In [1]:
import autoreg
import GPy
import numpy as np
from matplotlib import pyplot as plt
from __future__ import print_function
%matplotlib inline
from autoreg.benchmark import tasks
In [2]:
# Function to compute root mean square error:
def comp_RMSE(a,b):
return np.sqrt(np.square(a-b).mean())
In [3]:
# Define class for normalization
class Normalize(object):
def __init__(self, data, name, norm_name):
self.data_mean = data.mean(axis=0)
self.data_std = data.std(axis=0)
self.normalization_computed = True
setattr(self, name, data)
setattr(self, norm_name, (data-self.data_mean) / self.data_std )
def normalize(self, data, name, norm_name):
if hasattr(self,norm_name):
raise ValueError("This normalization name already exist, choose another one")
setattr(self, name, data )
setattr(self, norm_name, (data-self.data_mean) / self.data_std )
def denormalize(self, data):
return data*self.data_std + self.data_mean
In [4]:
trainned_models_folder_name = "/Users/grigoral/work/code/RGP/examples/identif_trainded"
task_name = 'IdentificationExample5'
# task names:
# Actuator, Ballbeam, Drive, Gas_furnace, Flutter, Dryer, Tank,
# IdentificationExample1..5
In [5]:
task = getattr( tasks, task_name)
task = task()
task.load_data()
print("Data OUT train shape: ", task.data_out_train.shape)
print("Data IN train shape: ", task.data_in_train.shape)
print("Data OUT test shape: ", task.data_out_test.shape)
print("Data IN test shape: ", task.data_in_test.shape)
Data OUT train shape: (300, 1)
Data IN train shape: (300, 1)
Data OUT test shape: (300, 1)
Data IN test shape: (300, 1)
In [6]:
normalize = False
in_data = Normalize(task.data_in_train,'in_train','in_train_norm' )
out_data = Normalize(task.data_out_train,'out_train','out_train_norm' )
in_data.normalize(task.data_in_test, 'in_test','in_test_norm')
out_data.normalize(task.data_out_test, 'out_test','out_test_norm')
if normalize:
out_train = out_data.out_train_norm #out_data.out_train
in_train = in_data.in_train_norm # in_data.in_train
out_test = out_data.out_test_norm #out_data.out_test
in_test = in_data.in_test_norm #in_data.in_test
else:
out_train = out_data.out_train #out_data.out_train
in_train = in_data.in_train # in_data.in_train
out_test = out_data.out_test #out_data.out_test
in_test = in_data.in_test #in_data.in_test
print("Training OUT mean: ", out_train.mean(0));
print("Training OUT std: ", out_train.std(0))
print("")
print("Test OUT mean: ", out_test.mean(0));
print("Test OUT std: ", out_test.std(0))
print("")
print("Training IN mean: ", in_train.mean(0));
print("Training IN std: ", in_train.std(0))
print("")
print("Test IN mean: ", in_test.mean(0));
print("Test IN std: ", in_test.std(0))
Training OUT mean: [ 0.05968174]
Training OUT std: [ 1.87727311]
Test OUT mean: [ 0.23458633]
Test OUT std: [ 1.70671659]
Training IN mean: [ 0.16261821]
Training IN std: [ 1.48605093]
Test IN mean: [ 9.17784367e-17]
Test IN std: [ 1.]
In [7]:
# Plot training:
fig1 = plt.figure(1,figsize=(20,8))
fig1.suptitle('Training data')
ax1 = plt.subplot(1,2,1)
ax1.plot(out_train)
ax1.set_title('Data OUT training')
ax2 = plt.subplot(1,2,2)
ax2.plot(in_train)
ax2.set_title('Data IN training')
fig2 = plt.figure(2,figsize=(20,8))
fig2.suptitle('Test data')
ax3 = plt.subplot(1,2,1)
ax3.plot(out_test)
ax3.set_title('Data OUT test')
ax4 = plt.subplot(1,2,2)
ax4.plot(in_test)
ax4.set_title('Data IN test')
del ax1, ax2, ax3, ax4
In [49]:
Q = 50 # 200 # Inducing points num
win_in = task.win_in # 20
win_out = task.win_out # 20
use_controls = True
back_cstr = False
inference_method = 'svi'
# 1 layer:
wins = [0, win_out] # 0-th is output layer
nDims = [out_train.shape[1],1]
# 2 layers:
# wins = [0, win_out, win_out]
# nDims = [out_train.shape[1],1,1]
MLP_dims = [300,200]
print("Input window: ", win_in)
print("Output window: ", win_out)
m = autoreg.DeepAutoreg_new(wins, out_train, U=in_train, U_win=win_in,
num_inducing=Q, back_cstr=back_cstr, MLP_dims=MLP_dims, nDims=nDims,
init='Y', # how to initialize hidden states means
X_variance=0.05, # how to initialize hidden states variances
inference_method=inference_method, # Inference method
# 1 layer:
kernels=[GPy.kern.RBF(win_out,ARD=True,inv_l=True),
GPy.kern.RBF(win_in + win_out,ARD=True,inv_l=True)] )
# 2 layers:
#kernels=[GPy.kern.RBF(win_out,ARD=True,inv_l=True),
# GPy.kern.RBF(win_out+win_out,ARD=True,inv_l=True),
# GPy.kern.RBF(win_out+win_in,ARD=True,inv_l=True)])
#m = autoreg.DeepAutoreg([0,win_out],out_train, U=in_train, U_win=win_in,X_variance=0.01,
# num_inducing=50)
# pattern for model name: #task_name, inf_meth=?, wins=layers, Q = ?, backcstr=?,MLP_dims=?, nDims=
model_file_name = '%s--inf_meth=%s--wins=%s--Q=%i--backcstr=%i--nDims=%s' % (task.name,
'reg' if inference_method is None else inference_method, str(wins), Q, back_cstr, str(nDims))
if back_cstr == True:
model_file_name += '--MLP_dims=%s' % (MLP_dims,)
print('Model file name: ', model_file_name)
print(m)
m.checkgrad(verbose=True)
Input window: 5
Output window: 5
Model file name: IdentificationExample5--inf_meth=svi--wins=[0, 5]--Q=3--backcstr=0--nDims=[1, 1]
Name : autoreg
Objective : 68283.6643485
Number of Parameters : 690
Number of Optimization Parameters : 690
Updates : True
Parameters:
autoreg. | value | constraints | priors
layer_1.inducing_inputs | (3, 10) | |
layer_1.rbf.variance | 1.0 | +ve |
layer_1.rbf.inv_lengthscale | (10,) | +ve |
layer_1.Gaussian_noise.variance | 0.01 | +ve |
layer_1.qU_m | (3, 1) | |
layer_1.qU_W | (3, 3) | |
layer_1.qU_a | 0.001 | +ve |
layer_1.qX_0.mean | (300, 1) | |
layer_1.qX_0.variance | (300, 1) | +ve |
layer_0.inducing_inputs | (3, 5) | |
layer_0.rbf.variance | 1.0 | +ve |
layer_0.rbf.inv_lengthscale | (5,) | +ve |
layer_0.Gaussian_noise.variance | 1.0 | +ve |
layer_0.qU_m | (3, 1) | |
layer_0.qU_W | (3, 3) | |
layer_0.qU_a | 0.001 | +ve |
Name | Ratio | Difference | Analytical | Numerical | dF_ratio
-----------------------------------------------------------------------------------------------------------------------------
autoreg.layer_1.inducing_inputs[[0 0]] | 0.999998 | 0.000003 | -1.346142 | -1.346139 | 4e-11
autoreg.layer_1.inducing_inputs[[0 1]] | 0.999999 | 0.000002 | 3.704345 | 3.704343 | 1e-10
autoreg.layer_1.inducing_inputs[[0 2]] | 1.000000 | 0.000000 | 0.634544 | 0.634544 | 2e-11
autoreg.layer_1.inducing_inputs[[0 3]] | 0.999999 | 0.000002 | 1.930219 | 1.930217 | 6e-11
autoreg.layer_1.inducing_inputs[[0 4]] | 0.999999 | 0.000005 | -7.912492 | -7.912487 | 2e-10
autoreg.layer_1.inducing_inputs[[0 5]] | 1.000002 | 0.000004 | 2.209479 | 2.209483 | 6e-11
autoreg.layer_1.inducing_inputs[[0 6]] | 0.999999 | 0.000005 | -4.889020 | -4.889014 | 1e-10
autoreg.layer_1.inducing_inputs[[0 7]] | 0.999997 | 0.000005 | 1.654711 | 1.654706 | 5e-11
autoreg.layer_1.inducing_inputs[[0 8]] | 0.999999 | 0.000002 | -4.098848 | -4.098845 | 1e-10
autoreg.layer_1.inducing_inputs[[0 9]] | 1.000000 | 0.000002 | -10.573926 | -10.573924 | 3e-10
autoreg.layer_1.inducing_inputs[[1 0]] | 1.000001 | 0.000006 | -10.878424 | -10.878430 | 3e-10
autoreg.layer_1.inducing_inputs[[1 1]] | 1.000001 | 0.000007 | -9.384894 | -9.384901 | 3e-10
autoreg.layer_1.inducing_inputs[[1 2]] | 0.999999 | 0.000004 | -6.970983 | -6.970979 | 2e-10
autoreg.layer_1.inducing_inputs[[1 3]] | 0.999997 | 0.000006 | 1.922947 | 1.922941 | 6e-11
autoreg.layer_1.inducing_inputs[[1 4]] | 1.000001 | 0.000002 | 2.013707 | 2.013709 | 6e-11
autoreg.layer_1.inducing_inputs[[1 5]] | 1.000000 | 0.000005 | 21.581060 | 21.581065 | 6e-10
autoreg.layer_1.inducing_inputs[[1 6]] | 0.999999 | 0.000004 | -4.200196 | -4.200192 | 1e-10
autoreg.layer_1.inducing_inputs[[1 7]] | 1.000000 | 0.000003 | -12.270476 | -12.270473 | 4e-10
autoreg.layer_1.inducing_inputs[[1 8]] | 1.000000 | 0.000002 | -8.393641 | -8.393639 | 2e-10
autoreg.layer_1.inducing_inputs[[1 9]] | 1.000000 | 0.000001 | -4.440735 | -4.440735 | 1e-10
autoreg.layer_1.inducing_inputs[[2 0]] | 0.999997 | 0.000009 | 2.562405 | 2.562396 | 8e-11
autoreg.layer_1.inducing_inputs[[2 1]] | 0.999999 | 0.000002 | -3.361960 | -3.361958 | 1e-10
autoreg.layer_1.inducing_inputs[[2 2]] | 1.000001 | 0.000003 | -4.007747 | -4.007750 | 1e-10
autoreg.layer_1.inducing_inputs[[2 3]] | 0.999998 | 0.000006 | -2.920183 | -2.920176 | 9e-11
autoreg.layer_1.inducing_inputs[[2 4]] | 0.999999 | 0.000006 | -5.247209 | -5.247202 | 2e-10
autoreg.layer_1.inducing_inputs[[2 5]] | 1.000002 | 0.000004 | -1.952637 | -1.952641 | 6e-11
autoreg.layer_1.inducing_inputs[[2 6]] | 1.000003 | 0.000005 | 1.564312 | 1.564316 | 5e-11
autoreg.layer_1.inducing_inputs[[2 7]] | 0.999998 | 0.000006 | -2.266517 | -2.266512 | 7e-11
autoreg.layer_1.inducing_inputs[[2 8]] | 1.000002 | 0.000009 | -4.030958 | -4.030968 | 1e-10
autoreg.layer_1.inducing_inputs[[2 9]] | 1.000001 | 0.000002 | -2.874241 | -2.874243 | 8e-11
autoreg.layer_1.rbf.variance | 1.000000 | 0.000007 | 9322.525722 | 9322.525715 | 3e-07
autoreg.layer_1.rbf.inv_lengthscale[[0]] | 0.999999 | 0.000006 | 6.791581 | 6.791575 | 2e-10
autoreg.layer_1.rbf.inv_lengthscale[[1]] | 1.000000 | 0.000001 | 8.761753 | 8.761752 | 3e-10
autoreg.layer_1.rbf.inv_lengthscale[[2]] | 1.000002 | 0.000008 | 4.270346 | 4.270354 | 1e-10
autoreg.layer_1.rbf.inv_lengthscale[[3]] | 1.000000 | 0.000000 | 4.599118 | 4.599118 | 1e-10
autoreg.layer_1.rbf.inv_lengthscale[[4]] | 1.000001 | 0.000004 | 3.296944 | 3.296947 | 1e-10
autoreg.layer_1.rbf.inv_lengthscale[[5]] | 1.000000 | 0.000003 | 13.694549 | 13.694553 | 4e-10
autoreg.layer_1.rbf.inv_lengthscale[[6]] | 1.000000 | 0.000000 | 5.931463 | 5.931463 | 2e-10
autoreg.layer_1.rbf.inv_lengthscale[[7]] | 1.000001 | 0.000010 | 9.279462 | 9.279473 | 3e-10
autoreg.layer_1.rbf.inv_lengthscale[[8]] | 1.000000 | 0.000001 | 6.184572 | 6.184571 | 2e-10
autoreg.layer_1.rbf.inv_lengthscale[[9]] | 1.000000 | 0.000001 | 8.201858 | 8.201860 | 2e-10
autoreg.layer_1.Gaussian_noise.variance | 1.000000 | 0.000009 | -67239.511364 | -67239.511372 | 2e-06
autoreg.layer_1.qU_m[[0 0]] | 1.000000 | 0.000001 | -28.488975 | -28.488976 | 8e-10
autoreg.layer_1.qU_m[[1 0]] | 1.000000 | 0.000001 | 43.766536 | 43.766537 | 1e-09
autoreg.layer_1.qU_m[[2 0]] | 1.000013 | 0.000004 | 0.289397 | 0.289401 | 8e-12
autoreg.layer_1.qU_W[[0 0]] | 1.000001 | 0.000006 | -5.010037 | -5.010043 | 1e-10
autoreg.layer_1.qU_W[[0 1]] | 0.999999 | 0.000002 | -3.092094 | -3.092093 | 9e-11
autoreg.layer_1.qU_W[[0 2]] | 1.000000 | 0.000000 | 11.586715 | 11.586715 | 3e-10
autoreg.layer_1.qU_W[[1 0]] | 0.999999 | 0.000005 | -7.609595 | -7.609589 | 2e-10
autoreg.layer_1.qU_W[[1 1]] | 0.999999 | 0.000005 | -3.408842 | -3.408837 | 1e-10
autoreg.layer_1.qU_W[[1 2]] | 0.999999 | 0.000001 | 1.839007 | 1.839006 | 5e-11
autoreg.layer_1.qU_W[[2 0]] | 1.000002 | 0.000003 | -1.709607 | -1.709610 | 5e-11
autoreg.layer_1.qU_W[[2 1]] | 1.000000 | 0.000000 | -1.607070 | -1.607070 | 5e-11
autoreg.layer_1.qU_W[[2 2]] | 0.999999 | 0.000001 | 1.756737 | 1.756736 | 5e-11
autoreg.layer_1.qU_a | 1.000003 | 0.000004 | -1.317970 | -1.317974 | 4e-11
autoreg.layer_1.qX_0.mean[[0 0]] | 1.000024 | 0.000002 | 0.088037 | 0.088039 | 3e-12
autoreg.layer_1.qX_0.mean[[1 0]] | 0.999992 | 0.000003 | 0.393218 | 0.393215 | 1e-11
autoreg.layer_1.qX_0.mean[[2 0]] | 0.999998 | 0.000001 | 0.647969 | 0.647968 | 2e-11
autoreg.layer_1.qX_0.mean[[3 0]] | 1.000008 | 0.000002 | 0.241865 | 0.241867 | 7e-12
autoreg.layer_1.qX_0.mean[[4 0]] | 1.000027 | 0.000007 | 0.259053 | 0.259060 | 8e-12
autoreg.layer_1.qX_0.mean[[5 0]] | 1.000000 | 0.000015 | -129.529108 | -129.529093 | 4e-09
autoreg.layer_1.qX_0.mean[[6 0]] | 1.000000 | 0.000007 | 65.375694 | 65.375687 | 2e-09
autoreg.layer_1.qX_0.mean[[7 0]] | 1.000000 | 0.000000 | -35.315280 | -35.315279 | 1e-09
autoreg.layer_1.qX_0.mean[[8 0]] | 1.000000 | 0.000005 | -138.437036 | -138.437041 | 4e-09
autoreg.layer_1.qX_0.mean[[9 0]] | 1.000000 | 0.000004 | -87.060111 | -87.060107 | 3e-09
autoreg.layer_1.qX_0.mean[[10 0]] | 1.000000 | 0.000003 | 135.077684 | 135.077687 | 4e-09
autoreg.layer_1.qX_0.mean[[11 0]] | 1.000000 | 0.000001 | -15.712591 | -15.712591 | 5e-10
autoreg.layer_1.qX_0.mean[[12 0]] | 1.000000 | 0.000011 | -166.242763 | -166.242775 | 5e-09
autoreg.layer_1.qX_0.mean[[13 0]] | 1.000000 | 0.000001 | -72.014655 | -72.014656 | 2e-09
autoreg.layer_1.qX_0.mean[[14 0]] | 1.000000 | 0.000004 | -417.327461 | -417.327457 | 1e-08
autoreg.layer_1.qX_0.mean[[15 0]] | 1.000000 | 0.000006 | 164.517782 | 164.517776 | 5e-09
autoreg.layer_1.qX_0.mean[[16 0]] | 1.000000 | 0.000006 | 74.792041 | 74.792035 | 2e-09
autoreg.layer_1.qX_0.mean[[17 0]] | 1.000000 | 0.000006 | -59.039093 | -59.039099 | 2e-09
autoreg.layer_1.qX_0.mean[[18 0]] | 1.000000 | 0.000001 | 35.325632 | 35.325633 | 1e-09
autoreg.layer_1.qX_0.mean[[19 0]] | 1.000000 | 0.000004 | -230.055379 | -230.055382 | 7e-09
autoreg.layer_1.qX_0.mean[[20 0]] | 1.000000 | 0.000003 | 130.899785 | 130.899782 | 4e-09
autoreg.layer_1.qX_0.mean[[21 0]] | 1.000000 | 0.000002 | 91.514228 | 91.514230 | 3e-09
autoreg.layer_1.qX_0.mean[[22 0]] | 0.999999 | 0.000005 | 6.446205 | 6.446200 | 2e-10
autoreg.layer_1.qX_0.mean[[23 0]] | 1.000000 | 0.000012 | 218.350159 | 218.350171 | 6e-09
autoreg.layer_1.qX_0.mean[[24 0]] | 1.000000 | 0.000012 | 52.769084 | 52.769072 | 2e-09
autoreg.layer_1.qX_0.mean[[25 0]] | 1.000000 | 0.000005 | 143.104710 | 143.104706 | 4e-09
autoreg.layer_1.qX_0.mean[[26 0]] | 1.000000 | 0.000010 | 199.361060 | 199.361049 | 6e-09
autoreg.layer_1.qX_0.mean[[27 0]] | 1.000000 | 0.000002 | 191.678735 | 191.678737 | 6e-09
autoreg.layer_1.qX_0.mean[[28 0]] | 1.000000 | 0.000002 | 146.565833 | 146.565835 | 4e-09
autoreg.layer_1.qX_0.mean[[29 0]] | 0.999999 | 0.000018 | -29.041813 | -29.041796 | 9e-10
autoreg.layer_1.qX_0.mean[[30 0]] | 1.000000 | 0.000001 | 21.369755 | 21.369757 | 6e-10
autoreg.layer_1.qX_0.mean[[31 0]] | 1.000000 | 0.000002 | 145.960648 | 145.960650 | 4e-09
autoreg.layer_1.qX_0.mean[[32 0]] | 1.000000 | 0.000012 | 136.889622 | 136.889634 | 4e-09
autoreg.layer_1.qX_0.mean[[33 0]] | 1.000000 | 0.000007 | -121.507073 | -121.507066 | 4e-09
autoreg.layer_1.qX_0.mean[[34 0]] | 1.000000 | 0.000001 | -51.500071 | -51.500072 | 2e-09
autoreg.layer_1.qX_0.mean[[35 0]] | 1.000000 | 0.000001 | -202.916386 | -202.916388 | 6e-09
autoreg.layer_1.qX_0.mean[[36 0]] | 1.000000 | 0.000013 | -187.102023 | -187.102036 | 5e-09
autoreg.layer_1.qX_0.mean[[37 0]] | 1.000000 | 0.000002 | 90.473191 | 90.473193 | 3e-09
autoreg.layer_1.qX_0.mean[[38 0]] | 1.000000 | 0.000005 | 145.562573 | 145.562568 | 4e-09
autoreg.layer_1.qX_0.mean[[39 0]] | 1.000000 | 0.000007 | 130.422719 | 130.422726 | 4e-09
autoreg.layer_1.qX_0.mean[[40 0]] | 1.000000 | 0.000011 | 105.442338 | 105.442326 | 3e-09
autoreg.layer_1.qX_0.mean[[41 0]] | 1.000000 | 0.000001 | 153.578198 | 153.578199 | 4e-09
autoreg.layer_1.qX_0.mean[[42 0]] | 1.000000 | 0.000008 | 130.089619 | 130.089611 | 4e-09
autoreg.layer_1.qX_0.mean[[43 0]] | 1.000000 | 0.000002 | 266.761361 | 266.761359 | 8e-09
autoreg.layer_1.qX_0.mean[[44 0]] | 1.000000 | 0.000003 | 51.091378 | 51.091381 | 1e-09
autoreg.layer_1.qX_0.mean[[45 0]] | 1.000000 | 0.000009 | -267.779788 | -267.779797 | 8e-09
autoreg.layer_1.qX_0.mean[[46 0]] | 1.000000 | 0.000001 | -257.747009 | -257.747008 | 8e-09
autoreg.layer_1.qX_0.mean[[47 0]] | 1.000000 | 0.000002 | -384.313260 | -384.313258 | 1e-08
autoreg.layer_1.qX_0.mean[[48 0]] | 1.000000 | 0.000002 | -417.606037 | -417.606039 | 1e-08
autoreg.layer_1.qX_0.mean[[49 0]] | 1.000000 | 0.000001 | -111.828792 | -111.828791 | 3e-09
autoreg.layer_1.qX_0.mean[[50 0]] | 1.000000 | 0.000006 | -168.905920 | -168.905914 | 5e-09
autoreg.layer_1.qX_0.mean[[51 0]] | 1.000000 | 0.000000 | 84.714556 | 84.714557 | 2e-09
autoreg.layer_1.qX_0.mean[[52 0]] | 1.000000 | 0.000009 | 142.541530 | 142.541539 | 4e-09
autoreg.layer_1.qX_0.mean[[53 0]] | 1.000000 | 0.000002 | 173.527135 | 173.527136 | 5e-09
autoreg.layer_1.qX_0.mean[[54 0]] | 1.000000 | 0.000005 | 109.080555 | 109.080560 | 3e-09
autoreg.layer_1.qX_0.mean[[55 0]] | 1.000000 | 0.000006 | 165.067117 | 165.067111 | 5e-09
autoreg.layer_1.qX_0.mean[[56 0]] | 1.000000 | 0.000001 | 93.367597 | 93.367598 | 3e-09
autoreg.layer_1.qX_0.mean[[57 0]] | 1.000000 | 0.000002 | 235.667094 | 235.667096 | 7e-09
autoreg.layer_1.qX_0.mean[[58 0]] | 1.000000 | 0.000008 | -50.848247 | -50.848255 | 1e-09
autoreg.layer_1.qX_0.mean[[59 0]] | 1.000000 | 0.000004 | 342.857977 | 342.857973 | 1e-08
autoreg.layer_1.qX_0.mean[[60 0]] | 1.000000 | 0.000014 | -361.907033 | -361.907019 | 1e-08
autoreg.layer_1.qX_0.mean[[61 0]] | 1.000000 | 0.000003 | 317.041037 | 317.041035 | 9e-09
autoreg.layer_1.qX_0.mean[[62 0]] | 1.000000 | 0.000008 | -501.611875 | -501.611867 | 1e-08
autoreg.layer_1.qX_0.mean[[63 0]] | 1.000000 | 0.000007 | 561.960340 | 561.960347 | 2e-08
autoreg.layer_1.qX_0.mean[[64 0]] | 1.000000 | 0.000007 | 290.159376 | 290.159369 | 8e-09
autoreg.layer_1.qX_0.mean[[65 0]] | 1.000000 | 0.000004 | 76.158755 | 76.158751 | 2e-09
autoreg.layer_1.qX_0.mean[[66 0]] | 1.000000 | 0.000010 | 60.196023 | 60.196013 | 2e-09
autoreg.layer_1.qX_0.mean[[67 0]] | 1.000000 | 0.000006 | 158.706900 | 158.706905 | 5e-09
autoreg.layer_1.qX_0.mean[[68 0]] | 1.000000 | 0.000006 | -140.090852 | -140.090859 | 4e-09
autoreg.layer_1.qX_0.mean[[69 0]] | 1.000000 | 0.000003 | 330.226826 | 330.226823 | 1e-08
autoreg.layer_1.qX_0.mean[[70 0]] | 1.000000 | 0.000002 | 40.557204 | 40.557206 | 1e-09
autoreg.layer_1.qX_0.mean[[71 0]] | 1.000000 | 0.000009 | -58.904888 | -58.904880 | 2e-09
autoreg.layer_1.qX_0.mean[[72 0]] | 1.000001 | 0.000012 | 17.115733 | 17.115744 | 5e-10
autoreg.layer_1.qX_0.mean[[73 0]] | 0.999998 | 0.000010 | -5.433556 | -5.433547 | 2e-10
autoreg.layer_1.qX_0.mean[[74 0]] | 1.000000 | 0.000005 | 42.499533 | 42.499538 | 1e-09
autoreg.layer_1.qX_0.mean[[75 0]] | 1.000000 | 0.000003 | -241.029328 | -241.029331 | 7e-09
autoreg.layer_1.qX_0.mean[[76 0]] | 1.000000 | 0.000003 | 38.490031 | 38.490034 | 1e-09
autoreg.layer_1.qX_0.mean[[77 0]] | 1.000000 | 0.000003 | 167.359173 | 167.359176 | 5e-09
autoreg.layer_1.qX_0.mean[[78 0]] | 1.000000 | 0.000002 | 183.070408 | 183.070406 | 5e-09
autoreg.layer_1.qX_0.mean[[79 0]] | 1.000000 | 0.000004 | 148.842405 | 148.842410 | 4e-09
autoreg.layer_1.qX_0.mean[[80 0]] | 1.000000 | 0.000009 | 161.752878 | 161.752869 | 5e-09
autoreg.layer_1.qX_0.mean[[81 0]] | 1.000000 | 0.000005 | 217.310297 | 217.310291 | 6e-09
autoreg.layer_1.qX_0.mean[[82 0]] | 1.000000 | 0.000002 | 8.064841 | 8.064839 | 2e-10
autoreg.layer_1.qX_0.mean[[83 0]] | 1.000000 | 0.000001 | -587.786236 | -587.786235 | 2e-08
autoreg.layer_1.qX_0.mean[[84 0]] | 1.000000 | 0.000006 | -273.065482 | -273.065489 | 8e-09
autoreg.layer_1.qX_0.mean[[85 0]] | 1.000000 | 0.000005 | -357.208252 | -357.208257 | 1e-08
autoreg.layer_1.qX_0.mean[[86 0]] | 1.000000 | 0.000001 | -218.342940 | -218.342939 | 6e-09
autoreg.layer_1.qX_0.mean[[87 0]] | 1.000000 | 0.000006 | -37.343713 | -37.343707 | 1e-09
autoreg.layer_1.qX_0.mean[[88 0]] | 1.000000 | 0.000003 | 8.740852 | 8.740855 | 3e-10
autoreg.layer_1.qX_0.mean[[89 0]] | 1.000000 | 0.000009 | 77.198018 | 77.198027 | 2e-09
autoreg.layer_1.qX_0.mean[[90 0]] | 1.000000 | 0.000010 | -35.885784 | -35.885794 | 1e-09
autoreg.layer_1.qX_0.mean[[91 0]] | 1.000000 | 0.000001 | 248.586263 | 248.586264 | 7e-09
autoreg.layer_1.qX_0.mean[[92 0]] | 1.000001 | 0.000018 | 29.079977 | 29.079994 | 9e-10
autoreg.layer_1.qX_0.mean[[93 0]] | 1.000000 | 0.000016 | -250.340452 | -250.340468 | 7e-09
autoreg.layer_1.qX_0.mean[[94 0]] | 1.000000 | 0.000009 | -163.094299 | -163.094308 | 5e-09
autoreg.layer_1.qX_0.mean[[95 0]] | 1.000000 | 0.000001 | -69.877257 | -69.877256 | 2e-09
autoreg.layer_1.qX_0.mean[[96 0]] | 1.000000 | 0.000006 | -55.184143 | -55.184137 | 2e-09
autoreg.layer_1.qX_0.mean[[97 0]] | 1.000000 | 0.000012 | -99.181062 | -99.181074 | 3e-09
autoreg.layer_1.qX_0.mean[[98 0]] | 1.000000 | 0.000001 | -12.899399 | -12.899400 | 4e-10
autoreg.layer_1.qX_0.mean[[99 0]] | 1.000000 | 0.000001 | -169.588696 | -169.588697 | 5e-09
autoreg.layer_1.qX_0.mean[[100 0]] | 1.000000 | 0.000006 | -75.923491 | -75.923497 | 2e-09
autoreg.layer_1.qX_0.mean[[101 0]] | 1.000000 | 0.000015 | -52.720905 | -52.720919 | 2e-09
autoreg.layer_1.qX_0.mean[[102 0]] | 1.000000 | 0.000003 | -151.934214 | -151.934211 | 4e-09
autoreg.layer_1.qX_0.mean[[103 0]] | 1.000000 | 0.000003 | 34.128814 | 34.128818 | 1e-09
autoreg.layer_1.qX_0.mean[[104 0]] | 1.000000 | 0.000011 | 82.313203 | 82.313192 | 2e-09
autoreg.layer_1.qX_0.mean[[105 0]] | 1.000000 | 0.000000 | 150.795910 | 150.795910 | 4e-09
autoreg.layer_1.qX_0.mean[[106 0]] | 1.000000 | 0.000011 | 284.645041 | 284.645030 | 8e-09
autoreg.layer_1.qX_0.mean[[107 0]] | 1.000000 | 0.000006 | -73.708471 | -73.708477 | 2e-09
autoreg.layer_1.qX_0.mean[[108 0]] | 1.000000 | 0.000007 | 339.311955 | 339.311962 | 1e-08
autoreg.layer_1.qX_0.mean[[109 0]] | 1.000000 | 0.000003 | -249.012377 | -249.012381 | 7e-09
autoreg.layer_1.qX_0.mean[[110 0]] | 1.000000 | 0.000003 | 222.698651 | 222.698654 | 7e-09
autoreg.layer_1.qX_0.mean[[111 0]] | 1.000000 | 0.000004 | 95.387205 | 95.387200 | 3e-09
autoreg.layer_1.qX_0.mean[[112 0]] | 1.000000 | 0.000001 | -89.871929 | -89.871930 | 3e-09
autoreg.layer_1.qX_0.mean[[113 0]] | 1.000000 | 0.000002 | -23.787599 | -23.787601 | 7e-10
autoreg.layer_1.qX_0.mean[[114 0]] | 1.000000 | 0.000002 | 17.046803 | 17.046805 | 5e-10
autoreg.layer_1.qX_0.mean[[115 0]] | 1.000000 | 0.000010 | 47.137747 | 47.137757 | 1e-09
autoreg.layer_1.qX_0.mean[[116 0]] | 1.000000 | 0.000001 | 182.492956 | 182.492957 | 5e-09
autoreg.layer_1.qX_0.mean[[117 0]] | 1.000000 | 0.000007 | 243.345029 | 243.345035 | 7e-09
autoreg.layer_1.qX_0.mean[[118 0]] | 1.000000 | 0.000010 | 96.953726 | 96.953736 | 3e-09
autoreg.layer_1.qX_0.mean[[119 0]] | 1.000000 | 0.000009 | -158.213921 | -158.213930 | 5e-09
autoreg.layer_1.qX_0.mean[[120 0]] | 1.000000 | 0.000003 | -303.217112 | -303.217115 | 9e-09
autoreg.layer_1.qX_0.mean[[121 0]] | 1.000000 | 0.000004 | -247.029110 | -247.029115 | 7e-09
autoreg.layer_1.qX_0.mean[[122 0]] | 1.000000 | 0.000002 | -187.369109 | -187.369107 | 5e-09
autoreg.layer_1.qX_0.mean[[123 0]] | 1.000000 | 0.000002 | -458.153592 | -458.153590 | 1e-08
autoreg.layer_1.qX_0.mean[[124 0]] | 1.000000 | 0.000002 | 314.974639 | 314.974641 | 9e-09
autoreg.layer_1.qX_0.mean[[125 0]] | 1.000000 | 0.000004 | -123.803290 | -123.803286 | 4e-09
autoreg.layer_1.qX_0.mean[[126 0]] | 1.000000 | 0.000007 | 341.128841 | 341.128834 | 1e-08
autoreg.layer_1.qX_0.mean[[127 0]] | 0.999999 | 0.000003 | -2.307799 | -2.307796 | 7e-11
autoreg.layer_1.qX_0.mean[[128 0]] | 1.000000 | 0.000015 | -250.492456 | -250.492471 | 7e-09
autoreg.layer_1.qX_0.mean[[129 0]] | 1.000000 | 0.000006 | -294.028804 | -294.028810 | 9e-09
autoreg.layer_1.qX_0.mean[[130 0]] | 1.000000 | 0.000003 | -10.104542 | -10.104544 | 3e-10
autoreg.layer_1.qX_0.mean[[131 0]] | 1.000000 | 0.000003 | -96.535489 | -96.535492 | 3e-09
autoreg.layer_1.qX_0.mean[[132 0]] | 1.000000 | 0.000012 | 101.801673 | 101.801685 | 3e-09
autoreg.layer_1.qX_0.mean[[133 0]] | 1.000000 | 0.000003 | 134.505314 | 134.505317 | 4e-09
autoreg.layer_1.qX_0.mean[[134 0]] | 1.000000 | 0.000002 | 161.791873 | 161.791875 | 5e-09
autoreg.layer_1.qX_0.mean[[135 0]] | 1.000000 | 0.000003 | 333.433530 | 333.433527 | 1e-08
autoreg.layer_1.qX_0.mean[[136 0]] | 1.000000 | 0.000013 | -30.076642 | -30.076655 | 9e-10
autoreg.layer_1.qX_0.mean[[137 0]] | 1.000000 | 0.000003 | -98.768734 | -98.768731 | 3e-09
autoreg.layer_1.qX_0.mean[[138 0]] | 1.000000 | 0.000002 | -134.321594 | -134.321592 | 4e-09
autoreg.layer_1.qX_0.mean[[139 0]] | 1.000000 | 0.000008 | -363.434222 | -363.434214 | 1e-08
autoreg.layer_1.qX_0.mean[[140 0]] | 1.000000 | 0.000001 | -69.220348 | -69.220347 | 2e-09
autoreg.layer_1.qX_0.mean[[141 0]] | 1.000000 | 0.000007 | -80.477126 | -80.477119 | 2e-09
autoreg.layer_1.qX_0.mean[[142 0]] | 1.000000 | 0.000001 | -67.190499 | -67.190500 | 2e-09
autoreg.layer_1.qX_0.mean[[143 0]] | 1.000000 | 0.000015 | -140.829557 | -140.829572 | 4e-09
autoreg.layer_1.qX_0.mean[[144 0]] | 1.000000 | 0.000009 | -150.631951 | -150.631960 | 4e-09
autoreg.layer_1.qX_0.mean[[145 0]] | 1.000000 | 0.000002 | -234.338693 | -234.338695 | 7e-09
autoreg.layer_1.qX_0.mean[[146 0]] | 1.000000 | 0.000007 | 354.379386 | 354.379379 | 1e-08
autoreg.layer_1.qX_0.mean[[147 0]] | 1.000000 | 0.000001 | -165.523371 | -165.523372 | 5e-09
autoreg.layer_1.qX_0.mean[[148 0]] | 1.000000 | 0.000001 | -217.658474 | -217.658475 | 6e-09
autoreg.layer_1.qX_0.mean[[149 0]] | 1.000000 | 0.000007 | 50.900264 | 50.900271 | 1e-09
autoreg.layer_1.qX_0.mean[[150 0]] | 1.000000 | 0.000003 | 54.283740 | 54.283737 | 2e-09
autoreg.layer_1.qX_0.mean[[151 0]] | 1.000000 | 0.000007 | 146.975114 | 146.975108 | 4e-09
autoreg.layer_1.qX_0.mean[[152 0]] | 1.000000 | 0.000009 | 97.235918 | 97.235927 | 3e-09
autoreg.layer_1.qX_0.mean[[153 0]] | 1.000000 | 0.000007 | 245.703370 | 245.703377 | 7e-09
autoreg.layer_1.qX_0.mean[[154 0]] | 1.000000 | 0.000002 | 121.383937 | 121.383935 | 4e-09
autoreg.layer_1.qX_0.mean[[155 0]] | 1.000000 | 0.000012 | -272.852599 | -272.852587 | 8e-09
autoreg.layer_1.qX_0.mean[[156 0]] | 1.000000 | 0.000007 | -47.135859 | -47.135865 | 1e-09
autoreg.layer_1.qX_0.mean[[157 0]] | 1.000000 | 0.000002 | -69.174208 | -69.174210 | 2e-09
autoreg.layer_1.qX_0.mean[[158 0]] | 1.000000 | 0.000001 | -118.854294 | -118.854296 | 3e-09
autoreg.layer_1.qX_0.mean[[159 0]] | 1.000000 | 0.000001 | 77.056852 | 77.056851 | 2e-09
autoreg.layer_1.qX_0.mean[[160 0]] | 1.000002 | 0.000004 | -1.742428 | -1.742432 | 5e-11
autoreg.layer_1.qX_0.mean[[161 0]] | 1.000000 | 0.000008 | 88.315188 | 88.315181 | 3e-09
autoreg.layer_1.qX_0.mean[[162 0]] | 1.000000 | 0.000004 | 153.496647 | 153.496643 | 4e-09
autoreg.layer_1.qX_0.mean[[163 0]] | 1.000000 | 0.000002 | 309.617995 | 309.617993 | 9e-09
autoreg.layer_1.qX_0.mean[[164 0]] | 1.000000 | 0.000003 | 141.137224 | 141.137221 | 4e-09
autoreg.layer_1.qX_0.mean[[165 0]] | 1.000000 | 0.000010 | -41.005633 | -41.005624 | 1e-09
autoreg.layer_1.qX_0.mean[[166 0]] | 1.000000 | 0.000006 | -44.602274 | -44.602268 | 1e-09
autoreg.layer_1.qX_0.mean[[167 0]] | 1.000000 | 0.000000 | 144.217272 | 144.217272 | 4e-09
autoreg.layer_1.qX_0.mean[[168 0]] | 1.000000 | 0.000007 | 64.028704 | 64.028711 | 2e-09
autoreg.layer_1.qX_0.mean[[169 0]] | 1.000000 | 0.000001 | 226.156290 | 226.156291 | 7e-09
autoreg.layer_1.qX_0.mean[[170 0]] | 1.000000 | 0.000001 | 16.160701 | 16.160702 | 5e-10
autoreg.layer_1.qX_0.mean[[171 0]] | 1.000000 | 0.000005 | 283.632838 | 283.632842 | 8e-09
autoreg.layer_1.qX_0.mean[[172 0]] | 1.000000 | 0.000004 | -53.290463 | -53.290467 | 2e-09
autoreg.layer_1.qX_0.mean[[173 0]] | 1.000000 | 0.000002 | 336.372798 | 336.372796 | 1e-08
autoreg.layer_1.qX_0.mean[[174 0]] | 1.000000 | 0.000008 | -180.809941 | -180.809933 | 5e-09
autoreg.layer_1.qX_0.mean[[175 0]] | 1.000000 | 0.000001 | 193.797126 | 193.797125 | 6e-09
autoreg.layer_1.qX_0.mean[[176 0]] | 1.000000 | 0.000009 | 76.049639 | 76.049648 | 2e-09
autoreg.layer_1.qX_0.mean[[177 0]] | 1.000000 | 0.000002 | 130.202055 | 130.202054 | 4e-09
autoreg.layer_1.qX_0.mean[[178 0]] | 1.000000 | 0.000003 | 231.086228 | 231.086226 | 7e-09
autoreg.layer_1.qX_0.mean[[179 0]] | 1.000000 | 0.000002 | 126.135975 | 126.135972 | 4e-09
autoreg.layer_1.qX_0.mean[[180 0]] | 1.000000 | 0.000003 | 233.810181 | 233.810184 | 7e-09
autoreg.layer_1.qX_0.mean[[181 0]] | 1.000000 | 0.000001 | 25.247349 | 25.247347 | 7e-10
autoreg.layer_1.qX_0.mean[[182 0]] | 1.000000 | 0.000002 | 45.242940 | 45.242938 | 1e-09
autoreg.layer_1.qX_0.mean[[183 0]] | 1.000000 | 0.000010 | -26.347904 | -26.347894 | 8e-10
autoreg.layer_1.qX_0.mean[[184 0]] | 1.000000 | 0.000011 | -270.558816 | -270.558827 | 8e-09
autoreg.layer_1.qX_0.mean[[185 0]] | 1.000000 | 0.000012 | 53.707374 | 53.707387 | 2e-09
autoreg.layer_1.qX_0.mean[[186 0]] | 1.000000 | 0.000012 | 77.695896 | 77.695884 | 2e-09
autoreg.layer_1.qX_0.mean[[187 0]] | 1.000000 | 0.000001 | 182.594332 | 182.594333 | 5e-09
autoreg.layer_1.qX_0.mean[[188 0]] | 1.000000 | 0.000006 | 337.549144 | 337.549151 | 1e-08
autoreg.layer_1.qX_0.mean[[189 0]] | 1.000000 | 0.000003 | -484.499765 | -484.499767 | 1e-08
autoreg.layer_1.qX_0.mean[[190 0]] | 1.000000 | 0.000008 | 142.420206 | 142.420198 | 4e-09
autoreg.layer_1.qX_0.mean[[191 0]] | 1.000000 | 0.000008 | 124.764315 | 124.764323 | 4e-09
autoreg.layer_1.qX_0.mean[[192 0]] | 1.000000 | 0.000006 | 210.199743 | 210.199738 | 6e-09
autoreg.layer_1.qX_0.mean[[193 0]] | 1.000000 | 0.000001 | 169.376798 | 169.376799 | 5e-09
autoreg.layer_1.qX_0.mean[[194 0]] | 1.000000 | 0.000010 | 27.085317 | 27.085327 | 8e-10
autoreg.layer_1.qX_0.mean[[195 0]] | 1.000000 | 0.000000 | 193.625295 | 193.625296 | 6e-09
autoreg.layer_1.qX_0.mean[[196 0]] | 1.000000 | 0.000003 | -111.754776 | -111.754773 | 3e-09
autoreg.layer_1.qX_0.mean[[197 0]] | 1.000000 | 0.000006 | 292.908792 | 292.908786 | 9e-09
autoreg.layer_1.qX_0.mean[[198 0]] | 1.000000 | 0.000008 | -101.807651 | -101.807644 | 3e-09
autoreg.layer_1.qX_0.mean[[199 0]] | 1.000000 | 0.000001 | -328.423535 | -328.423535 | 1e-08
autoreg.layer_1.qX_0.mean[[200 0]] | 1.000000 | 0.000000 | -250.348319 | -250.348319 | 7e-09
autoreg.layer_1.qX_0.mean[[201 0]] | 1.000000 | 0.000010 | -127.027251 | -127.027241 | 4e-09
autoreg.layer_1.qX_0.mean[[202 0]] | 1.000000 | 0.000000 | 0.674947 | 0.674947 | 2e-11
autoreg.layer_1.qX_0.mean[[203 0]] | 1.000000 | 0.000003 | -41.971044 | -41.971041 | 1e-09
autoreg.layer_1.qX_0.mean[[204 0]] | 1.000000 | 0.000002 | -122.479510 | -122.479512 | 4e-09
autoreg.layer_1.qX_0.mean[[205 0]] | 1.000000 | 0.000004 | -110.577825 | -110.577821 | 3e-09
autoreg.layer_1.qX_0.mean[[206 0]] | 1.000000 | 0.000002 | -33.126436 | -33.126438 | 1e-09
autoreg.layer_1.qX_0.mean[[207 0]] | 1.000000 | 0.000001 | -182.806731 | -182.806732 | 5e-09
autoreg.layer_1.qX_0.mean[[208 0]] | 1.000000 | 0.000008 | -297.509680 | -297.509672 | 9e-09
autoreg.layer_1.qX_0.mean[[209 0]] | 1.000000 | 0.000008 | -267.010125 | -267.010117 | 8e-09
autoreg.layer_1.qX_0.mean[[210 0]] | 1.000000 | 0.000003 | -141.402797 | -141.402794 | 4e-09
autoreg.layer_1.qX_0.mean[[211 0]] | 1.000000 | 0.000000 | -250.359379 | -250.359379 | 7e-09
autoreg.layer_1.qX_0.mean[[212 0]] | 1.000000 | 0.000009 | -54.288206 | -54.288197 | 2e-09
autoreg.layer_1.qX_0.mean[[213 0]] | 1.000000 | 0.000006 | -75.869656 | -75.869662 | 2e-09
autoreg.layer_1.qX_0.mean[[214 0]] | 1.000000 | 0.000004 | -303.144607 | -303.144610 | 9e-09
autoreg.layer_1.qX_0.mean[[215 0]] | 1.000000 | 0.000006 | 89.498491 | 89.498484 | 3e-09
autoreg.layer_1.qX_0.mean[[216 0]] | 1.000000 | 0.000007 | 59.711987 | 59.711994 | 2e-09
autoreg.layer_1.qX_0.mean[[217 0]] | 1.000000 | 0.000004 | 87.596647 | 87.596643 | 3e-09
autoreg.layer_1.qX_0.mean[[218 0]] | 1.000000 | 0.000003 | 77.003304 | 77.003300 | 2e-09
autoreg.layer_1.qX_0.mean[[219 0]] | 1.000000 | 0.000011 | 88.220750 | 88.220761 | 3e-09
autoreg.layer_1.qX_0.mean[[220 0]] | 1.000000 | 0.000004 | 87.390031 | 87.390035 | 3e-09
autoreg.layer_1.qX_0.mean[[221 0]] | 1.000000 | 0.000002 | -75.690840 | -75.690841 | 2e-09
autoreg.layer_1.qX_0.mean[[222 0]] | 1.000000 | 0.000004 | -374.946551 | -374.946554 | 1e-08
autoreg.layer_1.qX_0.mean[[223 0]] | 1.000000 | 0.000003 | -116.907494 | -116.907497 | 3e-09
autoreg.layer_1.qX_0.mean[[224 0]] | 1.000000 | 0.000009 | -144.569617 | -144.569625 | 4e-09
autoreg.layer_1.qX_0.mean[[225 0]] | 1.000000 | 0.000001 | 23.265079 | 23.265078 | 7e-10
autoreg.layer_1.qX_0.mean[[226 0]] | 1.000000 | 0.000015 | 105.884392 | 105.884406 | 3e-09
autoreg.layer_1.qX_0.mean[[227 0]] | 1.000000 | 0.000005 | 172.167840 | 172.167835 | 5e-09
autoreg.layer_1.qX_0.mean[[228 0]] | 1.000000 | 0.000012 | 156.686289 | 156.686277 | 5e-09
autoreg.layer_1.qX_0.mean[[229 0]] | 1.000000 | 0.000013 | 35.539959 | 35.539946 | 1e-09
autoreg.layer_1.qX_0.mean[[230 0]] | 1.000000 | 0.000012 | -45.252671 | -45.252658 | 1e-09
autoreg.layer_1.qX_0.mean[[231 0]] | 1.000000 | 0.000015 | -227.756690 | -227.756675 | 7e-09
autoreg.layer_1.qX_0.mean[[232 0]] | 1.000000 | 0.000000 | -178.264192 | -178.264192 | 5e-09
autoreg.layer_1.qX_0.mean[[233 0]] | 1.000000 | 0.000010 | -218.123659 | -218.123649 | 6e-09
autoreg.layer_1.qX_0.mean[[234 0]] | 1.000000 | 0.000005 | -114.953208 | -114.953204 | 3e-09
autoreg.layer_1.qX_0.mean[[235 0]] | 1.000000 | 0.000005 | -84.099751 | -84.099745 | 2e-09
autoreg.layer_1.qX_0.mean[[236 0]] | 1.000000 | 0.000006 | -67.743894 | -67.743887 | 2e-09
autoreg.layer_1.qX_0.mean[[237 0]] | 1.000000 | 0.000008 | 209.925107 | 209.925100 | 6e-09
autoreg.layer_1.qX_0.mean[[238 0]] | 1.000000 | 0.000008 | 122.492325 | 122.492333 | 4e-09
autoreg.layer_1.qX_0.mean[[239 0]] | 1.000000 | 0.000005 | 134.597374 | 134.597380 | 4e-09
autoreg.layer_1.qX_0.mean[[240 0]] | 1.000000 | 0.000003 | 168.751680 | 168.751678 | 5e-09
autoreg.layer_1.qX_0.mean[[241 0]] | 1.000000 | 0.000009 | 312.737628 | 312.737619 | 9e-09
autoreg.layer_1.qX_0.mean[[242 0]] | 1.000000 | 0.000010 | 64.213088 | 64.213098 | 2e-09
autoreg.layer_1.qX_0.mean[[243 0]] | 1.000000 | 0.000003 | 198.198384 | 198.198381 | 6e-09
autoreg.layer_1.qX_0.mean[[244 0]] | 1.000000 | 0.000001 | 89.373577 | 89.373578 | 3e-09
autoreg.layer_1.qX_0.mean[[245 0]] | 1.000000 | 0.000004 | 99.532273 | 99.532277 | 3e-09
autoreg.layer_1.qX_0.mean[[246 0]] | 1.000000 | 0.000002 | 145.332170 | 145.332167 | 4e-09
autoreg.layer_1.qX_0.mean[[247 0]] | 1.000000 | 0.000004 | 330.898820 | 330.898816 | 1e-08
autoreg.layer_1.qX_0.mean[[248 0]] | 1.000000 | 0.000006 | -243.753943 | -243.753937 | 7e-09
autoreg.layer_1.qX_0.mean[[249 0]] | 1.000000 | 0.000002 | 195.453785 | 195.453787 | 6e-09
autoreg.layer_1.qX_0.mean[[250 0]] | 1.000000 | 0.000011 | 56.667366 | 56.667377 | 2e-09
autoreg.layer_1.qX_0.mean[[251 0]] | 1.000000 | 0.000000 | -239.916124 | -239.916124 | 7e-09
autoreg.layer_1.qX_0.mean[[252 0]] | 1.000000 | 0.000016 | 115.046415 | 115.046430 | 3e-09
autoreg.layer_1.qX_0.mean[[253 0]] | 1.000000 | 0.000002 | -77.740015 | -77.740013 | 2e-09
autoreg.layer_1.qX_0.mean[[254 0]] | 1.000000 | 0.000009 | -141.507526 | -141.507517 | 4e-09
autoreg.layer_1.qX_0.mean[[255 0]] | 1.000000 | 0.000002 | 68.436350 | 68.436348 | 2e-09
autoreg.layer_1.qX_0.mean[[256 0]] | 1.000000 | 0.000004 | 154.696976 | 154.696972 | 5e-09
autoreg.layer_1.qX_0.mean[[257 0]] | 1.000000 | 0.000006 | 125.579971 | 125.579965 | 4e-09
autoreg.layer_1.qX_0.mean[[258 0]] | 1.000000 | 0.000006 | -142.255294 | -142.255289 | 4e-09
autoreg.layer_1.qX_0.mean[[259 0]] | 1.000000 | 0.000011 | -254.194023 | -254.194012 | 7e-09
autoreg.layer_1.qX_0.mean[[260 0]] | 1.000000 | 0.000009 | 198.385324 | 198.385314 | 6e-09
autoreg.layer_1.qX_0.mean[[261 0]] | 1.000000 | 0.000013 | 99.278519 | 99.278506 | 3e-09
autoreg.layer_1.qX_0.mean[[262 0]] | 1.000000 | 0.000001 | 262.339889 | 262.339890 | 8e-09
autoreg.layer_1.qX_0.mean[[263 0]] | 1.000000 | 0.000006 | 140.580479 | 140.580472 | 4e-09
autoreg.layer_1.qX_0.mean[[264 0]] | 1.000000 | 0.000002 | 294.106778 | 294.106780 | 9e-09
autoreg.layer_1.qX_0.mean[[265 0]] | 1.000000 | 0.000000 | -271.424811 | -271.424811 | 8e-09
autoreg.layer_1.qX_0.mean[[266 0]] | 1.000000 | 0.000007 | -68.266745 | -68.266752 | 2e-09
autoreg.layer_1.qX_0.mean[[267 0]] | 1.000000 | 0.000007 | -253.461455 | -253.461461 | 7e-09
autoreg.layer_1.qX_0.mean[[268 0]] | 1.000000 | 0.000001 | 230.067154 | 230.067155 | 7e-09
autoreg.layer_1.qX_0.mean[[269 0]] | 1.000000 | 0.000005 | 126.375589 | 126.375584 | 4e-09
autoreg.layer_1.qX_0.mean[[270 0]] | 1.000000 | 0.000004 | 174.239733 | 174.239736 | 5e-09
autoreg.layer_1.qX_0.mean[[271 0]] | 1.000000 | 0.000008 | 169.587868 | 169.587860 | 5e-09
autoreg.layer_1.qX_0.mean[[272 0]] | 1.000000 | 0.000003 | -41.830561 | -41.830564 | 1e-09
autoreg.layer_1.qX_0.mean[[273 0]] | 1.000000 | 0.000007 | -112.744769 | -112.744761 | 3e-09
autoreg.layer_1.qX_0.mean[[274 0]] | 1.000000 | 0.000000 | 363.856001 | 363.856001 | 1e-08
autoreg.layer_1.qX_0.mean[[275 0]] | 1.000000 | 0.000003 | -37.284332 | -37.284335 | 1e-09
autoreg.layer_1.qX_0.mean[[276 0]] | 1.000000 | 0.000002 | -137.678007 | -137.678006 | 4e-09
autoreg.layer_1.qX_0.mean[[277 0]] | 1.000000 | 0.000006 | -172.949191 | -172.949185 | 5e-09
autoreg.layer_1.qX_0.mean[[278 0]] | 1.000000 | 0.000006 | -475.091385 | -475.091379 | 1e-08
autoreg.layer_1.qX_0.mean[[279 0]] | 1.000000 | 0.000001 | 129.369213 | 129.369211 | 4e-09
autoreg.layer_1.qX_0.mean[[280 0]] | 1.000000 | 0.000009 | 131.134586 | 131.134577 | 4e-09
autoreg.layer_1.qX_0.mean[[281 0]] | 1.000000 | 0.000005 | 179.274463 | 179.274459 | 5e-09
autoreg.layer_1.qX_0.mean[[282 0]] | 1.000000 | 0.000003 | 170.017058 | 170.017054 | 5e-09
autoreg.layer_1.qX_0.mean[[283 0]] | 1.000000 | 0.000001 | -103.341271 | -103.341270 | 3e-09
autoreg.layer_1.qX_0.mean[[284 0]] | 1.000000 | 0.000005 | -417.177706 | -417.177711 | 1e-08
autoreg.layer_1.qX_0.mean[[285 0]] | 1.000000 | 0.000004 | -338.302398 | -338.302401 | 1e-08
autoreg.layer_1.qX_0.mean[[286 0]] | 1.000000 | 0.000008 | 153.823413 | 153.823421 | 5e-09
autoreg.layer_1.qX_0.mean[[287 0]] | 0.999999 | 0.000004 | -4.341211 | -4.341207 | 1e-10
autoreg.layer_1.qX_0.mean[[288 0]] | 1.000000 | 0.000006 | 175.525266 | 175.525260 | 5e-09
autoreg.layer_1.qX_0.mean[[289 0]] | 1.000000 | 0.000011 | 86.208132 | 86.208122 | 3e-09
autoreg.layer_1.qX_0.mean[[290 0]] | 1.000000 | 0.000007 | -49.170056 | -49.170063 | 1e-09
autoreg.layer_1.qX_0.mean[[291 0]] | 1.000000 | 0.000002 | 5.709042 | 5.709044 | 2e-10
autoreg.layer_1.qX_0.mean[[292 0]] | 1.000000 | 0.000011 | 93.379200 | 93.379211 | 3e-09
autoreg.layer_1.qX_0.mean[[293 0]] | 1.000000 | 0.000001 | 64.508683 | 64.508684 | 2e-09
autoreg.layer_1.qX_0.mean[[294 0]] | 1.000000 | 0.000002 | -127.571528 | -127.571526 | 4e-09
autoreg.layer_1.qX_0.mean[[295 0]] | 0.999997 | 0.000012 | -4.358448 | -4.358437 | 1e-10
autoreg.layer_1.qX_0.mean[[296 0]] | 1.000000 | 0.000005 | -121.849237 | -121.849233 | 4e-09
autoreg.layer_1.qX_0.mean[[297 0]] | 1.000000 | 0.000001 | -65.953310 | -65.953311 | 2e-09
autoreg.layer_1.qX_0.mean[[298 0]] | 1.000000 | 0.000003 | -39.551571 | -39.551574 | 1e-09
autoreg.layer_1.qX_0.mean[[299 0]] | 1.000000 | 0.000004 | -213.321026 | -213.321022 | 6e-09
autoreg.layer_1.qX_0.variance[[0 0]] | 1.000001 | 0.000000 | -0.476204 | -0.476204 | 1e-11
autoreg.layer_1.qX_0.variance[[1 0]] | 1.000008 | 0.000004 | -0.461583 | -0.461587 | 1e-11
autoreg.layer_1.qX_0.variance[[2 0]] | 0.999981 | 0.000009 | -0.458241 | -0.458233 | 1e-11
autoreg.layer_1.qX_0.variance[[3 0]] | 0.999988 | 0.000006 | -0.460938 | -0.460932 | 1e-11
autoreg.layer_1.qX_0.variance[[4 0]] | 1.000003 | 0.000001 | -0.456907 | -0.456908 | 1e-11
autoreg.layer_1.qX_0.variance[[5 0]] | 0.999993 | 0.000013 | 1.940766 | 1.940753 | 6e-11
autoreg.layer_1.qX_0.variance[[6 0]] | 0.999993 | 0.000013 | 1.936866 | 1.936853 | 6e-11
autoreg.layer_1.qX_0.variance[[7 0]] | 0.999995 | 0.000010 | 1.937794 | 1.937784 | 6e-11
autoreg.layer_1.qX_0.variance[[8 0]] | 0.999998 | 0.000004 | 1.963909 | 1.963905 | 6e-11
autoreg.layer_1.qX_0.variance[[9 0]] | 1.000000 | 0.000000 | 1.950262 | 1.950262 | 6e-11
autoreg.layer_1.qX_0.variance[[10 0]] | 0.999994 | 0.000011 | 1.953613 | 1.953602 | 6e-11
autoreg.layer_1.qX_0.variance[[11 0]] | 0.999995 | 0.000011 | 1.953612 | 1.953602 | 6e-11
autoreg.layer_1.qX_0.variance[[12 0]] | 0.999998 | 0.000005 | 1.955004 | 1.954999 | 6e-11
autoreg.layer_1.qX_0.variance[[13 0]] | 0.999996 | 0.000008 | 1.951783 | 1.951776 | 6e-11
autoreg.layer_1.qX_0.variance[[14 0]] | 0.999997 | 0.000007 | 1.937180 | 1.937173 | 6e-11
autoreg.layer_1.qX_0.variance[[15 0]] | 0.999997 | 0.000006 | 1.940889 | 1.940884 | 6e-11
autoreg.layer_1.qX_0.variance[[16 0]] | 0.999993 | 0.000014 | 1.935623 | 1.935608 | 6e-11
autoreg.layer_1.qX_0.variance[[17 0]] | 0.999993 | 0.000013 | 1.940460 | 1.940447 | 6e-11
autoreg.layer_1.qX_0.variance[[18 0]] | 0.999997 | 0.000006 | 1.939441 | 1.939436 | 6e-11
autoreg.layer_1.qX_0.variance[[19 0]] | 0.999990 | 0.000019 | 1.939578 | 1.939559 | 6e-11
autoreg.layer_1.qX_0.variance[[20 0]] | 0.999996 | 0.000007 | 1.952728 | 1.952722 | 6e-11
autoreg.layer_1.qX_0.variance[[21 0]] | 0.999994 | 0.000011 | 1.982979 | 1.982968 | 6e-11
autoreg.layer_1.qX_0.variance[[22 0]] | 1.000000 | 0.000001 | 1.988986 | 1.988985 | 6e-11
autoreg.layer_1.qX_0.variance[[23 0]] | 0.999992 | 0.000014 | 1.838932 | 1.838918 | 5e-11
autoreg.layer_1.qX_0.variance[[24 0]] | 0.999995 | 0.000010 | 2.042539 | 2.042529 | 6e-11
autoreg.layer_1.qX_0.variance[[25 0]] | 0.999997 | 0.000006 | 2.049891 | 2.049885 | 6e-11
autoreg.layer_1.qX_0.variance[[26 0]] | 0.999995 | 0.000010 | 1.964875 | 1.964865 | 6e-11
autoreg.layer_1.qX_0.variance[[27 0]] | 0.999999 | 0.000001 | 1.963200 | 1.963199 | 6e-11
autoreg.layer_1.qX_0.variance[[28 0]] | 0.999996 | 0.000009 | 1.957649 | 1.957640 | 6e-11
autoreg.layer_1.qX_0.variance[[29 0]] | 0.999997 | 0.000005 | 1.941013 | 1.941007 | 6e-11
autoreg.layer_1.qX_0.variance[[30 0]] | 1.000001 | 0.000002 | 1.956946 | 1.956949 | 6e-11
autoreg.layer_1.qX_0.variance[[31 0]] | 0.999993 | 0.000013 | 1.956125 | 1.956112 | 6e-11
autoreg.layer_1.qX_0.variance[[32 0]] | 0.999996 | 0.000008 | 1.950590 | 1.950582 | 6e-11
autoreg.layer_1.qX_0.variance[[33 0]] | 0.999996 | 0.000007 | 1.969631 | 1.969624 | 6e-11
autoreg.layer_1.qX_0.variance[[34 0]] | 0.999997 | 0.000006 | 1.966159 | 1.966153 | 6e-11
autoreg.layer_1.qX_0.variance[[35 0]] | 0.999994 | 0.000013 | 1.987222 | 1.987210 | 6e-11
autoreg.layer_1.qX_0.variance[[36 0]] | 0.999999 | 0.000001 | 1.966518 | 1.966517 | 6e-11
autoreg.layer_1.qX_0.variance[[37 0]] | 0.999997 | 0.000006 | 1.957901 | 1.957895 | 6e-11
autoreg.layer_1.qX_0.variance[[38 0]] | 0.999999 | 0.000003 | 1.976670 | 1.976667 | 6e-11
autoreg.layer_1.qX_0.variance[[39 0]] | 0.999998 | 0.000004 | 1.994119 | 1.994114 | 6e-11
autoreg.layer_1.qX_0.variance[[40 0]] | 0.999998 | 0.000003 | 1.994168 | 1.994165 | 6e-11
autoreg.layer_1.qX_0.variance[[41 0]] | 0.999999 | 0.000001 | 1.969050 | 1.969049 | 6e-11
autoreg.layer_1.qX_0.variance[[42 0]] | 0.999998 | 0.000004 | 1.957666 | 1.957662 | 6e-11
autoreg.layer_1.qX_0.variance[[43 0]] | 0.999995 | 0.000010 | 1.939438 | 1.939428 | 6e-11
autoreg.layer_1.qX_0.variance[[44 0]] | 0.999994 | 0.000011 | 1.947872 | 1.947861 | 6e-11
autoreg.layer_1.qX_0.variance[[45 0]] | 0.999998 | 0.000003 | 1.949349 | 1.949345 | 6e-11
autoreg.layer_1.qX_0.variance[[46 0]] | 0.999996 | 0.000007 | 1.951034 | 1.951026 | 6e-11
autoreg.layer_1.qX_0.variance[[47 0]] | 0.999997 | 0.000005 | 1.949831 | 1.949826 | 6e-11
autoreg.layer_1.qX_0.variance[[48 0]] | 0.999998 | 0.000003 | 1.947093 | 1.947090 | 6e-11
autoreg.layer_1.qX_0.variance[[49 0]] | 0.999996 | 0.000009 | 1.952512 | 1.952503 | 6e-11
autoreg.layer_1.qX_0.variance[[50 0]] | 0.999994 | 0.000012 | 1.948201 | 1.948189 | 6e-11
autoreg.layer_1.qX_0.variance[[51 0]] | 0.999997 | 0.000006 | 1.964900 | 1.964894 | 6e-11
autoreg.layer_1.qX_0.variance[[52 0]] | 0.999992 | 0.000015 | 1.971553 | 1.971537 | 6e-11
autoreg.layer_1.qX_0.variance[[53 0]] | 0.999997 | 0.000006 | 1.973966 | 1.973960 | 6e-11
autoreg.layer_1.qX_0.variance[[54 0]] | 0.999996 | 0.000008 | 1.988659 | 1.988650 | 6e-11
autoreg.layer_1.qX_0.variance[[55 0]] | 0.999994 | 0.000012 | 1.967263 | 1.967252 | 6e-11
autoreg.layer_1.qX_0.variance[[56 0]] | 0.999997 | 0.000006 | 1.972002 | 1.971996 | 6e-11
autoreg.layer_1.qX_0.variance[[57 0]] | 0.999995 | 0.000010 | 1.948344 | 1.948334 | 6e-11
autoreg.layer_1.qX_0.variance[[58 0]] | 0.999996 | 0.000008 | 1.948204 | 1.948196 | 6e-11
autoreg.layer_1.qX_0.variance[[59 0]] | 0.999998 | 0.000004 | 1.949837 | 1.949833 | 6e-11
autoreg.layer_1.qX_0.variance[[60 0]] | 0.999996 | 0.000008 | 1.950830 | 1.950822 | 6e-11
autoreg.layer_1.qX_0.variance[[61 0]] | 0.999996 | 0.000008 | 1.950823 | 1.950815 | 6e-11
autoreg.layer_1.qX_0.variance[[62 0]] | 0.999996 | 0.000008 | 1.950823 | 1.950815 | 6e-11
autoreg.layer_1.qX_0.variance[[63 0]] | 0.999997 | 0.000006 | 1.950821 | 1.950815 | 6e-11
autoreg.layer_1.qX_0.variance[[64 0]] | 0.999995 | 0.000009 | 1.951777 | 1.951768 | 6e-11
autoreg.layer_1.qX_0.variance[[65 0]] | 0.999999 | 0.000002 | 1.950497 | 1.950495 | 6e-11
autoreg.layer_1.qX_0.variance[[66 0]] | 0.999995 | 0.000010 | 1.950673 | 1.950662 | 6e-11
autoreg.layer_1.qX_0.variance[[67 0]] | 0.999997 | 0.000006 | 1.951018 | 1.951012 | 6e-11
autoreg.layer_1.qX_0.variance[[68 0]] | 0.999997 | 0.000006 | 1.951353 | 1.951346 | 6e-11
autoreg.layer_1.qX_0.variance[[69 0]] | 1.000001 | 0.000002 | 1.952705 | 1.952707 | 6e-11
autoreg.layer_1.qX_0.variance[[70 0]] | 0.999993 | 0.000013 | 1.952407 | 1.952394 | 6e-11
autoreg.layer_1.qX_0.variance[[71 0]] | 0.999998 | 0.000004 | 1.956014 | 1.956010 | 6e-11
autoreg.layer_1.qX_0.variance[[72 0]] | 0.999996 | 0.000007 | 1.951586 | 1.951579 | 6e-11
autoreg.layer_1.qX_0.variance[[73 0]] | 0.999997 | 0.000006 | 1.946834 | 1.946828 | 6e-11
autoreg.layer_1.qX_0.variance[[74 0]] | 0.999993 | 0.000013 | 1.916071 | 1.916058 | 6e-11
autoreg.layer_1.qX_0.variance[[75 0]] | 0.999996 | 0.000007 | 1.936481 | 1.936474 | 6e-11
autoreg.layer_1.qX_0.variance[[76 0]] | 0.999996 | 0.000007 | 1.959786 | 1.959779 | 6e-11
autoreg.layer_1.qX_0.variance[[77 0]] | 0.999995 | 0.000009 | 1.958071 | 1.958062 | 6e-11
autoreg.layer_1.qX_0.variance[[78 0]] | 0.999998 | 0.000004 | 1.959085 | 1.959081 | 6e-11
autoreg.layer_1.qX_0.variance[[79 0]] | 0.999998 | 0.000004 | 1.964979 | 1.964974 | 6e-11
autoreg.layer_1.qX_0.variance[[80 0]] | 0.999995 | 0.000011 | 1.956312 | 1.956301 | 6e-11
autoreg.layer_1.qX_0.variance[[81 0]] | 0.999993 | 0.000014 | 1.943044 | 1.943030 | 6e-11
autoreg.layer_1.qX_0.variance[[82 0]] | 0.999996 | 0.000009 | 1.951384 | 1.951375 | 6e-11
autoreg.layer_1.qX_0.variance[[83 0]] | 0.999995 | 0.000009 | 1.950817 | 1.950808 | 6e-11
autoreg.layer_1.qX_0.variance[[84 0]] | 0.999997 | 0.000006 | 1.946994 | 1.946988 | 6e-11
autoreg.layer_1.qX_0.variance[[85 0]] | 0.999999 | 0.000002 | 1.943017 | 1.943015 | 6e-11
autoreg.layer_1.qX_0.variance[[86 0]] | 0.999991 | 0.000018 | 1.949742 | 1.949724 | 6e-11
autoreg.layer_1.qX_0.variance[[87 0]] | 0.999998 | 0.000005 | 1.952464 | 1.952460 | 6e-11
autoreg.layer_1.qX_0.variance[[88 0]] | 0.999995 | 0.000011 | 1.953933 | 1.953922 | 6e-11
autoreg.layer_1.qX_0.variance[[89 0]] | 0.999995 | 0.000011 | 1.951903 | 1.951892 | 6e-11
autoreg.layer_1.qX_0.variance[[90 0]] | 0.999997 | 0.000005 | 1.949969 | 1.949964 | 6e-11
autoreg.layer_1.qX_0.variance[[91 0]] | 0.999998 | 0.000003 | 1.943353 | 1.943350 | 6e-11
autoreg.layer_1.qX_0.variance[[92 0]] | 0.999995 | 0.000010 | 1.969539 | 1.969529 | 6e-11
autoreg.layer_1.qX_0.variance[[93 0]] | 0.999992 | 0.000015 | 1.931483 | 1.931468 | 6e-11
autoreg.layer_1.qX_0.variance[[94 0]] | 0.999996 | 0.000008 | 1.925044 | 1.925036 | 6e-11
autoreg.layer_1.qX_0.variance[[95 0]] | 0.999999 | 0.000002 | 1.913142 | 1.913140 | 6e-11
autoreg.layer_1.qX_0.variance[[96 0]] | 1.000001 | 0.000002 | 1.941325 | 1.941327 | 6e-11
autoreg.layer_1.qX_0.variance[[97 0]] | 0.999994 | 0.000012 | 1.950558 | 1.950546 | 6e-11
autoreg.layer_1.qX_0.variance[[98 0]] | 0.999996 | 0.000007 | 1.955894 | 1.955887 | 6e-11
autoreg.layer_1.qX_0.variance[[99 0]] | 1.000000 | 0.000001 | 1.985908 | 1.985907 | 6e-11
autoreg.layer_1.qX_0.variance[[100 0]] | 0.999997 | 0.000005 | 1.969287 | 1.969282 | 6e-11
autoreg.layer_1.qX_0.variance[[101 0]] | 0.999996 | 0.000008 | 1.973044 | 1.973036 | 6e-11
autoreg.layer_1.qX_0.variance[[102 0]] | 1.000001 | 0.000002 | 1.971826 | 1.971828 | 6e-11
autoreg.layer_1.qX_0.variance[[103 0]] | 1.000001 | 0.000001 | 2.002539 | 2.002540 | 6e-11
autoreg.layer_1.qX_0.variance[[104 0]] | 0.999999 | 0.000001 | 1.973161 | 1.973160 | 6e-11
autoreg.layer_1.qX_0.variance[[105 0]] | 0.999995 | 0.000010 | 1.957592 | 1.957582 | 6e-11
autoreg.layer_1.qX_0.variance[[106 0]] | 0.999995 | 0.000011 | 1.943899 | 1.943888 | 6e-11
autoreg.layer_1.qX_0.variance[[107 0]] | 0.999999 | 0.000002 | 1.939059 | 1.939057 | 6e-11
autoreg.layer_1.qX_0.variance[[108 0]] | 0.999998 | 0.000005 | 1.952086 | 1.952081 | 6e-11
autoreg.layer_1.qX_0.variance[[109 0]] | 0.999995 | 0.000009 | 1.950810 | 1.950801 | 6e-11
autoreg.layer_1.qX_0.variance[[110 0]] | 0.999995 | 0.000009 | 1.951239 | 1.951230 | 6e-11
autoreg.layer_1.qX_0.variance[[111 0]] | 0.999999 | 0.000002 | 1.938171 | 1.938170 | 6e-11
autoreg.layer_1.qX_0.variance[[112 0]] | 1.000000 | 0.000001 | 1.946581 | 1.946581 | 6e-11
autoreg.layer_1.qX_0.variance[[113 0]] | 0.999998 | 0.000004 | 1.951896 | 1.951892 | 6e-11
autoreg.layer_1.qX_0.variance[[114 0]] | 1.000000 | 0.000000 | 1.959888 | 1.959888 | 6e-11
autoreg.layer_1.qX_0.variance[[115 0]] | 0.999997 | 0.000006 | 1.966952 | 1.966946 | 6e-11
autoreg.layer_1.qX_0.variance[[116 0]] | 0.999998 | 0.000003 | 1.955322 | 1.955319 | 6e-11
autoreg.layer_1.qX_0.variance[[117 0]] | 0.999997 | 0.000006 | 1.933723 | 1.933717 | 6e-11
autoreg.layer_1.qX_0.variance[[118 0]] | 0.999994 | 0.000012 | 1.953170 | 1.953158 | 6e-11
autoreg.layer_1.qX_0.variance[[119 0]] | 0.999999 | 0.000001 | 1.952563 | 1.952561 | 6e-11
autoreg.layer_1.qX_0.variance[[120 0]] | 0.999992 | 0.000016 | 1.940790 | 1.940774 | 6e-11
autoreg.layer_1.qX_0.variance[[121 0]] | 0.999994 | 0.000011 | 1.952624 | 1.952612 | 6e-11
autoreg.layer_1.qX_0.variance[[122 0]] | 0.999994 | 0.000012 | 1.952872 | 1.952860 | 6e-11
autoreg.layer_1.qX_0.variance[[123 0]] | 0.999996 | 0.000008 | 1.950808 | 1.950801 | 6e-11
autoreg.layer_1.qX_0.variance[[124 0]] | 0.999996 | 0.000007 | 1.950822 | 1.950815 | 6e-11
autoreg.layer_1.qX_0.variance[[125 0]] | 0.999998 | 0.000005 | 1.950798 | 1.950793 | 6e-11
autoreg.layer_1.qX_0.variance[[126 0]] | 0.999996 | 0.000008 | 1.950664 | 1.950655 | 6e-11
autoreg.layer_1.qX_0.variance[[127 0]] | 0.999997 | 0.000005 | 1.950944 | 1.950939 | 6e-11
autoreg.layer_1.qX_0.variance[[128 0]] | 0.999993 | 0.000014 | 1.950306 | 1.950291 | 6e-11
autoreg.layer_1.qX_0.variance[[129 0]] | 0.999994 | 0.000011 | 1.946810 | 1.946799 | 6e-11
autoreg.layer_1.qX_0.variance[[130 0]] | 0.999999 | 0.000002 | 1.949573 | 1.949571 | 6e-11
autoreg.layer_1.qX_0.variance[[131 0]] | 0.999999 | 0.000002 | 1.952418 | 1.952416 | 6e-11
autoreg.layer_1.qX_0.variance[[132 0]] | 0.999998 | 0.000005 | 1.954400 | 1.954395 | 6e-11
autoreg.layer_1.qX_0.variance[[133 0]] | 0.999995 | 0.000009 | 1.954826 | 1.954817 | 6e-11
autoreg.layer_1.qX_0.variance[[134 0]] | 0.999996 | 0.000008 | 1.950685 | 1.950677 | 6e-11
autoreg.layer_1.qX_0.variance[[135 0]] | 0.999993 | 0.000013 | 1.949410 | 1.949396 | 6e-11
autoreg.layer_1.qX_0.variance[[136 0]] | 0.999996 | 0.000007 | 1.952656 | 1.952649 | 6e-11
autoreg.layer_1.qX_0.variance[[137 0]] | 0.999998 | 0.000003 | 1.951190 | 1.951186 | 6e-11
autoreg.layer_1.qX_0.variance[[138 0]] | 0.999996 | 0.000008 | 1.949812 | 1.949804 | 6e-11
autoreg.layer_1.qX_0.variance[[139 0]] | 0.999996 | 0.000009 | 1.956528 | 1.956520 | 6e-11
autoreg.layer_1.qX_0.variance[[140 0]] | 0.999997 | 0.000006 | 1.943989 | 1.943983 | 6e-11
autoreg.layer_1.qX_0.variance[[141 0]] | 0.999993 | 0.000013 | 1.948966 | 1.948953 | 6e-11
autoreg.layer_1.qX_0.variance[[142 0]] | 0.999997 | 0.000006 | 1.979197 | 1.979191 | 6e-11
autoreg.layer_1.qX_0.variance[[143 0]] | 0.999995 | 0.000011 | 1.983379 | 1.983368 | 6e-11
autoreg.layer_1.qX_0.variance[[144 0]] | 0.999995 | 0.000011 | 1.988275 | 1.988265 | 6e-11
autoreg.layer_1.qX_0.variance[[145 0]] | 0.999998 | 0.000003 | 1.961514 | 1.961511 | 6e-11
autoreg.layer_1.qX_0.variance[[146 0]] | 0.999996 | 0.000007 | 1.950837 | 1.950830 | 6e-11
autoreg.layer_1.qX_0.variance[[147 0]] | 0.999996 | 0.000008 | 1.956804 | 1.956796 | 6e-11
autoreg.layer_1.qX_0.variance[[148 0]] | 0.999991 | 0.000017 | 1.942552 | 1.942535 | 6e-11
autoreg.layer_1.qX_0.variance[[149 0]] | 0.999998 | 0.000004 | 2.008387 | 2.008383 | 6e-11
autoreg.layer_1.qX_0.variance[[150 0]] | 0.999998 | 0.000004 | 1.997684 | 1.997680 | 6e-11
autoreg.layer_1.qX_0.variance[[151 0]] | 0.999997 | 0.000006 | 2.000188 | 2.000183 | 6e-11
autoreg.layer_1.qX_0.variance[[152 0]] | 0.999997 | 0.000006 | 2.000269 | 2.000263 | 6e-11
autoreg.layer_1.qX_0.variance[[153 0]] | 0.999999 | 0.000002 | 1.910632 | 1.910630 | 6e-11
autoreg.layer_1.qX_0.variance[[154 0]] | 1.000002 | 0.000003 | 1.924102 | 1.924105 | 6e-11
autoreg.layer_1.qX_0.variance[[155 0]] | 0.999995 | 0.000010 | 1.948628 | 1.948618 | 6e-11
autoreg.layer_1.qX_0.variance[[156 0]] | 0.999997 | 0.000007 | 1.956061 | 1.956054 | 6e-11
autoreg.layer_1.qX_0.variance[[157 0]] | 0.999996 | 0.000008 | 1.962770 | 1.962762 | 6e-11
autoreg.layer_1.qX_0.variance[[158 0]] | 0.999995 | 0.000010 | 1.870550 | 1.870540 | 5e-11
autoreg.layer_1.qX_0.variance[[159 0]] | 0.999997 | 0.000006 | 1.987434 | 1.987428 | 6e-11
autoreg.layer_1.qX_0.variance[[160 0]] | 0.999994 | 0.000012 | 1.950638 | 1.950626 | 6e-11
autoreg.layer_1.qX_0.variance[[161 0]] | 0.999997 | 0.000007 | 2.027081 | 2.027075 | 6e-11
autoreg.layer_1.qX_0.variance[[162 0]] | 0.999999 | 0.000003 | 1.998788 | 1.998786 | 6e-11
autoreg.layer_1.qX_0.variance[[163 0]] | 1.000000 | 0.000001 | 1.923335 | 1.923334 | 6e-11
autoreg.layer_1.qX_0.variance[[164 0]] | 0.999995 | 0.000011 | 1.944190 | 1.944180 | 6e-11
autoreg.layer_1.qX_0.variance[[165 0]] | 0.999999 | 0.000003 | 1.950127 | 1.950124 | 6e-11
autoreg.layer_1.qX_0.variance[[166 0]] | 0.999996 | 0.000007 | 1.947075 | 1.947068 | 6e-11
autoreg.layer_1.qX_0.variance[[167 0]] | 0.999997 | 0.000005 | 1.953404 | 1.953398 | 6e-11
autoreg.layer_1.qX_0.variance[[168 0]] | 0.999997 | 0.000005 | 1.955026 | 1.955021 | 6e-11
autoreg.layer_1.qX_0.variance[[169 0]] | 0.999993 | 0.000013 | 1.948297 | 1.948283 | 6e-11
autoreg.layer_1.qX_0.variance[[170 0]] | 0.999994 | 0.000011 | 1.951663 | 1.951652 | 6e-11
autoreg.layer_1.qX_0.variance[[171 0]] | 0.999998 | 0.000004 | 1.945333 | 1.945329 | 6e-11
autoreg.layer_1.qX_0.variance[[172 0]] | 0.999995 | 0.000009 | 1.950686 | 1.950677 | 6e-11
autoreg.layer_1.qX_0.variance[[173 0]] | 0.999998 | 0.000005 | 1.950362 | 1.950357 | 6e-11
autoreg.layer_1.qX_0.variance[[174 0]] | 0.999993 | 0.000013 | 1.943923 | 1.943910 | 6e-11
autoreg.layer_1.qX_0.variance[[175 0]] | 0.999996 | 0.000008 | 1.934379 | 1.934372 | 6e-11
autoreg.layer_1.qX_0.variance[[176 0]] | 0.999999 | 0.000003 | 1.971343 | 1.971341 | 6e-11
autoreg.layer_1.qX_0.variance[[177 0]] | 0.999997 | 0.000006 | 1.972438 | 1.972432 | 6e-11
autoreg.layer_1.qX_0.variance[[178 0]] | 0.999994 | 0.000011 | 1.939512 | 1.939501 | 6e-11
autoreg.layer_1.qX_0.variance[[179 0]] | 0.999997 | 0.000007 | 1.986758 | 1.986751 | 6e-11
autoreg.layer_1.qX_0.variance[[180 0]] | 0.999996 | 0.000007 | 1.913934 | 1.913926 | 6e-11
autoreg.layer_1.qX_0.variance[[181 0]] | 0.999998 | 0.000005 | 1.966842 | 1.966837 | 6e-11
autoreg.layer_1.qX_0.variance[[182 0]] | 0.999995 | 0.000010 | 1.947114 | 1.947104 | 6e-11
autoreg.layer_1.qX_0.variance[[183 0]] | 0.999998 | 0.000003 | 1.926226 | 1.926222 | 6e-11
autoreg.layer_1.qX_0.variance[[184 0]] | 0.999994 | 0.000012 | 1.949037 | 1.949025 | 6e-11
autoreg.layer_1.qX_0.variance[[185 0]] | 0.999998 | 0.000003 | 1.953445 | 1.953442 | 6e-11
autoreg.layer_1.qX_0.variance[[186 0]] | 0.999996 | 0.000008 | 1.951835 | 1.951827 | 6e-11
autoreg.layer_1.qX_0.variance[[187 0]] | 0.999995 | 0.000011 | 1.950709 | 1.950699 | 6e-11
autoreg.layer_1.qX_0.variance[[188 0]] | 0.999996 | 0.000008 | 1.950590 | 1.950582 | 6e-11
autoreg.layer_1.qX_0.variance[[189 0]] | 0.999998 | 0.000004 | 1.950820 | 1.950815 | 6e-11
autoreg.layer_1.qX_0.variance[[190 0]] | 0.999994 | 0.000012 | 1.953876 | 1.953864 | 6e-11
autoreg.layer_1.qX_0.variance[[191 0]] | 0.999996 | 0.000008 | 1.959605 | 1.959597 | 6e-11
autoreg.layer_1.qX_0.variance[[192 0]] | 0.999995 | 0.000009 | 1.949398 | 1.949389 | 6e-11
autoreg.layer_1.qX_0.variance[[193 0]] | 0.999997 | 0.000006 | 1.956184 | 1.956178 | 6e-11
autoreg.layer_1.qX_0.variance[[194 0]] | 0.999996 | 0.000008 | 1.956557 | 1.956549 | 6e-11
autoreg.layer_1.qX_0.variance[[195 0]] | 0.999998 | 0.000003 | 1.947712 | 1.947708 | 6e-11
autoreg.layer_1.qX_0.variance[[196 0]] | 0.999996 | 0.000007 | 1.951230 | 1.951223 | 6e-11
autoreg.layer_1.qX_0.variance[[197 0]] | 0.999995 | 0.000011 | 1.949043 | 1.949033 | 6e-11
autoreg.layer_1.qX_0.variance[[198 0]] | 0.999995 | 0.000009 | 1.949995 | 1.949986 | 6e-11
autoreg.layer_1.qX_0.variance[[199 0]] | 0.999997 | 0.000005 | 1.952094 | 1.952088 | 6e-11
autoreg.layer_1.qX_0.variance[[200 0]] | 0.999998 | 0.000004 | 1.950368 | 1.950364 | 6e-11
autoreg.layer_1.qX_0.variance[[201 0]] | 0.999997 | 0.000005 | 1.940038 | 1.940032 | 6e-11
autoreg.layer_1.qX_0.variance[[202 0]] | 0.999999 | 0.000001 | 1.955218 | 1.955217 | 6e-11
autoreg.layer_1.qX_0.variance[[203 0]] | 0.999996 | 0.000008 | 1.958317 | 1.958309 | 6e-11
autoreg.layer_1.qX_0.variance[[204 0]] | 0.999994 | 0.000012 | 1.942184 | 1.942171 | 6e-11
autoreg.layer_1.qX_0.variance[[205 0]] | 1.000000 | 0.000001 | 1.955100 | 1.955101 | 6e-11
autoreg.layer_1.qX_0.variance[[206 0]] | 0.999998 | 0.000003 | 1.957862 | 1.957858 | 6e-11
autoreg.layer_1.qX_0.variance[[207 0]] | 0.999998 | 0.000004 | 1.966463 | 1.966459 | 6e-11
autoreg.layer_1.qX_0.variance[[208 0]] | 0.999994 | 0.000011 | 1.946090 | 1.946079 | 6e-11
autoreg.layer_1.qX_0.variance[[209 0]] | 0.999994 | 0.000011 | 1.955003 | 1.954992 | 6e-11
autoreg.layer_1.qX_0.variance[[210 0]] | 0.999996 | 0.000007 | 1.948574 | 1.948567 | 6e-11
autoreg.layer_1.qX_0.variance[[211 0]] | 1.000001 | 0.000002 | 1.954415 | 1.954417 | 6e-11
autoreg.layer_1.qX_0.variance[[212 0]] | 0.999994 | 0.000011 | 1.948520 | 1.948509 | 6e-11
autoreg.layer_1.qX_0.variance[[213 0]] | 0.999995 | 0.000010 | 1.951800 | 1.951790 | 6e-11
autoreg.layer_1.qX_0.variance[[214 0]] | 0.999995 | 0.000009 | 1.945316 | 1.945307 | 6e-11
autoreg.layer_1.qX_0.variance[[215 0]] | 0.999999 | 0.000001 | 1.971546 | 1.971544 | 6e-11
autoreg.layer_1.qX_0.variance[[216 0]] | 1.000000 | 0.000001 | 1.988882 | 1.988883 | 6e-11
autoreg.layer_1.qX_0.variance[[217 0]] | 0.999997 | 0.000007 | 2.006134 | 2.006127 | 6e-11
autoreg.layer_1.qX_0.variance[[218 0]] | 0.999996 | 0.000008 | 2.004957 | 2.004948 | 6e-11
autoreg.layer_1.qX_0.variance[[219 0]] | 0.999995 | 0.000011 | 1.995195 | 1.995184 | 6e-11
autoreg.layer_1.qX_0.variance[[220 0]] | 0.999999 | 0.000002 | 1.941817 | 1.941815 | 6e-11
autoreg.layer_1.qX_0.variance[[221 0]] | 0.999993 | 0.000014 | 1.962187 | 1.962173 | 6e-11
autoreg.layer_1.qX_0.variance[[222 0]] | 0.999994 | 0.000011 | 1.902667 | 1.902656 | 6e-11
autoreg.layer_1.qX_0.variance[[223 0]] | 0.999996 | 0.000009 | 1.966307 | 1.966298 | 6e-11
autoreg.layer_1.qX_0.variance[[224 0]] | 0.999998 | 0.000003 | 1.944721 | 1.944718 | 6e-11
autoreg.layer_1.qX_0.variance[[225 0]] | 0.999994 | 0.000012 | 1.950703 | 1.950692 | 6e-11
autoreg.layer_1.qX_0.variance[[226 0]] | 0.999997 | 0.000005 | 1.963612 | 1.963606 | 6e-11
autoreg.layer_1.qX_0.variance[[227 0]] | 0.999995 | 0.000011 | 2.206976 | 2.206965 | 6e-11
autoreg.layer_1.qX_0.variance[[228 0]] | 0.999998 | 0.000006 | 2.242128 | 2.242123 | 7e-11
autoreg.layer_1.qX_0.variance[[229 0]] | 0.999998 | 0.000004 | 2.403100 | 2.403096 | 7e-11
autoreg.layer_1.qX_0.variance[[230 0]] | 0.999996 | 0.000009 | 2.480536 | 2.480527 | 7e-11
autoreg.layer_1.qX_0.variance[[231 0]] | 0.999996 | 0.000008 | 2.164642 | 2.164634 | 6e-11
autoreg.layer_1.qX_0.variance[[232 0]] | 0.999995 | 0.000011 | 2.059253 | 2.059242 | 6e-11
autoreg.layer_1.qX_0.variance[[233 0]] | 0.999992 | 0.000015 | 2.035173 | 2.035158 | 6e-11
autoreg.layer_1.qX_0.variance[[234 0]] | 0.999996 | 0.000008 | 1.995032 | 1.995024 | 6e-11
autoreg.layer_1.qX_0.variance[[235 0]] | 0.999994 | 0.000012 | 1.984937 | 1.984925 | 6e-11
autoreg.layer_1.qX_0.variance[[236 0]] | 0.999994 | 0.000011 | 1.946991 | 1.946981 | 6e-11
autoreg.layer_1.qX_0.variance[[237 0]] | 0.999999 | 0.000002 | 1.934512 | 1.934510 | 6e-11
autoreg.layer_1.qX_0.variance[[238 0]] | 0.999995 | 0.000009 | 1.967493 | 1.967484 | 6e-11
autoreg.layer_1.qX_0.variance[[239 0]] | 0.999993 | 0.000013 | 1.965257 | 1.965243 | 6e-11
autoreg.layer_1.qX_0.variance[[240 0]] | 0.999997 | 0.000005 | 1.954618 | 1.954613 | 6e-11
autoreg.layer_1.qX_0.variance[[241 0]] | 0.999998 | 0.000004 | 1.924961 | 1.924956 | 6e-11
autoreg.layer_1.qX_0.variance[[242 0]] | 0.999998 | 0.000003 | 1.987708 | 1.987704 | 6e-11
autoreg.layer_1.qX_0.variance[[243 0]] | 0.999995 | 0.000010 | 1.962642 | 1.962631 | 6e-11
autoreg.layer_1.qX_0.variance[[244 0]] | 0.999997 | 0.000006 | 1.984371 | 1.984365 | 6e-11
autoreg.layer_1.qX_0.variance[[245 0]] | 0.999994 | 0.000012 | 1.985665 | 1.985652 | 6e-11
autoreg.layer_1.qX_0.variance[[246 0]] | 0.999996 | 0.000007 | 1.976681 | 1.976674 | 6e-11
autoreg.layer_1.qX_0.variance[[247 0]] | 0.999997 | 0.000006 | 1.969164 | 1.969158 | 6e-11
autoreg.layer_1.qX_0.variance[[248 0]] | 0.999996 | 0.000009 | 1.950598 | 1.950590 | 6e-11
autoreg.layer_1.qX_0.variance[[249 0]] | 0.999997 | 0.000007 | 1.950756 | 1.950750 | 6e-11
autoreg.layer_1.qX_0.variance[[250 0]] | 0.999995 | 0.000009 | 1.951181 | 1.951172 | 6e-11
autoreg.layer_1.qX_0.variance[[251 0]] | 0.999997 | 0.000006 | 1.949868 | 1.949862 | 6e-11
autoreg.layer_1.qX_0.variance[[252 0]] | 0.999993 | 0.000013 | 1.947365 | 1.947352 | 6e-11
autoreg.layer_1.qX_0.variance[[253 0]] | 0.999995 | 0.000010 | 1.950418 | 1.950408 | 6e-11
autoreg.layer_1.qX_0.variance[[254 0]] | 0.999997 | 0.000006 | 1.952079 | 1.952074 | 6e-11
autoreg.layer_1.qX_0.variance[[255 0]] | 0.999995 | 0.000009 | 1.946001 | 1.945991 | 6e-11
autoreg.layer_1.qX_0.variance[[256 0]] | 0.999997 | 0.000006 | 1.945881 | 1.945875 | 6e-11
autoreg.layer_1.qX_0.variance[[257 0]] | 0.999996 | 0.000008 | 1.951922 | 1.951914 | 6e-11
autoreg.layer_1.qX_0.variance[[258 0]] | 0.999994 | 0.000013 | 1.950675 | 1.950662 | 6e-11
autoreg.layer_1.qX_0.variance[[259 0]] | 0.999996 | 0.000009 | 1.947506 | 1.947497 | 6e-11
autoreg.layer_1.qX_0.variance[[260 0]] | 0.999994 | 0.000012 | 1.951606 | 1.951594 | 6e-11
autoreg.layer_1.qX_0.variance[[261 0]] | 0.999997 | 0.000006 | 1.951236 | 1.951230 | 6e-11
autoreg.layer_1.qX_0.variance[[262 0]] | 0.999996 | 0.000008 | 1.950620 | 1.950611 | 6e-11
autoreg.layer_1.qX_0.variance[[263 0]] | 0.999992 | 0.000015 | 1.951034 | 1.951019 | 6e-11
autoreg.layer_1.qX_0.variance[[264 0]] | 0.999994 | 0.000013 | 1.948303 | 1.948290 | 6e-11
autoreg.layer_1.qX_0.variance[[265 0]] | 0.999996 | 0.000008 | 1.951194 | 1.951186 | 6e-11
autoreg.layer_1.qX_0.variance[[266 0]] | 0.999996 | 0.000008 | 1.951056 | 1.951048 | 6e-11
autoreg.layer_1.qX_0.variance[[267 0]] | 0.999995 | 0.000010 | 1.949821 | 1.949811 | 6e-11
autoreg.layer_1.qX_0.variance[[268 0]] | 0.999995 | 0.000009 | 1.949238 | 1.949229 | 6e-11
autoreg.layer_1.qX_0.variance[[269 0]] | 0.999999 | 0.000002 | 1.947783 | 1.947781 | 6e-11
autoreg.layer_1.qX_0.variance[[270 0]] | 0.999995 | 0.000010 | 1.949450 | 1.949440 | 6e-11
autoreg.layer_1.qX_0.variance[[271 0]] | 0.999995 | 0.000011 | 1.953940 | 1.953929 | 6e-11
autoreg.layer_1.qX_0.variance[[272 0]] | 0.999998 | 0.000004 | 1.949066 | 1.949062 | 6e-11
autoreg.layer_1.qX_0.variance[[273 0]] | 0.999997 | 0.000006 | 1.951113 | 1.951106 | 6e-11
autoreg.layer_1.qX_0.variance[[274 0]] | 0.999997 | 0.000005 | 1.950420 | 1.950415 | 6e-11
autoreg.layer_1.qX_0.variance[[275 0]] | 0.999997 | 0.000006 | 1.950880 | 1.950873 | 6e-11
autoreg.layer_1.qX_0.variance[[276 0]] | 0.999998 | 0.000005 | 1.950893 | 1.950888 | 6e-11
autoreg.layer_1.qX_0.variance[[277 0]] | 0.999996 | 0.000009 | 1.950889 | 1.950881 | 6e-11
autoreg.layer_1.qX_0.variance[[278 0]] | 0.999995 | 0.000010 | 1.950774 | 1.950764 | 6e-11
autoreg.layer_1.qX_0.variance[[279 0]] | 0.999996 | 0.000008 | 1.949012 | 1.949003 | 6e-11
autoreg.layer_1.qX_0.variance[[280 0]] | 0.999995 | 0.000011 | 1.949109 | 1.949098 | 6e-11
autoreg.layer_1.qX_0.variance[[281 0]] | 0.999995 | 0.000011 | 1.949050 | 1.949040 | 6e-11
autoreg.layer_1.qX_0.variance[[282 0]] | 0.999999 | 0.000003 | 1.949035 | 1.949033 | 6e-11
autoreg.layer_1.qX_0.variance[[283 0]] | 0.999999 | 0.000002 | 1.953619 | 1.953616 | 6e-11
autoreg.layer_1.qX_0.variance[[284 0]] | 0.999995 | 0.000010 | 1.950025 | 1.950015 | 6e-11
autoreg.layer_1.qX_0.variance[[285 0]] | 0.999995 | 0.000011 | 1.950491 | 1.950481 | 6e-11
autoreg.layer_1.qX_0.variance[[286 0]] | 0.999995 | 0.000010 | 1.949588 | 1.949578 | 6e-11
autoreg.layer_1.qX_0.variance[[287 0]] | 0.999994 | 0.000012 | 1.956961 | 1.956949 | 6e-11
autoreg.layer_1.qX_0.variance[[288 0]] | 0.999996 | 0.000007 | 1.951805 | 1.951797 | 6e-11
autoreg.layer_1.qX_0.variance[[289 0]] | 0.999997 | 0.000007 | 1.959917 | 1.959910 | 6e-11
autoreg.layer_1.qX_0.variance[[290 0]] | 0.999997 | 0.000007 | 1.932370 | 1.932363 | 6e-11
autoreg.layer_1.qX_0.variance[[291 0]] | 0.999997 | 0.000006 | 1.949941 | 1.949935 | 6e-11
autoreg.layer_1.qX_0.variance[[292 0]] | 0.999995 | 0.000009 | 1.959359 | 1.959350 | 6e-11
autoreg.layer_1.qX_0.variance[[293 0]] | 0.999994 | 0.000011 | 1.956975 | 1.956963 | 6e-11
autoreg.layer_1.qX_0.variance[[294 0]] | 0.999997 | 0.000006 | 1.949963 | 1.949957 | 6e-11
autoreg.layer_1.qX_0.variance[[295 0]] | 0.999992 | 0.000015 | 1.957211 | 1.957196 | 6e-11
autoreg.layer_1.qX_0.variance[[296 0]] | 0.999994 | 0.000011 | 1.958401 | 1.958389 | 6e-11
autoreg.layer_1.qX_0.variance[[297 0]] | 0.999994 | 0.000011 | 1.965444 | 1.965433 | 6e-11
autoreg.layer_1.qX_0.variance[[298 0]] | 0.999991 | 0.000017 | 1.944924 | 1.944907 | 6e-11
autoreg.layer_1.qX_0.variance[[299 0]] | 0.999995 | 0.000010 | 1.961499 | 1.961489 | 6e-11
autoreg.layer_0.inducing_inputs[[0 0]] | 1.000000 | 0.000001 | 2.611806 | 2.611807 | 8e-11
autoreg.layer_0.inducing_inputs[[0 1]] | 1.000001 | 0.000003 | 2.663122 | 2.663124 | 8e-11
autoreg.layer_0.inducing_inputs[[0 2]] | 1.000003 | 0.000002 | -0.558530 | -0.558532 | 2e-11
autoreg.layer_0.inducing_inputs[[0 3]] | 0.999999 | 0.000002 | -2.904804 | -2.904802 | 9e-11
autoreg.layer_0.inducing_inputs[[0 4]] | 1.000000 | 0.000001 | 2.451861 | 2.451859 | 7e-11
autoreg.layer_0.inducing_inputs[[1 0]] | 0.999998 | 0.000002 | -1.061404 | -1.061402 | 3e-11
autoreg.layer_0.inducing_inputs[[1 1]] | 0.999997 | 0.000001 | -0.351532 | -0.351531 | 1e-11
autoreg.layer_0.inducing_inputs[[1 2]] | 1.000002 | 0.000003 | -1.514589 | -1.514592 | 4e-11
autoreg.layer_0.inducing_inputs[[1 3]] | 1.000002 | 0.000002 | -1.008795 | -1.008797 | 3e-11
autoreg.layer_0.inducing_inputs[[1 4]] | 1.000001 | 0.000002 | -2.879590 | -2.879591 | 8e-11
autoreg.layer_0.inducing_inputs[[2 0]] | 1.000007 | 0.000005 | 0.701084 | 0.701089 | 2e-11
autoreg.layer_0.inducing_inputs[[2 1]] | 1.000001 | 0.000001 | -1.306048 | -1.306049 | 4e-11
autoreg.layer_0.inducing_inputs[[2 2]] | 1.000045 | 0.000004 | -0.077711 | -0.077715 | 2e-12
autoreg.layer_0.inducing_inputs[[2 3]] | 1.000002 | 0.000004 | -2.327510 | -2.327513 | 7e-11
autoreg.layer_0.inducing_inputs[[2 4]] | 1.000000 | 0.000000 | -8.039736 | -8.039737 | 2e-10
autoreg.layer_0.rbf.variance | 1.000000 | 0.000001 | 91.036273 | 91.036272 | 3e-09
autoreg.layer_0.rbf.inv_lengthscale[[0]] | 1.000000 | 0.000000 | 5.280265 | 5.280264 | 2e-10
autoreg.layer_0.rbf.inv_lengthscale[[1]] | 1.000000 | 0.000000 | 4.300433 | 4.300433 | 1e-10
autoreg.layer_0.rbf.inv_lengthscale[[2]] | 1.000001 | 0.000004 | 3.865014 | 3.865018 | 1e-10
autoreg.layer_0.rbf.inv_lengthscale[[3]] | 1.000000 | 0.000001 | 4.431705 | 4.431706 | 1e-10
autoreg.layer_0.rbf.inv_lengthscale[[4]] | 0.999999 | 0.000003 | 3.717988 | 3.717985 | 1e-10
autoreg.layer_0.Gaussian_noise.variance | 1.000000 | 0.000006 | -317.078704 | -317.078709 | 9e-09
autoreg.layer_0.qU_m[[0 0]] | 1.000000 | 0.000000 | 3.302979 | 3.302979 | 1e-10
autoreg.layer_0.qU_m[[1 0]] | 1.000000 | 0.000001 | -2.128268 | -2.128269 | 6e-11
autoreg.layer_0.qU_m[[2 0]] | 1.000000 | 0.000002 | -6.936379 | -6.936381 | 2e-10
autoreg.layer_0.qU_W[[0 0]] | 1.000001 | 0.000003 | 4.386599 | 4.386602 | 1e-10
autoreg.layer_0.qU_W[[0 1]] | 1.000000 | 0.000001 | 4.191564 | 4.191563 | 1e-10
autoreg.layer_0.qU_W[[0 2]] | 1.000000 | 0.000000 | -10.847777 | -10.847776 | 3e-10
autoreg.layer_0.qU_W[[1 0]] | 1.000004 | 0.000004 | -0.924326 | -0.924330 | 3e-11
autoreg.layer_0.qU_W[[1 1]] | 1.000001 | 0.000004 | -3.768381 | -3.768386 | 1e-10
autoreg.layer_0.qU_W[[1 2]] | 1.000000 | 0.000002 | 5.147185 | 5.147187 | 2e-10
autoreg.layer_0.qU_W[[2 0]] | 1.000000 | 0.000000 | 8.227296 | 8.227296 | 2e-10
autoreg.layer_0.qU_W[[2 1]] | 1.000001 | 0.000003 | 4.424427 | 4.424430 | 1e-10
autoreg.layer_0.qU_W[[2 2]] | 1.000000 | 0.000000 | 2.233297 | 2.233297 | 7e-11
autoreg.layer_0.qU_a | 1.000001 | 0.000001 | -1.304825 | -1.304827 | 4e-11
Out[49]:
True
In [36]:
# Here layer numbers are different than in initialization. 0-th layer is the top one
for i in range(m.nLayers):
m.layers[i].kern.inv_l[:] = np.mean( 1./((m.layers[i].X.mean.values.max(0)-m.layers[i].X.mean.values.min(0))/np.sqrt(2.)) )
m.layers[i].likelihood.variance[:] = 0.01*out_train.var()
m.layers[i].kern.variance.fix(warning=False)
m.layers[i].likelihood.fix(warning=False)
print(m)
Name : autoreg
Objective : 15409.2757564
Number of Parameters : 6471
Number of Optimization Parameters : 6467
Updates : True
Parameters:
autoreg. | value | constraints | priors
layer_1.inducing_inputs | (50, 10) | |
layer_1.rbf.variance | 1.0 | +ve fixed |
layer_1.rbf.inv_lengthscale | (10,) | +ve |
layer_1.Gaussian_noise.variance | 0.0352415434704 | +ve fixed |
layer_1.qU_m | (50, 1) | |
layer_1.qU_W | (50, 50) | |
layer_1.qU_a | 0.001 | +ve |
layer_1.qX_0.mean | (300, 1) | |
layer_1.qX_0.variance | (300, 1) | +ve |
layer_0.inducing_inputs | (50, 5) | |
layer_0.rbf.variance | 1.0 | +ve fixed |
layer_0.rbf.inv_lengthscale | (5,) | +ve |
layer_0.Gaussian_noise.variance | 0.0352415434704 | +ve fixed |
layer_0.qU_m | (50, 1) | |
layer_0.qU_W | (50, 50) | |
layer_0.qU_a | 0.001 | +ve |
In [37]:
print(m.layer_1.kern.inv_l)
print(m.layer_0.kern.inv_l)
print( np.mean(1./((m.layer_1.X.mean.values.max(0)-m.layer_1.X.mean.values.min(0))/np.sqrt(2.))) )
index | autoreg.layer_1.rbf.inv_lengthscale | constraints | priors
[0] | 0.20364643 | +ve |
[1] | 0.20364643 | +ve |
[2] | 0.20364643 | +ve |
[3] | 0.20364643 | +ve |
[4] | 0.20364643 | +ve |
[5] | 0.20364643 | +ve |
[6] | 0.20364643 | +ve |
[7] | 0.20364643 | +ve |
[8] | 0.20364643 | +ve |
[9] | 0.20364643 | +ve |
index | autoreg.layer_0.rbf.inv_lengthscale | constraints | priors
[0] | 0.12300233 | +ve |
[1] | 0.12300233 | +ve |
[2] | 0.12300233 | +ve |
[3] | 0.12300233 | +ve |
[4] | 0.12300233 | +ve |
0.203646429392
In [38]:
# Plot initialization of hidden layer:
def plot_hidden_states(fig_no, layer, layer_start_point=None, layer_end_point=None,
data_start_point=None, data_end_point=None):
if layer_start_point is None: layer_start_point=0;
if layer_end_point is None: layer_end_point = len(layer.mean)
if data_start_point is None: data_start_point=0;
if data_end_point is None: layer_end_point = len(out_train)
data = out_train[data_start_point:data_end_point]
layer_means = layer.mean[layer_start_point:layer_end_point]
layer_vars = layer.variance[layer_start_point:layer_end_point]
fig4 = plt.figure(fig_no,figsize=(10,8))
ax1 = plt.subplot(1,1,1)
fig4.suptitle('Hidden layer plotting')
ax1.plot(out_train[data_start_point:data_end_point], label="Orig data Train_out", color = 'b')
ax1.plot( layer_means, label = 'pred mean', color = 'r' )
ax1.plot( layer_means +\
2*np.sqrt( layer_vars ), label = 'pred var', color='r', linestyle='--' )
ax1.plot( layer_means -\
2*np.sqrt( layer_vars ), label = 'pred var', color='r', linestyle='--' )
ax1.legend(loc=4)
ax1.set_title('Hidden layer vs Training data')
del ax1
plot_hidden_states(5,m.layer_1.qX_0)
#plot_hidden_states(6,m.layer_2.qX_0)
In [39]:
#init_runs = 50 if out_train.shape[0]<1000 else 100
init_runs = 100
print("Init runs: ", init_runs)
m.optimize('bfgs',messages=1,max_iters=init_runs)
for i in range(m.nLayers):
m.layers[i].kern.variance.constrain_positive(warning=False)
m.layers[i].likelihood.constrain_positive(warning=False)
m.optimize('bfgs',messages=1,max_iters=10000)
print(m)
Init runs: 100
Running L-BFGS-B (Scipy implementation) Code:
runtime i f |g|
00s11 001 1.540928e+04 4.002577e+06
01s23 015 2.122730e+03 7.647174e+04
04s35 054 1.009164e+03 2.701847e+04
08s20 102 6.928191e+02 5.026328e+03
Runtime: 08s20
Optimization status: Maximum number of f evaluations reached
Running L-BFGS-B (Scipy implementation) Code:
runtime i f |g|
00s18 00002 6.303080e+02 5.814468e+05
03s29 00041 4.868468e+02 1.091685e+03
04s33 00053 4.747719e+02 3.797151e+03
14s86 00173 4.435939e+02 6.000586e+02
20s08 00229 4.404461e+02 4.280249e+02
59s53 00659 4.363617e+02 3.818287e+01
01m07s86 00760 4.360309e+02 7.395294e+01
03m09s39 02135 4.343032e+02 1.293935e+01
07m58s16 05615 4.337770e+02 1.346399e+01
Runtime: 07m58s16
Optimization status: Converged
Name : autoreg
Objective : 433.776951463
Number of Parameters : 6471
Number of Optimization Parameters : 6471
Updates : True
Parameters:
autoreg. | value | constraints | priors
layer_1.inducing_inputs | (50, 10) | |
layer_1.rbf.variance | 24.8967901913 | +ve |
layer_1.rbf.inv_lengthscale | (10,) | +ve |
layer_1.Gaussian_noise.variance | 0.0252376880307 | +ve |
layer_1.qU_m | (50, 1) | |
layer_1.qU_W | (50, 50) | |
layer_1.qU_a | 3.12959436094e-05 | +ve |
layer_1.qX_0.mean | (300, 1) | |
layer_1.qX_0.variance | (300, 1) | +ve |
layer_0.inducing_inputs | (50, 5) | |
layer_0.rbf.variance | 18.8966092153 | +ve |
layer_0.rbf.inv_lengthscale | (5,) | +ve |
layer_0.Gaussian_noise.variance | 0.116395672828 | +ve |
layer_0.qU_m | (50, 1) | |
layer_0.qU_W | (50, 50) | |
layer_0.qU_a | 0.00031457685719 | +ve |
In [40]:
if hasattr(m, 'layer_1'):
print("Layer 1: ")
print("States means (min and max), shapes: ", m.layer_1.qX_0.mean.min(),
m.layer_1.qX_0.mean.max(), m.layer_1.qX_0.mean.shape)
print("States variances (min and max), shapes: ", m.layer_1.qX_0.variance.min(),
m.layer_1.qX_0.variance.max(), m.layer_1.qX_0.mean.shape)
print("Inverse langthscales (min and max), shapes: ", m.layer_1.rbf.inv_lengthscale.min(),
m.layer_1.rbf.inv_lengthscale.max(), m.layer_1.rbf.inv_lengthscale.shape )
if hasattr(m, 'layer_0'):
print("")
print("Layer 0 (output): ")
print("Inverse langthscales (min and max), shapes: ", m.layer_0.rbf.inv_lengthscale.min(),
m.layer_0.rbf.inv_lengthscale.max(), m.layer_0.rbf.inv_lengthscale.shape )
Layer 1:
States means (min and max), shapes: -1.82693539616 2.12233556682 (300, 1)
States variances (min and max), shapes: 0.00070198191 0.865057396343 (300, 1)
Inverse langthscales (min and max), shapes: 1.31769235383e-08 0.753964668733 (10,)
Layer 0 (output):
Inverse langthscales (min and max), shapes: 1.44814063387e-08 0.746273629423 (5,)
In [41]:
print(m.layer_0.rbf.inv_lengthscale)
index | autoreg.layer_0.rbf.inv_lengthscale | constraints | priors
[0] | 0.00221387 | +ve |
[1] | 0.00000001 | +ve |
[2] | 0.01447472 | +ve |
[3] | 0.74627363 | +ve |
[4] | 0.32789038 | +ve |
In [42]:
print(m.layer_1.rbf.inv_lengthscale)
index | autoreg.layer_1.rbf.inv_lengthscale | constraints | priors
[0] | 0.00008952 | +ve |
[1] | 0.00000003 | +ve |
[2] | 0.00000004 | +ve |
[3] | 0.00167606 | +ve |
[4] | 0.75396467 | +ve |
[5] | 0.00000008 | +ve |
[6] | 0.00000001 | +ve |
[7] | 0.00007849 | +ve |
[8] | 0.00016292 | +ve |
[9] | 0.00149943 | +ve |
In [43]:
# Free-run on the train data
# initialize to last part of trained latent states
#init_Xs = [None, m.layer_1.qX_0[0:win_out]] # init_Xs for train prediction
# initialize to zeros
init_Xs = None
predictions_train = m.freerun(init_Xs = init_Xs, U=in_train, m_match=True)
# initialize to last part of trainig latent states
#init_Xs = [None, m.layer_1.qX_0[-win_out:] ] # init_Xs for test prediction
#U_test = np.vstack( (in_train[-win_in:], in_test) )
# initialize to zeros
init_Xs = None
U_test = in_test
# Free-run on the test data
predictions_test = m.freerun(init_Xs = init_Xs, U=U_test, m_match=True)
del init_Xs, U_test
In [44]:
# Plot predictions
def plot_predictions(fig_no,posterior_train, posterior_test=None, layer_no = None):
"""
Plots the output data along with posterior of the layer.
Used for plotting the hidden states or
layer_no: int or Normal posterior
plot states of this layer (0-th is output). There is also some logic about compting
the MSE, and aligning with actual data.
"""
if layer_no is None: #default
layer_no = 1
if posterior_test is None:
no_test_data = True
else:
no_test_data = False
if isinstance(posterior_train, list):
layer_in_list = len(predictions_train)-1-layer_no # standard layer no (like in printing the model)
predictions_train_layer = predictions_train[layer_in_list]
else:
predictions_train_layer = posterior_train
if not no_test_data:
if isinstance(posterior_test, list):
predictions_test_layer = predictions_test[layer_in_list]
else:
predictions_test_layer = posterior_test
# Aligning the data ->
# training of test data can be longer than leyer data because of the initial window.
if out_train.shape[0] > predictions_train_layer.mean.shape[0]:
out_train_tmp = out_train[win_out:]
else:
out_train_tmp = out_train
if out_test.shape[0] > predictions_test_layer.mean.shape[0]:
out_test_tmp = out_test[win_out:]
else:
out_test_tmp = out_test
# Aligning the data <-
if layer_no == 0:
# Not anymore! Compute RMSE ignoring first output values of length "win_out"
train_rmse = [comp_RMSE(predictions_train_layer.mean,
out_train_tmp)]
print("Train overall RMSE: ", str(train_rmse))
if not no_test_data:
# Compute RMSE ignoring first output values of length "win_out"
test_rmse = [comp_RMSE(predictions_test_layer.mean,
out_test_tmp)]
print("Test overall RMSE: ", str(test_rmse))
# Plot predictions:
if not no_test_data:
fig5 = plt.figure(10,figsize=(20,8))
else:
fig5 = plt.figure(10,figsize=(10,8))
fig5.suptitle('Predictions on Training and Test data')
if not no_test_data:
ax1 = plt.subplot(1,2,1)
else:
ax1 = plt.subplot(1,1,1)
ax1.plot(out_train_tmp, label="Train_out", color = 'b')
ax1.plot( predictions_train_layer.mean, label = 'pred mean', color = 'r' )
ax1.plot( predictions_train_layer.mean +\
2*np.sqrt( predictions_train_layer.variance ), label = 'pred var', color='r', linestyle='--' )
ax1.plot( predictions_train_layer.mean -\
2*np.sqrt( predictions_train_layer.variance ), label = 'pred var', color='r', linestyle='--' )
ax1.legend(loc=4)
ax1.set_title('Predictions on Train')
if not no_test_data:
ax2 = plt.subplot(1,2,2)
ax2.plot(out_test_tmp, label="Test_out", color = 'b')
ax2.plot( predictions_test_layer.mean, label = 'pred mean', color = 'r' )
#ax2.plot( predictions_test_layer.mean +\
# 2*np.sqrt( predictions_test_layer.variance ), label = 'pred var', color='r', linestyle='--' )
#ax2.plot( predictions_test_layer.mean -\
# 2*np.sqrt( predictions_test_layer.variance ), label = 'pred var', color='r', linestyle='--' )
ax2.legend(loc=4)
ax2.set_title('Predictions on Test')
del ax2
del ax1
plot_predictions(7,predictions_train, predictions_test , layer_no = 0)
Train overall RMSE: [1.7609920132306214]
Test overall RMSE: [1.9121407759057965]
In [ ]:
In [45]:
comp_RMSE(np.zeros( (len(out_train[20:]),1) ), out_train[20:] )
Out[45]:
1.8982884835906451
In [46]:
out_train[20:].mean(0)
Out[46]:
array([ 0.07638323])
In [47]:
plot_hidden_states(8,m.layer_1.qX_0)
#plot_hidden_states(9,m.layer_2.qX_0)
In [ ]:
Content source: zhenwendai/RGP
Similar notebooks: