter 120 | Accuracy 0.7725 | Perf loss 0.0937 | Spike loss 1.4866 | Weight loss 0.0402 | Mean activity 0.6305
DMS Iter 125 | Accuracy 0.7869 | Perf loss 0.0915 | Spike loss 1.5049 | Weight loss 0.0412 | Mean activity 0.6370
DMS Iter 130 | Accuracy 0.8078 | Perf loss 0.0841 | Spike loss 1.5549 | Weight loss 0.0422 | Mean activity 0.6438
DMS Iter 135 | Accuracy 0.8010 | Perf loss 0.0842 | Spike loss 1.5112 | Weight loss 0.0432 | Mean activity 0.6344
DMS Iter 140 | Accuracy 0.8080 | Perf loss 0.0833 | Spike loss 1.5285 | Weight loss 0.0440 | Mean activity 0.6270
DMS Iter 145 | Accuracy 0.8020 | Perf loss 0.0891 | Spike loss 1.4099 | Weight loss 0.0445 | Mean activity 0.6288
DMS Iter 150 | Accuracy 0.8230 | Perf loss 0.0769 | Spike loss 1.5634 | Weight loss 0.0458 | Mean activity 0.6360
DMS Iter 155 | Accuracy 0.8105 | Perf loss 0.0785 | Spike loss 1.4852 | Weight loss 0.0463 | Mean activity 0.6373
DMS Iter 160 | Accuracy 0.8307 | Perf loss 0.0774 | Spike loss 1.5121 | Weight loss 0.0471 | Mean activity 0.6303
DMS Iter 165 | Accuracy 0.8443 | Perf loss 0.0692 | Spike loss 1.5166 | Weight loss 0.0480 | Mean activity 0.6297
DMS Iter 170 | Accuracy 0.8475 | Perf loss 0.0688 | Spike loss 1.4785 | Weight loss 0.0487 | Mean activity 0.6227
DMS Iter 175 | Accuracy 0.8520 | Perf loss 0.0654 | Spike loss 1.5093 | Weight loss 0.0495 | Mean activity 0.6261
DMS Iter 180 | Accuracy 0.8693 | Perf loss 0.0602 | Spike loss 1.5176 | Weight loss 0.0504 | Mean activity 0.6212
DMS Iter 185 | Accuracy 0.8762 | Perf loss 0.0594 | Spike loss 1.5365 | Weight loss 0.0513 | Mean activity 0.6239
DMS Iter 190 | Accuracy 0.8658 | Perf loss 0.0594 | Spike loss 1.4823 | Weight loss 0.0520 | Mean activity 0.6207
DMS Iter 195 | Accuracy 0.8830 | Perf loss 0.0530 | Spike loss 1.5463 | Weight loss 0.0528 | Mean activity 0.6169
DMS Iter 200 | Accuracy 0.8783 | Perf loss 0.0532 | Spike loss 1.5334 | Weight loss 0.0536 | Mean activity 0.6191
DMS Iter 205 | Accuracy 0.8734 | Perf loss 0.0540 | Spike loss 1.4936 | Weight loss 0.0542 | Mean activity 0.6113
DMS Iter 210 | Accuracy 0.8865 | Perf loss 0.0515 | Spike loss 1.5378 | Weight loss 0.0553 | Mean activity 0.6094
DMS Iter 215 | Accuracy 0.8926 | Perf loss 0.0490 | Spike loss 1.4939 | Weight loss 0.0559 | Mean activity 0.5995
DMS Iter 220 | Accuracy 0.8986 | Perf loss 0.0470 | Spike loss 1.5006 | Weight loss 0.0565 | Mean activity 0.5876
DMS Iter 225 | Accuracy 0.8990 | Perf loss 0.0450 | Spike loss 1.5068 | Weight loss 0.0572 | Mean activity 0.5835
DMS Iter 230 | Accuracy 0.8959 | Perf loss 0.0449 | Spike loss 1.4902 | Weight loss 0.0578 | Mean activity 0.5868
DMS Iter 235 | Accuracy 0.9049 | Perf loss 0.0419 | Spike loss 1.4877 | Weight loss 0.0584 | Mean activity 0.5707
DMS Iter 240 | Accuracy 0.9023 | Perf loss 0.0445 | Spike loss 1.4508 | Weight loss 0.0589 | Mean activity 0.5667
DMS Iter 245 | Accuracy 0.9000 | Perf loss 0.0452 | Spike loss 1.4584 | Weight loss 0.0594 | Mean activity 0.5579
DMS Iter 250 | Accuracy 0.9043 | Perf loss 0.0413 | Spike loss 1.4353 | Weight loss 0.0598 | Mean activity 0.5533
DMS Iter 255 | Accuracy 0.9080 | Perf loss 0.0415 | Spike loss 1.4407 | Weight loss 0.0604 | Mean activity 0.5449
DMS Iter 260 | Accuracy 0.9102 | Perf loss 0.0414 | Spike loss 1.4392 | Weight loss 0.0608 | Mean activity 0.5452
DMS Iter 265 | Accuracy 0.9127 | Perf loss 0.0398 | Spike loss 1.4335 | Weight loss 0.0610 | Mean activity 0.5400
DMS Iter 270 | Accuracy 0.9062 | Perf loss 0.0432 | Spike loss 1.4092 | Weight loss 0.0612 | Mean activity 0.5347
DMS Iter 275 | Accuracy 0.9148 | Perf loss 0.0392 | Spike loss 1.4238 | Weight loss 0.0617 | Mean activity 0.5328
DMS Iter 280 | Accuracy 0.9182 | Perf loss 0.0378 | Spike loss 1.4112 | Weight loss 0.0620 | Mean activity 0.5328
DMS Iter 285 | Accuracy 0.9213 | Perf loss 0.0373 | Spike loss 1.3898 | Weight loss 0.0621 | Mean activity 0.5277
DMS Iter 290 | Accuracy 0.9176 | Perf loss 0.0371 | Spike loss 1.3848 | Weight loss 0.0625 | Mean activity 0.5201
DMS Iter 295 | Accuracy 0.9213 | Perf loss 0.0365 | Spike loss 1.3599 | Weight loss 0.0627 | Mean activity 0.5143
DMS Iter 300 | Accuracy 0.9256 | Perf loss 0.0334 | Spike loss 1.3885 | Weight loss 0.0632 | Mean activity 0.5118
DMS Iter 305 | Accuracy 0.9178 | Perf loss 0.0366 | Spike loss 1.3843 | Weight loss 0.0636 | Mean activity 0.5078
DMS Iter 310 | Accuracy 0.9162 | Perf loss 0.0366 | Spike loss 1.3603 | Weight loss 0.0639 | Mean activity 0.4967
DMS Iter 315 | Accuracy 0.9240 | Perf loss 0.0343 | Spike loss 1.3474 | Weight loss 0.0644 | Mean activity 0.4901
DMS Iter 320 | Accuracy 0.9201 | Perf loss 0.0363 | Spike loss 1.3436 | Weight loss 0.0645 | Mean activity 0.4903
DMS Iter 325 | Accuracy 0.9262 | Perf loss 0.0339 | Spike loss 1.3447 | Weight loss 0.0648 | Mean activity 0.4880
DMS Iter 330 | Accuracy 0.9238 | Perf loss 0.0336 | Spike loss 1.3444 | Weight loss 0.0651 | Mean activity 0.4815
DMS Iter 335 | Accuracy 0.9285 | Perf loss 0.0328 | Spike loss 1.2826 | Weight loss 0.0652 | Mean activity 0.4755
DMS Iter 340 | Accuracy 0.9295 | Perf loss 0.0324 | Spike loss 1.3139 | Weight loss 0.0657 | Mean activity 0.4716
DMS Iter 345 | Accuracy 0.9260 | Perf loss 0.0354 | Spike loss 1.3040 | Weight loss 0.0661 | Mean activity 0.4734
DMS Iter 350 | Accuracy 0.9277 | Perf loss 0.0332 | Spike loss 1.3203 | Weight loss 0.0664 | Mean activity 0.4695
DMS Iter 355 | Accuracy 0.9307 | Perf loss 0.0329 | Spike loss 1.3093 | Weight loss 0.0667 | Mean activity 0.4646
DMS Iter 360 | Accuracy 0.9271 | Perf loss 0.0322 | Spike loss 1.2897 | Weight loss 0.0670 | Mean activity 0.4529
DMS Iter 365 | Accuracy 0.9307 | Perf loss 0.0311 | Spike loss 1.3048 | Weight loss 0.0673 | Mean activity 0.4643
DMS Iter 370 | Accuracy 0.9291 | Perf loss 0.0309 | Spike loss 1.2920 | Weight loss 0.0676 | Mean activity 0.4562
DMS Iter 375 | Accuracy 0.9268 | Perf loss 0.0348 | Spike loss 1.2709 | Weight loss 0.0679 | Mean activity 0.4530
DMS Iter 380 | Accuracy 0.9311 | Perf loss 0.0321 | Spike loss 1.2744 | Weight loss 0.0683 | Mean activity 0.4460
DMS Iter 385 | Accuracy 0.9354 | Perf loss 0.0287 | Spike loss 1.2664 | Weight loss 0.0687 | Mean activity 0.4462
DMS Iter 390 | Accuracy 0.9316 | Perf loss 0.0319 | Spike loss 1.2684 | Weight loss 0.0691 | Mean activity 0.4372
DMS Iter 395 | Accuracy 0.9434 | Perf loss 0.0282 | Spike loss 1.2596 | Weight loss 0.0695 | Mean activity 0.4421
DMS Iter 400 | Accuracy 0.9371 | Perf loss 0.0297 | Spike loss 1.2364 | Weight loss 0.0697 | Mean activity 0.4407
DMS Iter 405 | Accuracy 0.9318 | Perf loss 0.0300 | Spike loss 1.2449 | Weight loss 0.0699 | Mean activity 0.4404
DMS Iter 410 | Accuracy 0.9437 | Perf loss 0.0263 | Spike loss 1.2660 | Weight loss 0.0703 | Mean activity 0.4422
DMS Iter 415 | Accuracy 0.9393 | Perf loss 0.0305 | Spike loss 1.2473 | Weight loss 0.0704 | Mean activity 0.4310
DMS Iter 420 | Accuracy 0.9422 | Perf loss 0.0289 | Spike loss 1.2085 | Weight loss 0.0707 | Mean activity 0.4189
DMS Iter 425 | Accuracy 0.9400 | Perf loss 0.0270 | Spike loss 1.2351 | Weight loss 0.0710 | Mean activity 0.4172
DMS Iter 430 | Accuracy 0.9352 | Perf loss 0.0301 | Spike loss 1.2222 | Weight loss 0.0711 | Mean activity 0.4254
DMS Iter 435 | Accuracy 0.9373 | Perf loss 0.0293 | Spike loss 1.2165 | Weight loss 0.0714 | Mean activity 0.4188
DMS Iter 440 | Accuracy 0.9465 | Perf loss 0.0274 | Spike loss 1.2483 | Weight loss 0.0720 | Mean activity 0.4210
DMS Iter 445 | Accuracy 0.9459 | Perf loss 0.0265 | Spike loss 1.1956 | Weight loss 0.0721 | Mean activity 0.4091
DMS Iter 450 | Accuracy 0.9469 | Perf loss 0.0263 | Spike loss 1.1811 | Weight loss 0.0722 | Mean activity 0.3961
DMS Iter 455 | Accuracy 0.9469 | Perf loss 0.0258 | Spike loss 1.1814 | Weight loss 0.0724 | Mean activity 0.4043
DMS Iter 460 | Accuracy 0.9400 | Perf loss 0.0294 | Spike loss 1.1727 | Weight loss 0.0726 | Mean activity 0.3990
DMS Iter 465 | Accuracy 0.9437 | Perf loss 0.0264 | Spike loss 1.1694 | Weight loss 0.0730 | Mean activity 0.4006
DMS Iter 470 | Accuracy 0.9496 | Perf loss 0.0253 | Spike loss 1.1539 | Weight loss 0.0730 | Mean activity 0.3957
DMS Iter 475 | Accuracy 0.9432 | Perf loss 0.0271 | Spike loss 1.1664 | Weight loss 0.0732 | Mean activity 0.3937
DMS Iter 480 | Accuracy 0.9379 | Perf loss 0.0290 | Spike loss 1.1903 | Weight loss 0.0736 | Mean activity 0.3936
DMS Iter 485 | Accuracy 0.9482 | Perf loss 0.0229 | Spike loss 1.1889 | Weight loss 0.0739 | Mean activity 0.3943
DMS Iter 490 | Accuracy 0.9439 | Perf loss 0.0274 | Spike loss 1.1412 | Weight loss 0.0741 | Mean activity 0.3878
DMS Iter 495 | Accuracy 0.9459 | Perf loss 0.0254 | Spike loss 1.1555 | Weight loss 0.0747 | Mean activity 0.3793
DMS Iter 500 | Accuracy 0.9555 | Perf loss 0.0232 | Spike loss 1.1706 | Weight loss 0.0750 | Mean activity 0.3827
DMS Iter 505 | Accuracy 0.9523 | Perf loss 0.0234 | Spike loss 1.1809 | Weight loss 0.0754 | Mean activity 0.3769
DMS Iter 510 | Accuracy 0.9437 | Perf loss 0.0259 | Spike loss 1.1405 | Weight loss 0.0754 | Mean activity 0.3764
DMS Iter 515 | Accuracy 0.9437 | Perf loss 0.0261 | Spike loss 1.1713 | Weight loss 0.0758 | Mean activity 0.3685
DMS Iter 520 | Accuracy 0.9537 | Perf loss 0.0227 | Spike loss 1.1599 | Weight loss 0.0760 | Mean activity 0.3626
DMS Iter 525 | Accuracy 0.9561 | Perf loss 0.0221 | Spike loss 1.1292 | Weight loss 0.0762 | Mean activity 0.3657
DMS Iter 530 | Accuracy 0.9498 | Perf loss 0.0246 | Spike loss 1.1414 | Weight loss 0.0763 | Mean activity 0.3633
DMS Iter 535 | Accuracy 0.9539 | Perf loss 0.0215 | Spike loss 1.1352 | Weight loss 0.0766 | Mean activity 0.3681
DMS Iter 540 | Accuracy 0.9490 | Perf loss 0.0263 | Spike loss 1.1245 | Weight loss 0.0771 | Mean activity 0.3622
DMS Iter 545 | Accuracy 0.9463 | Perf loss 0.0237 | Spike loss 1.1268 | Weight loss 0.0774 | Mean activity 0.3632
DMS Iter 550 | Accuracy 0.9596 | Perf loss 0.0223 | Spike loss 1.1361 | Weight loss 0.0777 | Mean activity 0.3580
DMS Iter 555 | Accuracy 0.9498 | Perf loss 0.0239 | Spike loss 1.1528 | Weight loss 0.0779 | Mean activity 0.3609
DMS Iter 560 | Accuracy 0.9602 | Perf loss 0.0211 | Spike loss 1.1440 | Weight loss 0.0780 | Mean activity 0.3579
DMS Iter 565 | Accuracy 0.9504 | Perf loss 0.0233 | Spike loss 1.1277 | Weight loss 0.0780 | Mean activity 0.3562
DMS Iter 570 | Accuracy 0.9545 | Perf loss 0.0239 | Spike loss 1.1364 | Weight loss 0.0783 | Mean activity 0.3605
DMS Iter 575 | Accuracy 0.9545 | Perf loss 0.0246 | Spike loss 1.0952 | Weight loss 0.0784 | Mean activity 0.3581
DMS Iter 580 | Accuracy 0.9563 | Perf loss 0.0216 | Spike loss 1.2180 | Weight loss 0.0796 | Mean activity 0.3674
DMS Iter 585 | Accuracy 0.9557 | Perf loss 0.0226 | Spike loss 1.1673 | Weight loss 0.0796 | Mean activity 0.3586
DMS Iter 590 | Accuracy 0.9588 | Perf loss 0.0201 | Spike loss 1.1738 | Weight loss 0.0797 | Mean activity 0.3545
DMS Iter 595 | Accuracy 0.9529 | Perf loss 0.0231 | Spike loss 1.1569 | Weight loss 0.0797 | Mean activity 0.3564
DMS Iter 600 | Accuracy 0.9604 | Perf loss 0.0209 | Spike loss 1.1535 | Weight loss 0.0800 | Mean activity 0.3508
DMS Iter 605 | Accuracy 0.9600 | Perf loss 0.0206 | Spike loss 1.1561 | Weight loss 0.0806 | Mean activity 0.3410
DMS Iter 610 | Accuracy 0.9613 | Perf loss 0.0197 | Spike loss 1.1349 | Weight loss 0.0809 | Mean activity 0.3484
DMS Iter 615 | Accuracy 0.9525 | Perf loss 0.0249 | Spike loss 1.1244 | Weight loss 0.0812 | Mean activity 0.3356
DMS Iter 620 | Accuracy 0.9619 | Perf loss 0.0192 | Spike loss 1.1498 | Weight loss 0.0816 | Mean activity 0.3365
DMS Iter 625 | Accuracy 0.9582 | Perf loss 0.0205 | Spike loss 1.1602 | Weight loss 0.0818 | Mean activity 0.3425
DMS Iter 630 | Accuracy 0.9691 | Perf loss 0.0173 | Spike loss 1.1452 | Weight loss 0.0821 | Mean activity 0.3397
DMS Iter 635 | Accuracy 0.9611 | Perf loss 0.0196 | Spike loss 1.1205 | Weight loss 0.0822 | Mean activity 0.3302
DMS Iter 640 | Accuracy 0.9646 | Perf loss 0.0180 | Spike loss 1.1164 | Weight loss 0.0823 | Mean activity 0.3342
DMS Iter 645 | Accuracy 0.9605 | Perf loss 0.0181 | Spike loss 1.1117 | Weight loss 0.0827 | Mean activity 0.3379
DMS Iter 650 | Accuracy 0.9686 | Perf loss 0.0171 | Spike loss 1.0920 | Weight loss 0.0828 | Mean activity 0.3318
DMS Iter 655 | Accuracy 0.9684 | Perf loss 0.0174 | Spike loss 1.1037 | Weight loss 0.0830 | Mean activity 0.3317
DMS Iter 660 | Accuracy 0.9639 | Perf loss 0.0190 | Spike loss 1.1031 | Weight loss 0.0830 | Mean activity 0.3317
DMS Iter 665 | Accuracy 0.9637 | Perf loss 0.0195 | Spike loss 1.0825 | Weight loss 0.0830 | Mean activity 0.3316
DMS Iter 670 | Accuracy 0.9662 | Perf loss 0.0180 | Spike loss 1.0904 | Weight loss 0.0835 | Mean activity 0.3244
DMS Iter 675 | Accuracy 0.9652 | Perf loss 0.0175 | Spike loss 1.0942 | Weight loss 0.0837 | Mean activity 0.3286
DMS Iter 680 | Accuracy 0.9713 | Perf loss 0.0163 | Spike loss 1.0984 | Weight loss 0.0842 | Mean activity 0.3236
DMS Iter 685 | Accuracy 0.9684 | Perf loss 0.0170 | Spike loss 1.0945 | Weight loss 0.0845 | Mean activity 0.3205
DMS Iter 690 | Accuracy 0.9664 | Perf loss 0.0180 | Spike loss 1.1058 | Weight loss 0.0849 | Mean activity 0.3193
DMS Iter 695 | Accuracy 0.9691 | Perf loss 0.0180 | Spike loss 1.0934 | Weight loss 0.0850 | Mean activity 0.3311
DMS Iter 700 | Accuracy 0.9619 | Perf loss 0.0177 | Spike loss 1.1012 | Weight loss 0.0852 | Mean activity 0.3207
DMS Iter 705 | Accuracy 0.9674 | Perf loss 0.0173 | Spike loss 1.0581 | Weight loss 0.0849 | Mean activity 0.3214
DMS Iter 710 | Accuracy 0.9613 | Perf loss 0.0194 | Spike loss 1.1214 | Weight loss 0.0856 | Mean activity 0.3170
DMS Iter 715 | Accuracy 0.9695 | Perf loss 0.0161 | Spike loss 1.1038 | Weight loss 0.0858 | Mean activity 0.3070
DMS Iter 720 | Accuracy 0.9729 | Perf loss 0.0154 | Spike loss 1.0821 | Weight loss 0.0861 | Mean activity 0.3047
DMS Iter 725 | Accuracy 0.9697 | Perf loss 0.0166 | Spike loss 1.0758 | Weight loss 0.0861 | Mean activity 0.3050
DMS Iter 730 | Accuracy 0.9760 | Perf loss 0.0138 | Spike loss 1.0794 | Weight loss 0.0862 | Mean activity 0.3033
DMS Iter 735 | Accuracy 0.9654 | Perf loss 0.0177 | Spike loss 1.0623 | Weight loss 0.0863 | Mean activity 0.3100
DMS Iter 740 | Accuracy 0.9723 | Perf loss 0.0160 | Spike loss 1.0798 | Weight loss 0.0867 | Mean activity 0.3059
DMS Iter 745 | Accuracy 0.9691 | Perf loss 0.0160 | Spike loss 1.0602 | Weight loss 0.0865 | Mean activity 0.3097
DMS Iter 750 | Accuracy 0.9715 | Perf loss 0.0153 | Spike loss 1.0657 | Weight loss 0.0867 | Mean activity 0.3049
DMS Iter 755 | Accuracy 0.9805 | Perf loss 0.0117 | Spike loss 1.0629 | Weight loss 0.0868 | Mean activity 0.3035
DMS Iter 760 | Accuracy 0.9758 | Perf loss 0.0131 | Spike loss 1.0551 | Weight loss 0.0871 | Mean activity 0.3066
DMS Iter 765 | Accuracy 0.9756 | Perf loss 0.0140 | Spike loss 1.0380 | Weight loss 0.0871 | Mean activity 0.3023
DMS Iter 770 | Accuracy 0.9758 | Perf loss 0.0132 | Spike loss 1.0508 | Weight loss 0.0876 | Mean activity 0.2929
DMS Iter 775 | Accuracy 0.9760 | Perf loss 0.0135 | Spike loss 1.0255 | Weight loss 0.0877 | Mean activity 0.2924
DMS Iter 780 | Accuracy 0.9770 | Perf loss 0.0128 | Spike loss 1.0267 | Weight loss 0.0880 | Mean activity 0.2884
DMS Iter 785 | Accuracy 0.9775 | Perf loss 0.0131 | Spike loss 1.0410 | Weight loss 0.0886 | Mean activity 0.2877
DMS Iter 790 | Accuracy 0.9793 | Perf loss 0.0122 | Spike loss 1.0329 | Weight loss 0.0889 | Mean activity 0.2928
DMS Iter 795 | Accuracy 0.9738 | Perf loss 0.0138 | Spike loss 1.0269 | Weight loss 0.0891 | Mean activity 0.2918
DMS Iter 800 | Accuracy 0.9746 | Perf loss 0.0153 | Spike loss 1.0111 | Weight loss 0.0890 | Mean activity 0.2920
DMS Iter 805 | Accuracy 0.9783 | Perf loss 0.0109 | Spike loss 1.0226 | Weight loss 0.0894 | Mean activity 0.2952
DMS Iter 810 | Accuracy 0.9725 | Perf loss 0.0151 | Spike loss 1.0158 | Weight loss 0.0894 | Mean activity 0.2947
DMS Iter 815 | Accuracy 0.9816 | Perf loss 0.0109 | Spike loss 1.0109 | Weight loss 0.0894 | Mean activity 0.2953
DMS Iter 820 | Accuracy 0.9793 | Perf loss 0.0118 | Spike loss 0.9978 | Weight loss 0.0896 | Mean activity 0.2907
DMS Iter 825 | Accuracy 0.9799 | Perf loss 0.0122 | Spike loss 1.0063 | Weight loss 0.0897 | Mean activity 0.2862
DMS Iter 830 | Accuracy 0.9814 | Perf loss 0.0111 | Spike loss 0.9986 | Weight loss 0.0897 | Mean activity 0.2804
DMS Iter 835 | Accuracy 0.9814 | Perf loss 0.0118 | Spike loss 0.9986 | Weight loss 0.0900 | Mean activity 0.2782
DMS Iter 840 | Accuracy 0.9838 | Perf loss 0.0106 | Spike loss 0.9984 | Weight loss 0.0903 | Mean activity 0.2789
DMS Iter 845 | Accuracy 0.9762 | Perf loss 0.0140 | Spike loss 0.9831 | Weight loss 0.0904 | Mean activity 0.2778
DMS Iter 850 | Accuracy 0.9803 | Perf loss 0.0106 | Spike loss 0.9999 | Weight loss 0.0909 | Mean activity 0.2752
DMS Iter 855 | Accuracy 0.9781 | Perf loss 0.0128 | Spike loss 0.9889 | Weight loss 0.0910 | Mean activity 0.2821
DMS Iter 860 | Accuracy 0.9801 | Perf loss 0.0115 | Spike loss 1.0079 | Weight loss 0.0909 | Mean activity 0.2753
DMS Iter 865 | Accuracy 0.9809 | Perf loss 0.0109 | Spike loss 1.0170 | Weight loss 0.0913 | Mean activity 0.2783
DMS Iter 870 | Accuracy 0.9797 | Perf loss 0.0115 | Spike loss 1.0154 | Weight loss 0.0916 | Mean activity 0.2808
DMS Iter 875 | Accuracy 0.9773 | Perf loss 0.0125 | Spike loss 1.0115 | Weight loss 0.0918 | Mean activity 0.2851
DMS Iter 880 | Accuracy 0.9828 | Perf loss 0.0112 | Spike loss 1.0216 | Weight loss 0.0923 | Mean activity 0.2763
DMS Iter 885 | Accuracy 0.9803 | Perf loss 0.0110 | Spike loss 1.0089 | Weight loss 0.0925 | Mean activity 0.2743
DMS Iter 890 | Accuracy 0.9836 | Perf loss 0.0102 | Spike loss 0.9905 | Weight loss 0.0924 | Mean activity 0.2796
DMS Iter 895 | Accuracy 0.9816 | Perf loss 0.0103 | Spike loss 0.9914 | Weight loss 0.0925 | Mean activity 0.2764
DMS Iter 900 | Accuracy 0.9820 | Perf loss 0.0106 | Spike loss 0.9942 | Weight loss 0.0928 | Mean activity 0.2765
DMS Iter 905 | Accuracy 0.9822 | Perf loss 0.0106 | Spike loss 0.9963 | Weight loss 0.0931 | Mean activity 0.2748
DMS Iter 910 | Accuracy 0.9828 | Perf loss 0.0104 | Spike loss 0.9907 | Weight loss 0.0934 | Mean activity 0.2707
DMS Iter 915 | Accuracy 0.9832 | Perf loss 0.0101 | Spike loss 0.9815 | Weight loss 0.0936 | Mean activity 0.2690
DMS Iter 920 | Accuracy 0.9842 | Perf loss 0.0094 | Spike loss 0.9863 | Weight loss 0.0937 | Mean activity 0.2608
DMS Iter 925 | Accuracy 0.9830 | Perf loss 0.0108 | Spike loss 0.9753 | Weight loss 0.0936 | Mean activity 0.2661
DMS Iter 930 | Accuracy 0.9822 | Perf loss 0.0110 | Spike loss 0.9808 | Weight loss 0.0938 | Mean activity 0.2702
DMS Iter 935 | Accuracy 0.9807 | Perf loss 0.0109 | Spike loss 0.9825 | Weight loss 0.0938 | Mean activity 0.2681
DMS Iter 940 | Accuracy 0.9854 | Perf loss 0.0090 | Spike loss 0.9957 | Weight loss 0.0943 | Mean activity 0.2655
DMS Iter 945 | Accuracy 0.9848 | Perf loss 0.0091 | Spike loss 0.9684 | Weight loss 0.0943 | Mean activity 0.2561
DMS Iter 950 | Accuracy 0.9846 | Perf loss 0.0095 | Spike loss 0.9679 | Weight loss 0.0945 | Mean activity 0.2621
DMS Iter 955 | Accuracy 0.9844 | Perf loss 0.0091 | Spike loss 0.9778 | Weight loss 0.0950 | Mean activity 0.2624
DMS Iter 960 | Accuracy 0.9824 | Perf loss 0.0100 | Spike loss 0.9568 | Weight loss 0.0954 | Mean activity 0.2707
DMS Iter 965 | Accuracy 0.9824 | Perf loss 0.0097 | Spike loss 0.9826 | Weight loss 0.0958 | Mean activity 0.2603
DMS Iter 970 | Accuracy 0.9818 | Perf loss 0.0102 | Spike loss 0.9781 | Weight loss 0.0961 | Mean activity 0.2702
DMS Iter 975 | Accuracy 0.9869 | Perf loss 0.0085 | Spike loss 0.9676 | Weight loss 0.0962 | Mean activity 0.2646
DMS Iter 980 | Accuracy 0.9840 | Perf loss 0.0088 | Spike loss 0.9723 | Weight loss 0.0963 | Mean activity 0.2578
DMS Iter 985 | Accuracy 0.9836 | Perf loss 0.0095 | Spike loss 0.9537 | Weight loss 0.0963 | Mean activity 0.2568
DMS Iter 990 | Accuracy 0.9854 | Perf loss 0.0108 | Spike loss 0.9620 | Weight loss 0.0962 | Mean activity 0.2500
DMS Iter 995 | Accuracy 0.9811 | Perf loss 0.0124 | Spike loss 0.9923 | Weight loss 0.0971 | Mean activity 0.2550
DMS Iter 1000 | Accuracy 0.9809 | Perf loss 0.0120 | Spike loss 1.0522 | Weight loss 0.0985 | Mean activity 0.2516
DMS Iter 1005 | Accuracy 0.9865 | Perf loss 0.0086 | Spike loss 1.0562 | Weight loss 0.0990 | Mean activity 0.2552
DMS Iter 1010 | Accuracy 0.9812 | Perf loss 0.0119 | Spike loss 1.0279 | Weight loss 0.0992 | Mean activity 0.2534
DMS Iter 1015 | Accuracy 0.9805 | Perf loss 0.0139 | Spike loss 0.9988 | Weight loss 0.0991 | Mean activity 0.2551
DMS Iter 1020 | Accuracy 0.9840 | Perf loss 0.0098 | Spike loss 1.1007 | Weight loss 0.0996 | Mean activity 0.2626
DMS Iter 1025 | Accuracy 0.9805 | Perf loss 0.0119 | Spike loss 1.1064 | Weight loss 0.1000 | Mean activity 0.2641
DMS Iter 1030 | Accuracy 0.9840 | Perf loss 0.0087 | Spike loss 1.1182 | Weight loss 0.1005 | Mean activity 0.2715
DMS Iter 1035 | Accuracy 0.9871 | Perf loss 0.0082 | Spike loss 1.0944 | Weight loss 0.1006 | Mean activity 0.2706
DMS Iter 1040 | Accuracy 0.9836 | Perf loss 0.0090 | Spike loss 1.0599 | Weight loss 0.1007 | Mean activity 0.2604
DMS Iter 1045 | Accuracy 0.9891 | Perf loss 0.0081 | Spike loss 1.0744 | Weight loss 0.1012 | Mean activity 0.2524
DMS Iter 1050 | Accuracy 0.9861 | Perf loss 0.0077 | Spike loss 1.0416 | Weight loss 0.1014 | Mean activity 0.2527
DMS Iter 1055 | Accuracy 0.9859 | Perf loss 0.0077 | Spike loss 1.0153 | Weight loss 0.1012 | Mean activity 0.2563
DMS Iter 1060 | Accuracy 0.9877 | Perf loss 0.0077 | Spike loss 0.9900 | Weight loss 0.1011 | Mean activity 0.2584
DMS Iter 1065 | Accuracy 0.9855 | Perf loss 0.0086 | Spike loss 0.9899 | Weight loss 0.1013 | Mean activity 0.2575
DMS Iter 1070 | Accuracy 0.9871 | Perf loss 0.0080 | Spike loss 0.9903 | Weight loss 0.1011 | Mean activity 0.2509
DMS Iter 1075 | Accuracy 0.9879 | Perf loss 0.0078 | Spike loss 0.9906 | Weight loss 0.1010 | Mean activity 0.2498
DMS Iter 1080 | Accuracy 0.9848 | Perf loss 0.0107 | Spike loss 0.9698 | Weight loss 0.1010 | Mean activity 0.2507
DMS Iter 1085 | Accuracy 0.9896 | Perf loss 0.0081 | Spike loss 0.9592 | Weight loss 0.1009 | Mean activity 0.2431
DMS Iter 1090 | Accuracy 0.9887 | Perf loss 0.0074 | Spike loss 0.9576 | Weight loss 0.1008 | Mean activity 0.2395
DMS Iter 1095 | Accuracy 0.9881 | Perf loss 0.0071 | Spike loss 0.9382 | Weight loss 0.1003 | Mean activity 0.2380
DMS Iter 1100 | Accuracy 0.9873 | Perf loss 0.0080 | Spike loss 0.9296 | Weight loss 0.1004 | Mean activity 0.2378
DMS Iter 1105 | Accuracy 0.9885 | Perf loss 0.0072 | Spike loss 0.9388 | Weight loss 0.1005 | Mean activity 0.2405
DMS Iter 1110 | Accuracy 0.9852 | Perf loss 0.0082 | Spike loss 0.9156 | Weight loss 0.1003 | Mean activity 0.2445
DMS Iter 1115 | Accuracy 0.9887 | Perf loss 0.0070 | Spike loss 0.9402 | Weight loss 0.1007 | Mean activity 0.2457
DMS Iter 1120 | Accuracy 0.9906 | Perf loss 0.0064 | Spike loss 0.9364 | Weight loss 0.1007 | Mean activity 0.2403
DMS Iter 1125 | Accuracy 0.9918 | Perf loss 0.0059 | Spike loss 0.9255 | Weight loss 0.1005 | Mean activity 0.2379
DMS Iter 1130 | Accuracy 0.9871 | Perf loss 0.0074 | Spike loss 0.9086 | Weight loss 0.1004 | Mean activity 0.2361
DMS Iter 1135 | Accuracy 0.9885 | Perf loss 0.0086 | Spike loss 0.9164 | Weight loss 0.1009 | Mean activity 0.2327
DMS Iter 1140 | Accuracy 0.9896 | Perf loss 0.0068 | Spike loss 0.9165 | Weight loss 0.1015 | Mean activity 0.2291
DMS Iter 1145 | Accuracy 0.9908 | Perf loss 0.0064 | Spike loss 0.9209 | Weight loss 0.1016 | Mean activity 0.2241
DMS Iter 1150 | Accuracy 0.9914 | Perf loss 0.0057 | Spike loss 0.9296 | Weight loss 0.1015 | Mean activity 0.2291
DMS Iter 1155 | Accuracy 0.9891 | Perf loss 0.0064 | Spike loss 0.9030 | Weight loss 0.1013 | Mean activity 0.2314
DMS Iter 1160 | Accuracy 0.9885 | Perf loss 0.0067 | Spike loss 0.8997 | Weight loss 0.1013 | Mean activity 0.2353
DMS Iter 1165 | Accuracy 0.9893 | Perf loss 0.0078 | Spike loss 0.8913 | Weight loss 0.1013 | Mean activity 0.2326
DMS Iter 1170 | Accuracy 0.9918 | Perf loss 0.0057 | Spike loss 0.9102 | Weight loss 0.1016 | Mean activity 0.2268
DMS Iter 1175 | Accuracy 0.9889 | Perf loss 0.0059 | Spike loss 0.9160 | Weight loss 0.1019 | Mean activity 0.2223
DMS Iter 1180 | Accuracy 0.9904 | Perf loss 0.0060 | Spike loss 0.8863 | Weight loss 0.1019 | Mean activity 0.2205
DMS Iter 1185 | Accuracy 0.9916 | Perf loss 0.0058 | Spike loss 0.8728 | Weight loss 0.1019 | Mean activity 0.2194
DMS Iter 1190 | Accuracy 0.9889 | Perf loss 0.0065 | Spike loss 0.8755 | Weight loss 0.1022 | Mean activity 0.2228
DMS Iter 1195 | Accuracy 0.9920 | Perf loss 0.0055 | Spike loss 0.8721 | Weight loss 0.1021 | Mean activity 0.2223
DMS Iter 1200 | Accuracy 0.9908 | Perf loss 0.0062 | Spike loss 0.9001 | Weight loss 0.1022 | Mean activity 0.2271
DMS Iter 1205 | Accuracy 0.9908 | Perf loss 0.0062 | Spike loss 0.8858 | Weight loss 0.1022 | Mean activity 0.2241
DMS Iter 1210 | Accuracy 0.9877 | Perf loss 0.0074 | Spike loss 0.8708 | Weight loss 0.1023 | Mean activity 0.2211
DMS Iter 1215 | Accuracy 0.9918 | Perf loss 0.0058 | Spike loss 0.8792 | Weight loss 0.1025 | Mean activity 0.2187
DMS Iter 1220 | Accuracy 0.9898 | Perf loss 0.0062 | Spike loss 0.8800 | Weight loss 0.1025 | Mean activity 0.2163
DMS Iter 1225 | Accuracy 0.9914 | Perf loss 0.0055 | Spike loss 0.8738 | Weight loss 0.1026 | Mean activity 0.2138
DMS Iter 1230 | Accuracy 0.9881 | Perf loss 0.0075 | Spike loss 0.8666 | Weight loss 0.1028 | Mean activity 0.2167
DMS Iter 1235 | Accuracy 0.9906 | Perf loss 0.0069 | Spike loss 0.8475 | Weight loss 0.1029 | Mean activity 0.2159
DMS Iter 1240 | Accuracy 0.9918 | Perf loss 0.0062 | Spike loss 0.8590 | Weight loss 0.1032 | Mean activity 0.2185
DMS Iter 1245 | Accuracy 0.9906 | Perf loss 0.0059 | Spike loss 0.8780 | Weight loss 0.1035 | Mean activity 0.2150
DMS Iter 1250 | Accuracy 0.9914 | Perf loss 0.0056 | Spike loss 0.8884 | Weight loss 0.1035 | Mean activity 0.2235
DMS Iter 1255 | Accuracy 0.9902 | Perf loss 0.0065 | Spike loss 0.9063 | Weight loss 0.1035 | Mean activity 0.2232
DMS Iter 1260 | Accuracy 0.9883 | Perf loss 0.0088 | Spike loss 0.8891 | Weight loss 0.1036 | Mean activity 0.2282
DMS Iter 1265 | Accuracy 0.9922 | Perf loss 0.0059 | Spike loss 0.9288 | Weight loss 0.1041 | Mean activity 0.2217
DMS Iter 1270 | Accuracy 0.9922 | Perf loss 0.0056 | Spike loss 0.9277 | Weight loss 0.1044 | Mean activity 0.2131
DMS Iter 1275 | Accuracy 0.9898 | Perf loss 0.0065 | Spike loss 0.8973 | Weight loss 0.1042 | Mean activity 0.2098
DMS Iter 1280 | Accuracy 0.9908 | Perf loss 0.0065 | Spike loss 0.9115 | Weight loss 0.1045 | Mean activity 0.2178
DMS Iter 1285 | Accuracy 0.9936 | Perf loss 0.0046 | Spike loss 0.8839 | Weight loss 0.1044 | Mean activity 0.2173
DMS Iter 1290 | Accuracy 0.9871 | Perf loss 0.0079 | Spike loss 0.9266 | Weight loss 0.1053 | Mean activity 0.2178
DMS Iter 1295 | Accuracy 0.9836 | Perf loss 0.0099 | Spike loss 0.9714 | Weight loss 0.1061 | Mean activity 0.2144
DMS Iter 1300 | Accuracy 0.9850 | Perf loss 0.0093 | Spike loss 0.9930 | Weight loss 0.1068 | Mean activity 0.2098
DMS Iter 1305 | Accuracy 0.9904 | Perf loss 0.0070 | Spike loss 1.0005 | Weight loss 0.1070 | Mean activity 0.2029
DMS Iter 1310 | Accuracy 0.9895 | Perf loss 0.0075 | Spike loss 0.9642 | Weight loss 0.1065 | Mean activity 0.2018
DMS Iter 1315 | Accuracy 0.9896 | Perf loss 0.0064 | Spike loss 0.9606 | Weight loss 0.1063 | Mean activity 0.2036
DMS Iter 1320 | Accuracy 0.9867 | Perf loss 0.0080 | Spike loss 0.9600 | Weight loss 0.1064 | Mean activity 0.2105
DMS Iter 1325 | Accuracy 0.9916 | Perf loss 0.0054 | Spike loss 0.9356 | Weight loss 0.1062 | Mean activity 0.2139
DMS Iter 1330 | Accuracy 0.9898 | Perf loss 0.0067 | Spike loss 0.9386 | Weight loss 0.1064 | Mean activity 0.2110
DMS Iter 1335 | Accuracy 0.9912 | Perf loss 0.0065 | Spike loss 0.9217 | Weight loss 0.1068 | Mean activity 0.2010
DMS Iter 1340 | Accuracy 0.9906 | Perf loss 0.0090 | Spike loss 0.9437 | Weight loss 0.1070 | Mean activity 0.1987
DMS Iter 1345 | Accuracy 0.9877 | Perf loss 0.0099 | Spike loss 0.9245 | Weight loss 0.1068 | Mean activity 0.2053
DMS Iter 1350 | Accuracy 0.9854 | Perf loss 0.0095 | Spike loss 0.9272 | Weight loss 0.1070 | Mean activity 0.1989
DMS Iter 1355 | Accuracy 0.9934 | Perf loss 0.0050 | Spike loss 0.9560 | Weight loss 0.1076 | Mean activity 0.1901
DMS Iter 1360 | Accuracy 0.9895 | Perf loss 0.0065 | Spike loss 0.9569 | Weight loss 0.1081 | Mean activity 0.1889
DMS Iter 1365 | Accuracy 0.9883 | Perf loss 0.0076 | Spike loss 0.9145 | Weight loss 0.1086 | Mean activity 0.1965
DMS Iter 1370 | Accuracy 0.9908 | Perf loss 0.0054 | Spike loss 1.0013 | Weight loss 0.1098 | Mean activity 0.1908
DMS Iter 1375 | Accuracy 0.9932 | Perf loss 0.0045 | Spike loss 1.0142 | Weight loss 0.1096 | Mean activity 0.1965
DMS Iter 1380 | Accuracy 0.9938 | Perf loss 0.0044 | Spike loss 0.9836 | Weight loss 0.1088 | Mean activity 0.2099
DMS Iter 1385 | Accuracy 0.9934 | Perf loss 0.0063 | Spike loss 0.9681 | Weight loss 0.1084 | Mean activity 0.2165
DMS Iter 1390 | Accuracy 0.9945 | Perf loss 0.0044 | Spike loss 0.9420 | Weight loss 0.1080 | Mean activity 0.2191
DMS Iter 1395 | Accuracy 0.9904 | Perf loss 0.0057 | Spike loss 0.9378 | Weight loss 0.1081 | Mean activity 0.2172
DMS Iter 1400 | Accuracy 0.9924 | Perf loss 0.0053 | Spike loss 0.9191 | Weight loss 0.1083 | Mean activity 0.2116
DMS Iter 1405 | Accuracy 0.9938 | Perf loss 0.0043 | Spike loss 0.8989 | Weight loss 0.1084 | Mean activity 0.2085
DMS Iter 1410 | Accuracy 0.9938 | Perf loss 0.0046 | Spike loss 0.8878 | Weight loss 0.1085 | Mean activity 0.2112
DMS Iter 1415 | Accuracy 0.9914 | Perf loss 0.0049 | Spike loss 0.8759 | Weight loss 0.1084 | Mean activity 0.2122
DMS Iter 1420 | Accuracy 0.9928 | Perf loss 0.0047 | Spike loss 0.8669 | Weight loss 0.1083 | Mean activity 0.2064
DMS Iter 1425 | Accuracy 0.9910 | Perf loss 0.0061 | Spike loss 0.8610 | Weight loss 0.1083 | Mean activity 0.2010
DMS Iter 1430 | Accuracy 0.9941 | Perf loss 0.0043 | Spike loss 0.8739 | Weight loss 0.1085 | Mean activity 0.1978
DMS Iter 1435 | Accuracy 0.9900 | Perf loss 0.0060 | Spike loss 0.8548 | Weight loss 0.1084 | Mean activity 0.1994
DMS Iter 1440 | Accuracy 0.9938 | Perf loss 0.0049 | Spike loss 0.8715 | Weight loss 0.1088 | Mean activity 0.2017
DMS Iter 1445 | Accuracy 0.9932 | Perf loss 0.0074 | Spike loss 0.8556 | Weight loss 0.1088 | Mean activity 0.2016
DMS Iter 1450 | Accuracy 0.9916 | Perf loss 0.0046 | Spike loss 0.8578 | Weight loss 0.1091 | Mean activity 0.2130
DMS Iter 1455 | Accuracy 0.9949 | Perf loss 0.0045 | Spike loss 0.8400 | Weight loss 0.1090 | Mean activity 0.2020
DMS Iter 1460 | Accuracy 0.9910 | Perf loss 0.0054 | Spike loss 0.8529 | Weight loss 0.1097 | Mean activity 0.1940
DMS Iter 1465 | Accuracy 0.9938 | Perf loss 0.0042 | Spike loss 0.8708 | Weight loss 0.1104 | Mean activity 0.1943
DMS Iter 1470 | Accuracy 0.9936 | Perf loss 0.0048 | Spike loss 0.8869 | Weight loss 0.1111 | Mean activity 0.1944
DMS Iter 1475 | Accuracy 0.9902 | Perf loss 0.0065 | Spike loss 0.8790 | Weight loss 0.1111 | Mean activity 0.1957
DMS Iter 1480 | Accuracy 0.9932 | Perf loss 0.0042 | Spike loss 0.8940 | Weight loss 0.1116 | Mean activity 0.1974
DMS Iter 1485 | Accuracy 0.9939 | Perf loss 0.0045 | Spike loss 0.8810 | Weight loss 0.1117 | Mean activity 0.1948
DMS Iter 1490 | Accuracy 0.9947 | Perf loss 0.0042 | Spike loss 0.8667 | Weight loss 0.1114 | Mean activity 0.1968
DMS Iter 1495 | Accuracy 0.9930 | Perf loss 0.0046 | Spike loss 0.8704 | Weight loss 0.1114 | Mean activity 0.1973
DMS Iter 1500 | Accuracy 0.9936 | Perf loss 0.0045 | Spike loss 0.8542 | Weight loss 0.1115 | Mean activity 0.1979
DMS Iter 1505 | Accuracy 0.9926 | Perf loss 0.0049 | Spike loss 0.8395 | Weight loss 0.1112 | Mean activity 0.1927
DMS Iter 1510 | Accuracy 0.9902 | Perf loss 0.0055 | Spike loss 0.8411 | Weight loss 0.1111 | Mean activity 0.1968
DMS Iter 1515 | Accuracy 0.9955 | Perf loss 0.0042 | Spike loss 0.8502 | Weight loss 0.1115 | Mean activity 0.1946
DMS Iter 1520 | Accuracy 0.9947 | Perf loss 0.0037 | Spike loss 0.8713 | Weight loss 0.1120 | Mean activity 0.1923
DMS Iter 1525 | Accuracy 0.9938 | Perf loss 0.0041 | Spike loss 0.8592 | Weight loss 0.1121 | Mean activity 0.1904
DMS Iter 1530 | Accuracy 0.9891 | Perf loss 0.0059 | Spike loss 0.8584 | Weight loss 0.1120 | Mean activity 0.1975
DMS Iter 1535 | Accuracy 0.9969 | Perf loss 0.0030 | Spike loss 0.8562 | Weight loss 0.1122 | Mean activity 0.2045
DMS Iter 1540 | Accuracy 0.9934 | Perf loss 0.0045 | Spike loss 0.8505 | Weight loss 0.1125 | Mean activity 0.2083
DMS Iter 1545 | Accuracy 0.9951 | Perf loss 0.0037 | Spike loss 0.8539 | Weight loss 0.1131 | Mean activity 0.1993
DMS Iter 1550 | Accuracy 0.9840 | Perf loss 0.0105 | Spike loss 0.8402 | Weight loss 0.1132 | Mean activity 0.1983
DMS Iter 1555 | Accuracy 0.9947 | Perf loss 0.0045 | Spike loss 0.8586 | Weight loss 0.1134 | Mean activity 0.1870
DMS Iter 1560 | Accuracy 0.9928 | Perf loss 0.0059 | Spike loss 0.8491 | Weight loss 0.1131 | Mean activity 0.1896
DMS Iter 1565 | Accuracy 0.9945 | Perf loss 0.0044 | Spike loss 0.8768 | Weight loss 0.1135 | Mean activity 0.1866
DMS Iter 1570 | Accuracy 0.9961 | Perf loss 0.0040 | Spike loss 0.8761 | Weight loss 0.1134 | Mean activity 0.1890
DMS Iter 1575 | Accuracy 0.9920 | Perf loss 0.0050 | Spike loss 0.8622 | Weight loss 0.1136 | Mean activity 0.1876
DMS Iter 1580 | Accuracy 0.9947 | Perf loss 0.0040 | Spike loss 0.8654 | Weight loss 0.1137 | Mean activity 0.1895
DMS Iter 1585 | Accuracy 0.9916 | Perf loss 0.0053 | Spike loss 0.8556 | Weight loss 0.1135 | Mean activity 0.1887
DMS Iter 1590 | Accuracy 0.9934 | Perf loss 0.0052 | Spike loss 0.8511 | Weight loss 0.1139 | Mean activity 0.1921
DMS Iter 1595 | Accuracy 0.9924 | Perf loss 0.0052 | Spike loss 0.8525 | Weight loss 0.1139 | Mean activity 0.1903
DMS Iter 1600 | Accuracy 0.9930 | Perf loss 0.0048 | Spike loss 0.8391 | Weight loss 0.1138 | Mean activity 0.1895
DMS Iter 1605 | Accuracy 0.9936 | Perf loss 0.0040 | Spike loss 0.8435 | Weight loss 0.1139 | Mean activity 0.1859
DMS Iter 1610 | Accuracy 0.9953 | Perf loss 0.0038 | Spike loss 0.8391 | Weight loss 0.1139 | Mean activity 0.1853
DMS Iter 1615 | Accuracy 0.9951 | Perf loss 0.0035 | Spike loss 0.8159 | Weight loss 0.1134 | Mean activity 0.1875
DMS Iter 1620 | Accuracy 0.9928 | Perf loss 0.0050 | Spike loss 0.8111 | Weight loss 0.1132 | Mean activity 0.1818
DMS Iter 1625 | Accuracy 0.9945 | Perf loss 0.0063 | Spike loss 0.8065 | Weight loss 0.1131 | Mean activity 0.1789
DMS Iter 1630 | Accuracy 0.9949 | Perf loss 0.0039 | Spike loss 0.7948 | Weight loss 0.1133 | Mean activity 0.1852
DMS Iter 1635 | Accuracy 0.9943 | Perf loss 0.0037 | Spike loss 0.8035 | Weight loss 0.1136 | Mean activity 0.1839
DMS Iter 1640 | Accuracy 0.9945 | Perf loss 0.0043 | Spike loss 0.7917 | Weight loss 0.1140 | Mean activity 0.1823
DMS Iter 1645 | Accuracy 0.9951 | Perf loss 0.0042 | Spike loss 0.7960 | Weight loss 0.1148 | Mean activity 0.1879
DMS Iter 1650 | Accuracy 0.9943 | Perf loss 0.0041 | Spike loss 0.8118 | Weight loss 0.1152 | Mean activity 0.1861
DMS Iter 1655 | Accuracy 0.9961 | Perf loss 0.0033 | Spike loss 0.8231 | Weight loss 0.1154 | Mean activity 0.1872
DMS Iter 1660 | Accuracy 0.9922 | Perf loss 0.0049 | Spike loss 0.8300 | Weight loss 0.1156 | Mean activity 0.1857
DMS Iter 1665 | Accuracy 0.9928 | Perf loss 0.0051 | Spike loss 0.7958 | Weight loss 0.1151 | Mean activity 0.1819
DMS Iter 1670 | Accuracy 0.9945 | Perf loss 0.0039 | Spike loss 0.8031 | Weight loss 0.1152 | Mean activity 0.1879
DMS Iter 1675 | Accuracy 0.9943 | Perf loss 0.0040 | Spike loss 0.8033 | Weight loss 0.1154 | Mean activity 0.1932
DMS Iter 1680 | Accuracy 0.9949 | Perf loss 0.0036 | Spike loss 0.8184 | Weight loss 0.1154 | Mean activity 0.1991
DMS Iter 1685 | Accuracy 0.9938 | Perf loss 0.0046 | Spike loss 0.8355 | Weight loss 0.1154 | Mean activity 0.1971
DMS Iter 1690 | Accuracy 0.9959 | Perf loss 0.0031 | Spike loss 0.8267 | Weight loss 0.1156 | Mean activity 0.1946
DMS Iter 1695 | Accuracy 0.9971 | Perf loss 0.0031 | Spike loss 0.8272 | Weight loss 0.1160 | Mean activity 0.1895
DMS Iter 1700 | Accuracy 0.9947 | Perf loss 0.0044 | Spike loss 0.7990 | Weight loss 0.1160 | Mean activity 0.1873
DMS Iter 1705 | Accuracy 0.9949 | Perf loss 0.0049 | Spike loss 0.8047 | Weight loss 0.1163 | Mean activity 0.1845
DMS Iter 1710 | Accuracy 0.9914 | Perf loss 0.0059 | Spike loss 0.8207 | Weight loss 0.1168 | Mean activity 0.1894
DMS Iter 1715 | Accuracy 0.9934 | Perf loss 0.0055 | Spike loss 0.8967 | Weight loss 0.1183 | Mean activity 0.1889
DMS Iter 1720 | Accuracy 0.9938 | Perf loss 0.0042 | Spike loss 0.9271 | Weight loss 0.1190 | Mean activity 0.1797
DMS Iter 1725 | Accuracy 0.9941 | Perf loss 0.0043 | Spike loss 0.8997 | Weight loss 0.1188 | Mean activity 0.1849
DMS Iter 1730 | Accuracy 0.9957 | Perf loss 0.0041 | Spike loss 0.9010 | Weight loss 0.1187 | Mean activity 0.1859
DMS Iter 1735 | Accuracy 0.9943 | Perf loss 0.0039 | Spike loss 0.8702 | Weight loss 0.1183 | Mean activity 0.1818
DMS Iter 1740 | Accuracy 0.9947 | Perf loss 0.0038 | Spike loss 0.8373 | Weight loss 0.1179 | Mean activity 0.1787
DMS Iter 1745 | Accuracy 0.9941 | Perf loss 0.0044 | Spike loss 0.8255 | Weight loss 0.1177 | Mean activity 0.1740
DMS Iter 1750 | Accuracy 0.9941 | Perf loss 0.0040 | Spike loss 0.8066 | Weight loss 0.1174 | Mean activity 0.1753
DMS Iter 1755 | Accuracy 0.9938 | Perf loss 0.0043 | Spike loss 0.8010 | Weight loss 0.1173 | Mean activity 0.1759
DMS Iter 1760 | Accuracy 0.9965 | Perf loss 0.0031 | Spike loss 0.8057 | Weight loss 0.1174 | Mean activity 0.1760
DMS Iter 1765 | Accuracy 0.9949 | Perf loss 0.0037 | Spike loss 0.8215 | Weight loss 0.1175 | Mean activity 0.1807
DMS Iter 1770 | Accuracy 0.9947 | Perf loss 0.0038 | Spike loss 0.8170 | Weight loss 0.1174 | Mean activity 0.1763
DMS Iter 1775 | Accuracy 0.9922 | Perf loss 0.0070 | Spike loss 0.8104 | Weight loss 0.1178 | Mean activity 0.1820
DMS Iter 1780 | Accuracy 0.9877 | Perf loss 0.0089 | Spike loss 0.8513 | Weight loss 0.1187 | Mean activity 0.1823
DMS Iter 1785 | Accuracy 0.9936 | Perf loss 0.0050 | Spike loss 0.9072 | Weight loss 0.1201 | Mean activity 0.1844
DMS Iter 1790 | Accuracy 0.9945 | Perf loss 0.0040 | Spike loss 0.9034 | Weight loss 0.1210 | Mean activity 0.1862
DMS Iter 1795 | Accuracy 0.9928 | Perf loss 0.0052 | Spike loss 0.9049 | Weight loss 0.1215 | Mean activity 0.1787
DMS Iter 1800 | Accuracy 0.9900 | Perf loss 0.0071 | Spike loss 0.8681 | Weight loss 0.1212 | Mean activity 0.1876
DMS Iter 1805 | Accuracy 0.9936 | Perf loss 0.0051 | Spike loss 0.9025 | Weight loss 0.1215 | Mean activity 0.1839
DMS Iter 1810 | Accuracy 0.9914 | Perf loss 0.0062 | Spike loss 0.8534 | Weight loss 0.1211 | Mean activity 0.1962
DMS Iter 1815 | Accuracy 0.9938 | Perf loss 0.0056 | Spike loss 0.8340 | Weight loss 0.1211 | Mean activity 0.1940
DMS Iter 1820 | Accuracy 0.9922 | Perf loss 0.0054 | Spike loss 0.8712 | Weight loss 0.1226 | Mean activity 0.1901
DMS Iter 1825 | Accuracy 0.9949 | Perf loss 0.0042 | Spike loss 0.9605 | Weight loss 0.1246 | Mean activity 0.1883
DMS Iter 1830 | Accuracy 0.9951 | Perf loss 0.0041 | Spike loss 1.0019 | Weight loss 0.1256 | Mean activity 0.1950
DMS Iter 1835 | Accuracy 0.9959 | Perf loss 0.0027 | Spike loss 0.9699 | Weight loss 0.1257 | Mean activity 0.1963
DMS Iter 1840 | Accuracy 0.9902 | Perf loss 0.0067 | Spike loss 0.9282 | Weight loss 0.1255 | Mean activity 0.1898
DMS Iter 1845 | Accuracy 0.9865 | Perf loss 0.0109 | Spike loss 0.9986 | Weight loss 0.1266 | Mean activity 0.1946
DMS Iter 1850 | Accuracy 0.9916 | Perf loss 0.0062 | Spike loss 0.9811 | Weight loss 0.1262 | Mean activity 0.1922
DMS Iter 1855 | Accuracy 0.9934 | Perf loss 0.0049 | Spike loss 1.0203 | Weight loss 0.1258 | Mean activity 0.1922
DMS Iter 1860 | Accuracy 0.9924 | Perf loss 0.0052 | Spike loss 0.9674 | Weight loss 0.1245 | Mean activity 0.1937
DMS Iter 1865 | Accuracy 0.9945 | Perf loss 0.0040 | Spike loss 0.9687 | Weight loss 0.1242 | Mean activity 0.1910
DMS Iter 1870 | Accuracy 0.9916 | Perf loss 0.0062 | Spike loss 0.9064 | Weight loss 0.1235 | Mean activity 0.1899
DMS Iter 1875 | Accuracy 0.9934 | Perf loss 0.0043 | Spike loss 0.9104 | Weight loss 0.1237 | Mean activity 0.1890
DMS Iter 1880 | Accuracy 0.9949 | Perf loss 0.0038 | Spike loss 0.8810 | Weight loss 0.1235 | Mean activity 0.1868
DMS Iter 1885 | Accuracy 0.9957 | Perf loss 0.0034 | Spike loss 0.8970 | Weight loss 0.1235 | Mean activity 0.1915
DMS Iter 1890 | Accuracy 0.9955 | Perf loss 0.0035 | Spike loss 0.8865 | Weight loss 0.1231 | Mean activity 0.2022
DMS Iter 1895 | Accuracy 0.9961 | Perf loss 0.0028 | Spike loss 0.8642 | Weight loss 0.1228 | Mean activity 0.1918
DMS Iter 1900 | Accuracy 0.9961 | Perf loss 0.0038 | Spike loss 0.8516 | Weight loss 0.1226 | Mean activity 0.1910
DMS Iter 1905 | Accuracy 0.9959 | Perf loss 0.0031 | Spike loss 0.8262 | Weight loss 0.1225 | Mean activity 0.1843
DMS Iter 1910 | Accuracy 0.9965 | Perf loss 0.0029 | Spike loss 0.8059 | Weight loss 0.1223 | Mean activity 0.1810
DMS Iter 1915 | Accuracy 0.9922 | Perf loss 0.0060 | Spike loss 0.7895 | Weight loss 0.1221 | Mean activity 0.1806
DMS Iter 1920 | Accuracy 0.9930 | Perf loss 0.0057 | Spike loss 0.7994 | Weight loss 0.1227 | Mean activity 0.1768
DMS Iter 1925 | Accuracy 0.9943 | Perf loss 0.0050 | Spike loss 0.8451 | Weight loss 0.1243 | Mean activity 0.1780
DMS Iter 1930 | Accuracy 0.9910 | Perf loss 0.0062 | Spike loss 0.8501 | Weight loss 0.1252 | Mean activity 0.1829
DMS Iter 1935 | Accuracy 0.9951 | Perf loss 0.0036 | Spike loss 0.8845 | Weight loss 0.1262 | Mean activity 0.1847
DMS Iter 1940 | Accuracy 0.9934 | Perf loss 0.0049 | Spike loss 0.8389 | Weight loss 0.1260 | Mean activity 0.1858
DMS Iter 1945 | Accuracy 0.9941 | Perf loss 0.0045 | Spike loss 0.8717 | Weight loss 0.1266 | Mean activity 0.1840
DMS Iter 1950 | Accuracy 0.9949 | Perf loss 0.0041 | Spike loss 0.8284 | Weight loss 0.1265 | Mean activity 0.1850
DMS Iter 1955 | Accuracy 0.9953 | Perf loss 0.0037 | Spike loss 0.8459 | Weight loss 0.1268 | Mean activity 0.1735
DMS Iter 1960 | Accuracy 0.9918 | Perf loss 0.0058 | Spike loss 0.8244 | Weight loss 0.1265 | Mean activity 0.1813
DMS Iter 1965 | Accuracy 0.9922 | Perf loss 0.0075 | Spike loss 0.8114 | Weight loss 0.1264 | Mean activity 0.1778
DMS Iter 1970 | Accuracy 0.9914 | Perf loss 0.0068 | Spike loss 0.8257 | Weight loss 0.1273 | Mean activity 0.1749
DMS Iter 1975 | Accuracy 0.9965 | Perf loss 0.0031 | Spike loss 0.8691 | Weight loss 0.1282 | Mean activity 0.1744
DMS Iter 1980 | Accuracy 0.9967 | Perf loss 0.0034 | Spike loss 0.8379 | Weight loss 0.1278 | Mean activity 0.1774
DMS Iter 1985 | Accuracy 0.9941 | Perf loss 0.0044 | Spike loss 0.8190 | Weight loss 0.1273 | Mean activity 0.1703
DMS Iter 1990 | Accuracy 0.9953 | Perf loss 0.0031 | Spike loss 0.8146 | Weight loss 0.1273 | Mean activity 0.1712
DMS Iter 1995 | Accuracy 0.9959 | Perf loss 0.0034 | Spike loss 0.8150 | Weight loss 0.1271 | Mean activity 0.1716