3.7.1 梯度消失
[1]:
import warnings
warnings.simplefilter("ignore")
导入功能模块
[2]:
import sys
load_dataset_module_path = '../../'
sys.path.append(load_dataset_module_path)
from load_hyperspectral_dataset import (load_hyperspectral_data, y_labels,
extract_features,
plot_selected_categories,
plot_decision_function)
[3]:
#%matplotlib inline
#%matplotlib notebook
#%config InlineBackend.figure_format = 'retina'
import matplotlib as mpl
import matplotlib.pyplot as plt
from matplotlib import cm
from sklearn.utils import shuffle
import seaborn as sns
from ipywidgets import interact_manual
mpl.style.use('ggplot')
加载数据集
[4]:
img_file_path = '../../Hyperspectral_Project/dc.tif'
label_file_path = '../../Hyperspectral_Project/dctest.project'
raw_X, raw_y, pixel_position = load_hyperspectral_data(img_file_path,
label_file_path)
hyperspectral_df = extract_features(raw_X, raw_y)
画出指定类别，训练模型分类
[5]:
import torch
x = torch.rand(5, 3) print(x)
[30]:
x = torch.ones((100, 2)).uniform_(-1, 1)
y = 0.5*torch.ones((100,1))
print(x)
tensor([[ 0.8818, -0.8668],
[-0.1363, -0.2439],
[ 0.8219, -0.2049],
[ 0.4047, -0.0252],
[-0.7581, -0.2658],
[-0.7607, -0.4452],
[-0.8348, 0.8814],
[-0.9340, -0.7795],
[ 0.3956, 0.6170],
[ 0.1705, 0.7711],
[-0.1255, -0.2936],
[-0.9499, 0.2035],
[-0.1676, 0.8705],
[-0.1025, 0.0206],
[ 0.3198, 0.5712],
[ 0.0584, 0.3835],
[-0.2871, -0.6258],
[ 0.9953, -0.6273],
[-0.0421, 0.6482],
[-0.8480, -0.4230],
[-0.3497, -0.7190],
[ 0.8050, 0.2406],
[-0.6468, -0.6207],
[-0.8344, -0.0309],
[-0.9921, -0.1161],
[-0.0555, -0.8394],
[ 0.8810, -0.9356],
[ 0.0206, 0.2708],
[ 0.2776, -0.5339],
[ 0.0023, -0.2588],
[-0.0672, 0.2231],
[ 0.5111, 0.9055],
[ 0.5452, -0.1510],
[ 0.4759, 0.6258],
[ 0.0124, -0.6646],
[ 0.8518, 0.9648],
[ 0.6189, -0.3919],
[ 0.8161, 0.8570],
[-0.5948, 0.9784],
[-0.9799, 0.3437],
[-0.7858, -0.0701],
[-0.7290, 0.1747],
[ 0.0072, 0.9585],
[-0.5910, 0.8373],
[ 0.5548, 0.9877],
[-0.2046, 0.7552],
[ 0.1421, 0.4855],
[ 0.7442, 0.4239],
[ 0.9459, 0.5004],
[-0.3639, 0.9261],
[ 0.1297, -0.6969],
[-0.7324, -0.8119],
[ 0.7554, -0.6773],
[ 0.8009, -0.0191],
[-0.9201, 0.7688],
[-0.3360, 0.6330],
[ 0.9081, 0.1893],
[ 0.8211, -0.2364],
[ 0.6889, -0.1804],
[ 0.2079, -0.5484],
[-0.0785, -0.4735],
[-0.3998, 0.7603],
[ 0.7154, 0.4627],
[-0.8816, -0.4077],
[-0.4585, 0.1315],
[-0.1251, 0.4851],
[-0.3616, 0.0839],
[-0.0925, -0.2280],
[-0.7314, 0.4661],
[ 0.5474, 0.9387],
[-0.3722, 0.4936],
[-0.1138, -0.4359],
[-0.0136, -0.0831],
[ 0.8213, 0.3289],
[ 0.2020, 0.1772],
[-0.1760, -0.8098],
[ 0.7836, 0.6880],
[ 0.7268, 0.4244],
[ 0.8658, 0.3465],
[-0.4633, 0.7277],
[-0.6041, 0.3271],
[-0.1271, -0.7205],
[ 0.8001, -0.2382],
[-0.4161, 0.7591],
[ 0.5995, -0.9781],
[-0.2896, 0.6675],
[ 0.3702, -0.2581],
[-0.3622, 0.8973],
[ 0.2533, 0.2385],
[-0.9053, 0.6196],
[ 0.5036, 0.7443],
[-0.4245, 0.2807],
[-0.2366, 0.4772],
[-0.0633, 0.4845],
[ 0.3749, -0.9075],
[-0.3182, -0.2987],
[ 0.9127, 0.0938],
[-0.7642, 0.0669],
[ 0.0822, -0.0687],
[ 0.9339, 0.2288]])
[31]:
print(y)
tensor([[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000],
[0.5000]])
[32]:
import collections
[33]:
model_layers = collections.OrderedDict()
n_layers =100
for i in range(n_layers):
if i==0:
model_layers[f'linear_{i+1}']=torch.nn.Linear(2, 10)
model_layers[f'sigmoid_{i+1}']=torch.nn.Sigmoid()
else:
model_layers[f'linear_{i+1}']=torch.nn.Linear(10, 10)
model_layers[f'sigmoid_{i+1}']=torch.nn.Sigmoid()
# Use the nn package to define our model and loss function.
model = torch.nn.Sequential(
model_layers
)
loss_fn = torch.nn.MSELoss(reduction='mean')
[37]:
y_pred = model(x)
loss = loss_fn(y_pred, y)
[38]:
loss.backward()
for i in range(n_layers):
print(f'{i+1}-th layer:',model[i*2].weight.grad)
1-th layer: tensor([[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.],
[0., 0.]])
2-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
3-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
4-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
5-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
6-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
7-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
8-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
9-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
10-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
11-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
12-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
13-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
14-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
15-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
16-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
17-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
18-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
19-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
20-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
21-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
22-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
23-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
24-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
25-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
26-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
27-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
28-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
29-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
30-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
31-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
32-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
33-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
34-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
35-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
36-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
37-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
38-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
39-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
40-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
41-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
42-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
43-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
44-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
45-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
46-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
47-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
48-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
49-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
50-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
51-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
52-th layer: tensor([[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 0.]])
53-th layer: tensor([[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[-2.8026e-43, -0.0000e+00, -0.0000e+00, -2.8026e-43, -2.8026e-43,
-2.8026e-43, -2.8026e-43, -0.0000e+00, -0.0000e+00, -0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00]])
54-th layer: tensor([[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43,
-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43],
[ 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,
0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 0.0000e+00, 2.8026e-43,
0.0000e+00, 2.8026e-43, 0.0000e+00, 2.8026e-43, 2.8026e-43],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 0.0000e+00, 2.8026e-43,
0.0000e+00, 2.8026e-43, 0.0000e+00, 2.8026e-43, 2.8026e-43],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 0.0000e+00, 2.8026e-43,
0.0000e+00, 2.8026e-43, 0.0000e+00, 2.8026e-43, 2.8026e-43],
[-5.6052e-43, -5.6052e-43, -5.6052e-43, -2.8026e-43, -5.6052e-43,
-2.8026e-43, -5.6052e-43, -2.8026e-43, -5.6052e-43, -5.6052e-43],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43,
2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 0.0000e+00, 2.8026e-43,
0.0000e+00, 2.8026e-43, 0.0000e+00, 2.8026e-43, 2.8026e-43],
[-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43,
-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43]])
55-th layer: tensor([[-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43,
-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43],
[ 1.4013e-42, 1.6816e-42, 1.4013e-42, 2.2421e-42, 1.4013e-42,
2.2421e-42, 2.2421e-42, 1.6816e-42, 2.2421e-42, 1.6816e-42],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43,
2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43],
[-5.6052e-43, -5.6052e-43, -5.6052e-43, -8.4078e-43, -5.6052e-43,
-8.4078e-43, -8.4078e-43, -5.6052e-43, -8.4078e-43, -5.6052e-43],
[-1.9618e-42, -2.2421e-42, -1.9618e-42, -2.8026e-42, -1.9618e-42,
-2.8026e-42, -3.0829e-42, -2.2421e-42, -3.0829e-42, -2.5223e-42],
[-1.9618e-42, -2.2421e-42, -1.9618e-42, -2.8026e-42, -1.9618e-42,
-2.8026e-42, -3.0829e-42, -2.2421e-42, -3.0829e-42, -2.5223e-42],
[ 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43,
2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43, 2.8026e-43],
[-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43,
-2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43, -2.8026e-43],
[-2.2421e-42, -2.5223e-42, -2.2421e-42, -3.0829e-42, -2.2421e-42,
-3.0829e-42, -3.6434e-42, -2.5223e-42, -3.6434e-42, -2.8026e-42],
[-1.1210e-42, -1.4013e-42, -1.1210e-42, -1.6816e-42, -1.1210e-42,
-1.6816e-42, -1.6816e-42, -1.4013e-42, -1.6816e-42, -1.4013e-42]])
56-th layer: tensor([[-1.0370e-41, -1.0089e-41, -1.1771e-41, -9.8091e-42, -1.2051e-41,
-9.5288e-42, -7.5670e-42, -9.2486e-42, -8.6881e-42, -1.0089e-41],
[ 9.5288e-42, 9.5288e-42, 1.0930e-41, 9.2486e-42, 1.1210e-41,
8.6881e-42, 7.0065e-42, 8.6881e-42, 8.1275e-42, 9.2486e-42],
[-4.2039e-42, -3.9236e-42, -4.7644e-42, -3.9236e-42, -4.7644e-42,
-3.9236e-42, -3.0829e-42, -3.6434e-42, -3.3631e-42, -3.9236e-42],
[ 8.6881e-42, 8.6881e-42, 1.0089e-41, 8.4078e-42, 1.0370e-41,
8.1275e-42, 6.4460e-42, 7.8473e-42, 7.2868e-42, 8.6881e-42],
[-1.5134e-41, -1.4854e-41, -1.7376e-41, -1.4574e-41, -1.7656e-41,
-1.4013e-41, -1.1210e-41, -1.3733e-41, -1.2612e-41, -1.4854e-41],
[-1.0370e-41, -1.0370e-41, -1.2051e-41, -1.0089e-41, -1.2051e-41,
-9.5288e-42, -7.8473e-42, -9.5288e-42, -8.6881e-42, -1.0370e-41],
[-2.4663e-41, -2.4383e-41, -2.8306e-41, -2.3822e-41, -2.8586e-41,
-2.2701e-41, -1.8497e-41, -2.2421e-41, -2.0739e-41, -2.4102e-41],
[ 1.1491e-41, 1.1210e-41, 1.3172e-41, 1.1210e-41, 1.3172e-41,
1.0650e-41, 8.6881e-42, 1.0370e-41, 9.5288e-42, 1.1210e-41],
[-5.0447e-42, -5.0447e-42, -5.8855e-42, -5.0447e-42, -5.8855e-42,
-4.7644e-42, -3.9236e-42, -4.7644e-42, -4.2039e-42, -5.0447e-42],
[ 1.3172e-41, 1.2892e-41, 1.4854e-41, 1.2612e-41, 1.5134e-41,
1.2051e-41, 9.8091e-42, 1.1771e-41, 1.0930e-41, 1.2892e-41]])
57-th layer: tensor([[-2.8895e-40, -2.2869e-40, -2.5055e-40, -2.3766e-40, -2.3486e-40,
-2.3850e-40, -2.3346e-40, -2.4439e-40, -2.3402e-40, -2.3458e-40],
[ 1.5246e-40, 1.2051e-40, 1.3200e-40, 1.2528e-40, 1.2387e-40,
1.2584e-40, 1.2303e-40, 1.2892e-40, 1.2331e-40, 1.2359e-40],
[ 1.6591e-40, 1.3116e-40, 1.4377e-40, 1.3649e-40, 1.3480e-40,
1.3677e-40, 1.3396e-40, 1.4041e-40, 1.3424e-40, 1.3452e-40],
[-2.0347e-40, -1.6087e-40, -1.7628e-40, -1.6732e-40, -1.6535e-40,
-1.6788e-40, -1.6423e-40, -1.7208e-40, -1.6451e-40, -1.6507e-40],
[-1.3789e-40, -1.0902e-40, -1.1939e-40, -1.1351e-40, -1.1210e-40,
-1.1379e-40, -1.1126e-40, -1.1659e-40, -1.1154e-40, -1.1182e-40],
[-1.2387e-40, -9.7811e-41, -1.0734e-40, -1.0173e-40, -1.0061e-40,
-1.0201e-40, -1.0005e-40, -1.0454e-40, -1.0005e-40, -1.0033e-40],
[-7.0065e-42, -5.6052e-42, -6.1657e-42, -5.8855e-42, -5.6052e-42,
-5.8855e-42, -5.6052e-42, -5.8855e-42, -5.6052e-42, -5.6052e-42],
[ 9.5849e-41, 7.5950e-41, 8.3237e-41, 7.8753e-41, 7.7912e-41,
7.9033e-41, 7.7352e-41, 8.1275e-41, 7.7632e-41, 7.7912e-41],
[-2.8026e-41, -2.2141e-41, -2.4102e-41, -2.2981e-41, -2.2701e-41,
-2.2981e-41, -2.2701e-41, -2.3542e-41, -2.2701e-41, -2.2701e-41],
[-3.4472e-41, -2.7465e-41, -2.9988e-41, -2.8586e-41, -2.8026e-41,
-2.8586e-41, -2.8026e-41, -2.9147e-41, -2.8026e-41, -2.8026e-41]])
58-th layer: tensor([[-1.3046e-39, -1.6325e-39, -1.5367e-39, -1.2671e-39, -1.2699e-39,
-1.6149e-39, -1.5165e-39, -1.1897e-39, -1.3542e-39, -1.7354e-39],
[ 5.2409e-41, 6.5581e-41, 6.1657e-41, 5.1007e-41, 5.1007e-41,
6.5020e-41, 6.1097e-41, 4.7924e-41, 5.4370e-41, 6.9785e-41],
[ 2.6485e-40, 3.3127e-40, 3.1193e-40, 2.5700e-40, 2.5756e-40,
3.2762e-40, 3.0773e-40, 2.4130e-40, 2.7465e-40, 3.5201e-40],
[-3.2902e-40, -4.1170e-40, -3.8760e-40, -3.1950e-40, -3.2034e-40,
-4.0722e-40, -3.8255e-40, -3.0016e-40, -3.4164e-40, -4.3777e-40],
[ 5.3754e-40, 6.7234e-40, 6.3311e-40, 5.2184e-40, 5.2296e-40,
6.6534e-40, 6.2470e-40, 4.9017e-40, 5.5772e-40, 7.1494e-40],
[-1.9338e-41, -2.4383e-41, -2.2701e-41, -1.8777e-41, -1.8777e-41,
-2.4102e-41, -2.2421e-41, -1.7656e-41, -2.0179e-41, -2.5784e-41],
[-1.2208e-39, -1.5274e-39, -1.4377e-39, -1.1855e-39, -1.1880e-39,
-1.5109e-39, -1.4190e-39, -1.1132e-39, -1.2671e-39, -1.6235e-39],
[-4.4057e-40, -5.5127e-40, -5.1904e-40, -4.2796e-40, -4.2880e-40,
-5.4539e-40, -5.1231e-40, -4.0189e-40, -4.5738e-40, -5.8602e-40],
[-7.0177e-40, -8.7805e-40, -8.2649e-40, -6.8159e-40, -6.8299e-40,
-8.6852e-40, -8.1556e-40, -6.3983e-40, -7.2839e-40, -9.3326e-40],
[ 7.2559e-40, 9.0804e-40, 8.5479e-40, 7.0485e-40, 7.0625e-40,
8.9823e-40, 8.4358e-40, 6.6169e-40, 7.5334e-40, 9.6521e-40]])
59-th layer: tensor([[ 2.5610e-39, 2.4072e-39, 2.0069e-39, 2.4767e-39, 2.0097e-39,
2.4528e-39, 3.0383e-39, 2.2723e-39, 3.2132e-39, 2.2956e-39],
[-1.2320e-39, -1.1580e-39, -9.6549e-40, -1.1914e-39, -9.6662e-40,
-1.1799e-39, -1.4616e-39, -1.0930e-39, -1.5456e-39, -1.1042e-39],
[-5.2608e-39, -4.9446e-39, -4.1223e-39, -5.0873e-39, -4.1279e-39,
-5.0385e-39, -6.2414e-39, -4.6677e-39, -6.6007e-39, -4.7154e-39],
[ 8.0838e-39, 7.5978e-39, 6.3344e-39, 7.8173e-39, 6.3434e-39,
7.7422e-39, 9.5908e-39, 7.1727e-39, 1.0143e-38, 7.2458e-39],
[ 7.1889e-39, 6.7568e-39, 5.6332e-39, 6.9516e-39, 5.6411e-39,
6.8849e-39, 8.5289e-39, 6.3784e-39, 9.0196e-39, 6.4437e-39],
[-7.5452e-39, -7.0914e-39, -5.9121e-39, -7.2960e-39, -5.9205e-39,
-7.2259e-39, -8.9512e-39, -6.6946e-39, -9.4666e-39, -6.7629e-39],
[ 7.6631e-39, 7.2024e-39, 6.0048e-39, 7.4103e-39, 6.0133e-39,
7.3392e-39, 9.0913e-39, 6.7994e-39, 9.6149e-39, 6.8686e-39],
[-2.8483e-39, -2.6770e-39, -2.2320e-39, -2.7544e-39, -2.2351e-39,
-2.7280e-39, -3.3791e-39, -2.5271e-39, -3.5736e-39, -2.5532e-39],
[ 1.4204e-39, 1.3349e-39, 1.1129e-39, 1.3733e-39, 1.1143e-39,
1.3601e-39, 1.6849e-39, 1.2600e-39, 1.7819e-39, 1.2729e-39],
[-1.4153e-39, -1.3301e-39, -1.1090e-39, -1.3685e-39, -1.1104e-39,
-1.3553e-39, -1.6790e-39, -1.2556e-39, -1.7757e-39, -1.2685e-39]])
60-th layer: tensor([[-7.2163e-38, -9.1649e-38, -8.5942e-38, -7.5590e-38, -6.1810e-38,
-9.2917e-38, -6.6311e-38, -8.0736e-38, -9.5158e-38, -7.5757e-38],
[ 1.7525e-38, 2.2257e-38, 2.0871e-38, 1.8357e-38, 1.5011e-38,
2.2565e-38, 1.6104e-38, 1.9607e-38, 2.3109e-38, 1.8398e-38],
[-1.6666e-38, -2.1166e-38, -1.9848e-38, -1.7457e-38, -1.4275e-38,
-2.1459e-38, -1.5314e-38, -1.8645e-38, -2.1976e-38, -1.7496e-38],
[-3.7426e-38, -4.7531e-38, -4.4572e-38, -3.9203e-38, -3.2056e-38,
-4.8190e-38, -3.4391e-38, -4.1872e-38, -4.9352e-38, -3.9290e-38],
[-5.0432e-38, -6.4049e-38, -6.0061e-38, -5.2826e-38, -4.3196e-38,
-6.4936e-38, -4.6342e-38, -5.6423e-38, -6.6502e-38, -5.2943e-38],
[ 6.0984e-38, 7.7451e-38, 7.2628e-38, 6.3880e-38, 5.2235e-38,
7.8523e-38, 5.6038e-38, 6.8228e-38, 8.0417e-38, 6.4021e-38],
[-4.1691e-38, -5.2949e-38, -4.9652e-38, -4.3671e-38, -3.5710e-38,
-5.3682e-38, -3.8310e-38, -4.6644e-38, -5.4976e-38, -4.3768e-38],
[-6.6461e-38, -8.4407e-38, -7.9152e-38, -6.9617e-38, -5.6926e-38,
-8.5576e-38, -6.1071e-38, -7.4357e-38, -8.7640e-38, -6.9771e-38],
[-2.5964e-38, -3.2975e-38, -3.0922e-38, -2.7197e-38, -2.2239e-38,
-3.3432e-38, -2.3859e-38, -2.9049e-38, -3.4238e-38, -2.7257e-38],
[-2.9750e-38, -3.7783e-38, -3.5430e-38, -3.1162e-38, -2.5481e-38,
-3.8306e-38, -2.7337e-38, -3.3284e-38, -3.9229e-38, -3.1231e-38]])
61-th layer: tensor([[ 3.0521e-37, 4.2739e-37, 4.1745e-37, 3.9104e-37, 3.5448e-37,
2.6814e-37, 3.6426e-37, 3.4668e-37, 3.7770e-37, 3.6312e-37],
[ 2.0566e-37, 2.8799e-37, 2.8129e-37, 2.6349e-37, 2.3886e-37,
1.8068e-37, 2.4545e-37, 2.3360e-37, 2.5451e-37, 2.4468e-37],
[-4.5967e-37, -6.4368e-37, -6.2871e-37, -5.8893e-37, -5.3387e-37,
-4.0384e-37, -5.4860e-37, -5.2212e-37, -5.6885e-37, -5.4688e-37],
[ 1.9924e-37, 2.7900e-37, 2.7251e-37, 2.5526e-37, 2.3140e-37,
1.7504e-37, 2.3778e-37, 2.2631e-37, 2.4656e-37, 2.3704e-37],
[ 5.1638e-37, 7.2310e-37, 7.0628e-37, 6.6159e-37, 5.9974e-37,
4.5367e-37, 6.1628e-37, 5.8654e-37, 6.3903e-37, 6.1436e-37],
[-2.8356e-37, -3.9707e-37, -3.8783e-37, -3.6329e-37, -3.2933e-37,
-2.4912e-37, -3.3841e-37, -3.2208e-37, -3.5090e-37, -3.3736e-37],
[-3.1893e-37, -4.4660e-37, -4.3621e-37, -4.0861e-37, -3.7040e-37,
-2.8019e-37, -3.8062e-37, -3.6225e-37, -3.9467e-37, -3.7943e-37],
[ 2.5305e-37, 3.5435e-37, 3.4611e-37, 3.2421e-37, 2.9390e-37,
2.2232e-37, 3.0200e-37, 2.8743e-37, 3.1315e-37, 3.0106e-37],
[ 3.1863e-37, 4.4619e-37, 4.3581e-37, 4.0823e-37, 3.7007e-37,
2.7993e-37, 3.8027e-37, 3.6192e-37, 3.9431e-37, 3.7909e-37],
[-1.3384e-37, -1.8742e-37, -1.8306e-37, -1.7147e-37, -1.5544e-37,
-1.1758e-37, -1.5973e-37, -1.5202e-37, -1.6563e-37, -1.5923e-37]])
62-th layer: tensor([[-6.0479e-36, -4.6008e-36, -5.8828e-36, -4.6576e-36, -3.6275e-36,
-2.8739e-36, -4.7922e-36, -3.8203e-36, -4.5513e-36, -5.3218e-36],
[-6.9698e-36, -5.3022e-36, -6.7796e-36, -5.3676e-36, -4.1805e-36,
-3.3120e-36, -5.5228e-36, -4.4027e-36, -5.2452e-36, -6.1331e-36],
[ 2.5342e-36, 1.9279e-36, 2.4651e-36, 1.9517e-36, 1.5200e-36,
1.2042e-36, 2.0081e-36, 1.6008e-36, 1.9071e-36, 2.2300e-36],
[-3.0489e-36, -2.3194e-36, -2.9656e-36, -2.3480e-36, -1.8287e-36,
-1.4488e-36, -2.4159e-36, -1.9259e-36, -2.2944e-36, -2.6829e-36],
[ 2.0166e-36, 1.5341e-36, 1.9615e-36, 1.5530e-36, 1.2095e-36,
9.5825e-37, 1.5979e-36, 1.2738e-36, 1.5176e-36, 1.7745e-36],
[ 2.2894e-36, 1.7416e-36, 2.2269e-36, 1.7631e-36, 1.3732e-36,
1.0879e-36, 1.8141e-36, 1.4462e-36, 1.7229e-36, 2.0145e-36],
[-4.8950e-37, -3.7237e-37, -4.7613e-37, -3.7697e-37, -2.9360e-37,
-2.3260e-37, -3.8787e-37, -3.0921e-37, -3.6837e-37, -4.3073e-37],
[-5.0127e-36, -3.8133e-36, -4.8759e-36, -3.8604e-36, -3.0067e-36,
-2.3820e-36, -3.9720e-36, -3.1664e-36, -3.7723e-36, -4.4109e-36],
[ 3.5275e-36, 2.6835e-36, 3.4312e-36, 2.7166e-36, 2.1158e-36,
1.6763e-36, 2.7952e-36, 2.2283e-36, 2.6547e-36, 3.1041e-36],
[ 3.4011e-36, 2.5873e-36, 3.3083e-36, 2.6193e-36, 2.0400e-36,
1.6162e-36, 2.6950e-36, 2.1484e-36, 2.5595e-36, 2.9928e-36]])
63-th layer: tensor([[ 8.1517e-36, 9.1610e-36, 1.2544e-35, 1.2432e-35, 9.6349e-36,
9.5161e-36, 9.7820e-36, 9.2577e-36, 7.8680e-36, 9.9181e-36],
[ 1.0122e-35, 1.1375e-35, 1.5575e-35, 1.5436e-35, 1.1964e-35,
1.1816e-35, 1.2146e-35, 1.1495e-35, 9.7696e-36, 1.2315e-35],
[-6.7773e-37, -7.6164e-37, -1.0429e-36, -1.0336e-36, -8.0105e-37,
-7.9117e-37, -8.1327e-37, -7.6968e-37, -6.5414e-37, -8.2459e-37],
[ 3.4791e-35, 3.9099e-35, 5.3536e-35, 5.3058e-35, 4.1122e-35,
4.0614e-35, 4.1749e-35, 3.9512e-35, 3.3580e-35, 4.2330e-35],
[ 1.1475e-35, 1.2896e-35, 1.7657e-35, 1.7499e-35, 1.3563e-35,
1.3395e-35, 1.3770e-35, 1.3032e-35, 1.1075e-35, 1.3961e-35],
[ 4.2723e-35, 4.8013e-35, 6.5741e-35, 6.5154e-35, 5.0497e-35,
4.9874e-35, 5.1267e-35, 4.8519e-35, 4.1236e-35, 5.1981e-35],
[-1.5158e-35, -1.7034e-35, -2.3324e-35, -2.3116e-35, -1.7916e-35,
-1.7695e-35, -1.8189e-35, -1.7214e-35, -1.4630e-35, -1.8442e-35],
[-9.0697e-37, -1.0193e-36, -1.3956e-36, -1.3832e-36, -1.0720e-36,
-1.0588e-36, -1.0884e-36, -1.0300e-36, -8.7540e-37, -1.1035e-36],
[-2.3317e-35, -2.6204e-35, -3.5880e-35, -3.5559e-35, -2.7559e-35,
-2.7220e-35, -2.7980e-35, -2.6480e-35, -2.2505e-35, -2.8370e-35],
[ 1.8871e-36, 2.1208e-36, 2.9039e-36, 2.8779e-36, 2.2305e-36,
2.2030e-36, 2.2645e-36, 2.1432e-36, 1.8214e-36, 2.2961e-36]])
64-th layer: tensor([[ 1.2025e-34, 2.3151e-34, 2.5629e-34, 1.8183e-34, 2.2574e-34,
2.1077e-34, 2.0190e-34, 1.9282e-34, 2.7084e-34, 1.4375e-34],
[-8.7061e-35, -1.6762e-34, -1.8556e-34, -1.3165e-34, -1.6344e-34,
-1.5260e-34, -1.4618e-34, -1.3961e-34, -1.9609e-34, -1.0408e-34],
[ 1.0490e-36, 2.0196e-36, 2.2357e-36, 1.5862e-36, 1.9693e-36,
1.8387e-36, 1.7613e-36, 1.6821e-36, 2.3627e-36, 1.2540e-36],
[-5.1030e-35, -9.8248e-35, -1.0876e-34, -7.7164e-35, -9.5799e-35,
-8.9447e-35, -8.5683e-35, -8.1828e-35, -1.1494e-34, -6.1003e-35],
[-9.2706e-35, -1.7849e-34, -1.9759e-34, -1.4018e-34, -1.7404e-34,
-1.6250e-34, -1.5566e-34, -1.4866e-34, -2.0881e-34, -1.1082e-34],
[-1.0830e-34, -2.0851e-34, -2.3082e-34, -1.6376e-34, -2.0331e-34,
-1.8983e-34, -1.8184e-34, -1.7366e-34, -2.4393e-34, -1.2947e-34],
[ 1.4726e-34, 2.8353e-34, 3.1387e-34, 2.2268e-34, 2.7646e-34,
2.5813e-34, 2.4727e-34, 2.3615e-34, 3.3169e-34, 1.7605e-34],
[ 1.2155e-34, 2.3403e-34, 2.5907e-34, 1.8380e-34, 2.2819e-34,
2.1306e-34, 2.0410e-34, 1.9492e-34, 2.7378e-34, 1.4531e-34],
[-5.9990e-35, -1.1550e-34, -1.2786e-34, -9.0714e-35, -1.1262e-34,
-1.0515e-34, -1.0073e-34, -9.6198e-35, -1.3512e-34, -7.1715e-35],
[ 5.0219e-35, 9.6686e-35, 1.0703e-34, 7.5938e-35, 9.4276e-35,
8.8026e-35, 8.4321e-35, 8.0528e-35, 1.1311e-34, 6.0034e-35]])
65-th layer: tensor([[-2.4549e-33, -2.1396e-33, -2.4054e-33, -2.7345e-33, -1.8059e-33,
-1.8800e-33, -2.0739e-33, -2.0836e-33, -2.7101e-33, -2.4347e-33],
[-9.1564e-34, -7.9803e-34, -8.9718e-34, -1.0199e-33, -6.7355e-34,
-7.0119e-34, -7.7351e-34, -7.7716e-34, -1.0108e-33, -9.0810e-34],
[ 1.3029e-33, 1.1355e-33, 1.2766e-33, 1.4512e-33, 9.5840e-34,
9.9772e-34, 1.1006e-33, 1.1058e-33, 1.4383e-33, 1.2921e-33],
[-6.5575e-34, -5.7152e-34, -6.4254e-34, -7.3042e-34, -4.8238e-34,
-5.0217e-34, -5.5396e-34, -5.5658e-34, -7.2393e-34, -6.5036e-34],
[ 1.1182e-33, 9.7460e-34, 1.0957e-33, 1.2456e-33, 8.2258e-34,
8.5634e-34, 9.4466e-34, 9.4912e-34, 1.2345e-33, 1.1090e-33],
[-1.0874e-33, -9.4774e-34, -1.0655e-33, -1.2112e-33, -7.9991e-34,
-8.3273e-34, -9.1862e-34, -9.2295e-34, -1.2005e-33, -1.0785e-33],
[-2.6796e-33, -2.3355e-33, -2.6256e-33, -2.9848e-33, -1.9712e-33,
-2.0521e-33, -2.2637e-33, -2.2744e-33, -2.9582e-33, -2.6576e-33],
[-1.3695e-33, -1.1936e-33, -1.3419e-33, -1.5254e-33, -1.0074e-33,
-1.0487e-33, -1.1569e-33, -1.1624e-33, -1.5119e-33, -1.3582e-33],
[ 2.4585e-33, 2.1427e-33, 2.4090e-33, 2.7385e-33, 1.8085e-33,
1.8827e-33, 2.0769e-33, 2.0867e-33, 2.7141e-33, 2.4383e-33],
[ 5.8617e-34, 5.1087e-34, 5.7435e-34, 6.5291e-34, 4.3119e-34,
4.4888e-34, 4.9518e-34, 4.9752e-34, 6.4711e-34, 5.8134e-34]])
66-th layer: tensor([[-3.9977e-33, -5.0687e-33, -4.5106e-33, -4.0393e-33, -3.2233e-33,
-4.7100e-33, -4.4990e-33, -3.8970e-33, -3.9408e-33, -4.1827e-33],
[ 1.2037e-32, 1.5262e-32, 1.3582e-32, 1.2162e-32, 9.7055e-33,
1.4182e-32, 1.3547e-32, 1.1734e-32, 1.1866e-32, 1.2594e-32],
[ 2.7554e-34, 3.4937e-34, 3.1090e-34, 2.7841e-34, 2.2217e-34,
3.2464e-34, 3.1010e-34, 2.6861e-34, 2.7162e-34, 2.8830e-34],
[-1.0854e-32, -1.3762e-32, -1.2247e-32, -1.0967e-32, -8.7518e-33,
-1.2788e-32, -1.2216e-32, -1.0581e-32, -1.0700e-32, -1.1357e-32],
[-3.0413e-33, -3.8561e-33, -3.4315e-33, -3.0729e-33, -2.4522e-33,
-3.5832e-33, -3.4227e-33, -2.9647e-33, -2.9980e-33, -3.1820e-33],
[-6.2242e-33, -7.8918e-33, -7.0229e-33, -6.2890e-33, -5.0185e-33,
-7.3332e-33, -7.0048e-33, -6.0675e-33, -6.1356e-33, -6.5123e-33],
[ 5.5354e-33, 7.0184e-33, 6.2457e-33, 5.5930e-33, 4.4631e-33,
6.5217e-33, 6.2295e-33, 5.3960e-33, 5.4566e-33, 5.7915e-33],
[-2.0540e-32, -2.6043e-32, -2.3176e-32, -2.0754e-32, -1.6561e-32,
-2.4200e-32, -2.3116e-32, -2.0023e-32, -2.0248e-32, -2.1491e-32],
[-1.9153e-32, -2.4285e-32, -2.1611e-32, -1.9353e-32, -1.5443e-32,
-2.2566e-32, -2.1555e-32, -1.8671e-32, -1.8881e-32, -2.0040e-32],
[ 2.9266e-33, 3.7107e-33, 3.3022e-33, 2.9571e-33, 2.3597e-33,
3.4481e-33, 3.2937e-33, 2.8529e-33, 2.8850e-33, 3.0621e-33]])
67-th layer: tensor([[ 9.2513e-33, 7.8830e-33, 7.8949e-33, 1.1779e-32, 8.5238e-33,
9.3295e-33, 1.0580e-32, 1.0528e-32, 8.7628e-33, 1.0440e-32],
[-7.5041e-33, -6.3942e-33, -6.4039e-33, -9.5542e-33, -6.9139e-33,
-7.5675e-33, -8.5817e-33, -8.5398e-33, -7.1079e-33, -8.4685e-33],
[-7.9818e-32, -6.8013e-32, -6.8115e-32, -1.0162e-31, -7.3541e-32,
-8.0493e-32, -9.1280e-32, -9.0834e-32, -7.5603e-32, -9.0076e-32],
[ 1.1520e-31, 9.8159e-32, 9.8307e-32, 1.4667e-31, 1.0614e-31,
1.1617e-31, 1.3174e-31, 1.3110e-31, 1.0911e-31, 1.3000e-31],
[-1.3628e-31, -1.1612e-31, -1.1630e-31, -1.7351e-31, -1.2556e-31,
-1.3743e-31, -1.5585e-31, -1.5509e-31, -1.2908e-31, -1.5379e-31],
[ 1.1972e-31, 1.0201e-31, 1.0217e-31, 1.5242e-31, 1.1030e-31,
1.2073e-31, 1.3691e-31, 1.3624e-31, 1.1340e-31, 1.3510e-31],
[-5.9780e-32, -5.0938e-32, -5.1015e-32, -7.6112e-32, -5.5079e-32,
-6.0285e-32, -6.8364e-32, -6.8031e-32, -5.6624e-32, -6.7463e-32],
[ 3.7669e-32, 3.2098e-32, 3.2146e-32, 4.7960e-32, 3.4707e-32,
3.7988e-32, 4.3079e-32, 4.2868e-32, 3.5680e-32, 4.2510e-32],
[ 1.7183e-32, 1.4642e-32, 1.4664e-32, 2.1878e-32, 1.5832e-32,
1.7329e-32, 1.9651e-32, 1.9555e-32, 1.6276e-32, 1.9392e-32],
[ 1.0128e-31, 8.6303e-32, 8.6433e-32, 1.2895e-31, 9.3318e-32,
1.0214e-31, 1.1583e-31, 1.1526e-31, 9.5935e-32, 1.1430e-31]])
68-th layer: tensor([[-9.4977e-31, -8.2370e-31, -6.7312e-31, -5.5722e-31, -4.9661e-31,
-4.0336e-31, -6.2315e-31, -5.3176e-31, -5.9240e-31, -6.1116e-31],
[-1.0164e-30, -8.8153e-31, -7.2037e-31, -5.9634e-31, -5.3147e-31,
-4.3167e-31, -6.6690e-31, -5.6909e-31, -6.3398e-31, -6.5407e-31],
[ 8.6152e-31, 7.4717e-31, 6.1058e-31, 5.0545e-31, 4.5046e-31,
3.6588e-31, 5.6525e-31, 4.8235e-31, 5.3735e-31, 5.5438e-31],
[-4.6700e-31, -4.0501e-31, -3.3097e-31, -2.7398e-31, -2.4418e-31,
-1.9833e-31, -3.0640e-31, -2.6146e-31, -2.9128e-31, -3.0051e-31],
[ 1.0135e-30, 8.7893e-31, 7.1825e-31, 5.9459e-31, 5.2990e-31,
4.3040e-31, 6.6494e-31, 5.6741e-31, 6.3211e-31, 6.5214e-31],
[-3.3468e-31, -2.9025e-31, -2.3719e-31, -1.9635e-31, -1.7499e-31,
-1.4213e-31, -2.1958e-31, -1.8738e-31, -2.0875e-31, -2.1536e-31],
[ 1.4614e-31, 1.2674e-31, 1.0357e-31, 8.5740e-32, 7.6413e-32,
6.2065e-32, 9.5885e-32, 8.1822e-32, 9.1152e-32, 9.4040e-32],
[ 6.0813e-31, 5.2741e-31, 4.3099e-31, 3.5679e-31, 3.1797e-31,
2.5827e-31, 3.9900e-31, 3.4048e-31, 3.7931e-31, 3.9132e-31],
[-9.8811e-32, -8.5696e-32, -7.0029e-32, -5.7972e-32, -5.1665e-32,
-4.1964e-32, -6.4831e-32, -5.5322e-32, -6.1631e-32, -6.3584e-32],
[ 1.9211e-32, 1.6661e-32, 1.3615e-32, 1.1271e-32, 1.0045e-32,
8.1585e-33, 1.2604e-32, 1.0756e-32, 1.1982e-32, 1.2362e-32]])
69-th layer: tensor([[-1.9098e-30, -1.5509e-30, -1.8639e-30, -1.6007e-30, -1.7738e-30,
-1.3558e-30, -1.6076e-30, -1.7207e-30, -1.4763e-30, -1.7882e-30],
[ 8.3063e-30, 6.7450e-30, 8.1062e-30, 6.9619e-30, 7.7147e-30,
5.8966e-30, 6.9920e-30, 7.4837e-30, 6.4208e-30, 7.7773e-30],
[ 2.1213e-31, 1.7226e-31, 2.0702e-31, 1.7780e-31, 1.9702e-31,
1.5059e-31, 1.7857e-31, 1.9113e-31, 1.6398e-31, 1.9862e-31],
[-4.6340e-30, -3.7630e-30, -4.5224e-30, -3.8840e-30, -4.3040e-30,
-3.2897e-30, -3.9008e-30, -4.1751e-30, -3.5821e-30, -4.3389e-30],
[ 3.7470e-30, 3.0428e-30, 3.6568e-30, 3.1406e-30, 3.4802e-30,
2.6600e-30, 3.1542e-30, 3.3760e-30, 2.8965e-30, 3.5084e-30],
[-4.5342e-30, -3.6819e-30, -4.4250e-30, -3.8003e-30, -4.2113e-30,
-3.2188e-30, -3.8167e-30, -4.0852e-30, -3.5049e-30, -4.2454e-30],
[-6.3182e-31, -5.1307e-31, -6.1661e-31, -5.2956e-31, -5.8683e-31,
-4.4853e-31, -5.3185e-31, -5.6925e-31, -4.8840e-31, -5.9158e-31],
[ 8.4945e-31, 6.8979e-31, 8.2900e-31, 7.1196e-31, 7.8895e-31,
6.0303e-31, 7.1504e-31, 7.6533e-31, 6.5663e-31, 7.9535e-31],
[-1.0574e-30, -8.5868e-31, -1.0320e-30, -8.8628e-31, -9.8212e-31,
-7.5067e-31, -8.9011e-31, -9.5271e-31, -8.1740e-31, -9.9008e-31],
[ 2.4445e-30, 1.9850e-30, 2.3856e-30, 2.0488e-30, 2.2704e-30,
1.7353e-30, 2.0577e-30, 2.2024e-30, 1.8896e-30, 2.2888e-30]])
70-th layer: tensor([[-1.6422e-29, -1.9691e-29, -1.7418e-29, -1.5591e-29, -1.5552e-29,
-1.5642e-29, -1.7851e-29, -1.0771e-29, -1.6419e-29, -1.7532e-29],
[-1.8337e-29, -2.1986e-29, -1.9449e-29, -1.7409e-29, -1.7366e-29,
-1.7466e-29, -1.9933e-29, -1.2027e-29, -1.8333e-29, -1.9576e-29],
[ 6.1899e-29, 7.4219e-29, 6.5654e-29, 5.8768e-29, 5.8621e-29,
5.8960e-29, 6.7286e-29, 4.0600e-29, 6.1887e-29, 6.6082e-29],
[ 2.3835e-29, 2.8579e-29, 2.5281e-29, 2.2629e-29, 2.2573e-29,
2.2703e-29, 2.5909e-29, 1.5633e-29, 2.3831e-29, 2.5446e-29],
[-6.4945e-30, -7.7870e-30, -6.8884e-30, -6.1659e-30, -6.1505e-30,
-6.1861e-30, -7.0597e-30, -4.2597e-30, -6.4932e-30, -6.9333e-30],
[-3.2207e-29, -3.8617e-29, -3.4160e-29, -3.0577e-29, -3.0501e-29,
-3.0678e-29, -3.5010e-29, -2.1124e-29, -3.2200e-29, -3.4383e-29],
[ 4.9894e-29, 5.9824e-29, 5.2920e-29, 4.7370e-29, 4.7251e-29,
4.7525e-29, 5.4236e-29, 3.2725e-29, 4.9884e-29, 5.3265e-29],
[ 1.5505e-29, 1.8590e-29, 1.6445e-29, 1.4720e-29, 1.4683e-29,
1.4769e-29, 1.6854e-29, 1.0169e-29, 1.5502e-29, 1.6552e-29],
[-3.5674e-30, -4.2773e-30, -3.7837e-30, -3.3869e-30, -3.3784e-30,
-3.3980e-30, -3.8778e-30, -2.3398e-30, -3.5667e-30, -3.8084e-30],
[-2.6719e-29, -3.2036e-29, -2.8339e-29, -2.5367e-29, -2.5304e-29,
-2.5450e-29, -2.9044e-29, -1.7525e-29, -2.6713e-29, -2.8524e-29]])
71-th layer: tensor([[ 1.8351e-28, 2.4312e-28, 2.2969e-28, 1.8620e-28, 2.3568e-28,
2.2513e-28, 2.6343e-28, 1.7291e-28, 1.7982e-28, 2.4929e-28],
[-8.4116e-29, -1.1144e-28, -1.0528e-28, -8.5346e-29, -1.0803e-28,
-1.0319e-28, -1.2075e-28, -7.9257e-29, -8.2422e-29, -1.1427e-28],
[-7.7593e-31, -1.0280e-30, -9.7116e-31, -7.8727e-31, -9.9652e-31,
-9.5189e-31, -1.1138e-30, -7.3110e-31, -7.6030e-31, -1.0541e-30],
[ 2.6567e-28, 3.5197e-28, 3.3252e-28, 2.6956e-28, 3.4120e-28,
3.2592e-28, 3.8136e-28, 2.5032e-28, 2.6032e-28, 3.6090e-28],
[-4.2160e-29, -5.5855e-29, -5.2768e-29, -4.2777e-29, -5.4146e-29,
-5.1721e-29, -6.0519e-29, -3.9725e-29, -4.1311e-29, -5.7273e-29],
[ 5.4837e-29, 7.2650e-29, 6.8634e-29, 5.5639e-29, 7.0427e-29,
6.7273e-29, 7.8716e-29, 5.1669e-29, 5.3733e-29, 7.4494e-29],
[-7.9410e-29, -1.0521e-28, -9.9390e-29, -8.0571e-29, -1.0199e-28,
-9.7419e-29, -1.1399e-28, -7.4823e-29, -7.7811e-29, -1.0788e-28],
[-4.0689e-29, -5.3907e-29, -5.0927e-29, -4.1284e-29, -5.2257e-29,
-4.9917e-29, -5.8408e-29, -3.8339e-29, -3.9870e-29, -5.5275e-29],
[ 2.0772e-28, 2.7520e-28, 2.5999e-28, 2.1076e-28, 2.6678e-28,
2.5483e-28, 2.9818e-28, 1.9572e-28, 2.0354e-28, 2.8218e-28],
[-1.9189e-28, -2.5422e-28, -2.4017e-28, -1.9469e-28, -2.4644e-28,
-2.3540e-28, -2.7545e-28, -1.8080e-28, -1.8802e-28, -2.6067e-28]])
72-th layer: tensor([[-2.5963e-27, -1.6708e-27, -2.6782e-27, -2.7544e-27, -2.7319e-27,
-1.8756e-27, -2.7171e-27, -2.5418e-27, -1.7303e-27, -2.3342e-27],
[-1.6609e-27, -1.0688e-27, -1.7133e-27, -1.7620e-27, -1.7476e-27,
-1.1998e-27, -1.7382e-27, -1.6260e-27, -1.1069e-27, -1.4932e-27],
[ 1.7727e-27, 1.1408e-27, 1.8287e-27, 1.8806e-27, 1.8653e-27,
1.2807e-27, 1.8552e-27, 1.7355e-27, 1.1814e-27, 1.5938e-27],
[-5.3058e-28, -3.4145e-28, -5.4733e-28, -5.6288e-28, -5.5829e-28,
-3.8331e-28, -5.5527e-28, -5.1944e-28, -3.5361e-28, -4.7703e-28],
[-5.3762e-29, -3.4598e-29, -5.5458e-29, -5.7034e-29, -5.6569e-29,
-3.8839e-29, -5.6264e-29, -5.2633e-29, -3.5830e-29, -4.8335e-29],
[ 2.5027e-28, 1.6106e-28, 2.5816e-28, 2.6550e-28, 2.6333e-28,
1.8080e-28, 2.6191e-28, 2.4501e-28, 1.6679e-28, 2.2500e-28],
[-1.1360e-27, -7.3108e-28, -1.1719e-27, -1.2052e-27, -1.1953e-27,
-8.2069e-28, -1.1889e-27, -1.1122e-27, -7.5711e-28, -1.0214e-27],
[ 1.3894e-27, 8.9415e-28, 1.4333e-27, 1.4740e-27, 1.4620e-27,
1.0038e-27, 1.4541e-27, 1.3603e-27, 9.2599e-28, 1.2492e-27],
[ 1.1858e-27, 7.6312e-28, 1.2232e-27, 1.2580e-27, 1.2477e-27,
8.5666e-28, 1.2410e-27, 1.1609e-27, 7.9029e-28, 1.0661e-27],
[-1.0186e-27, -6.5554e-28, -1.0508e-27, -1.0807e-27, -1.0718e-27,
-7.3589e-28, -1.0660e-27, -9.9726e-28, -6.7888e-28, -9.1582e-28]])
73-th layer: tensor([[-2.9849e-27, -4.6218e-27, -3.4407e-27, -3.6334e-27, -4.5626e-27,
-3.7353e-27, -4.0818e-27, -3.0661e-27, -3.9911e-27, -3.6236e-27],
[-7.9554e-27, -1.2318e-26, -9.1702e-27, -9.6839e-27, -1.2160e-26,
-9.9553e-27, -1.0879e-26, -8.1719e-27, -1.0637e-26, -9.6577e-27],
[ 1.0228e-26, 1.5837e-26, 1.1790e-26, 1.2450e-26, 1.5634e-26,
1.2799e-26, 1.3987e-26, 1.0506e-26, 1.3676e-26, 1.2417e-26],
[ 4.6424e-27, 7.1883e-27, 5.3513e-27, 5.6511e-27, 7.0963e-27,
5.8095e-27, 6.3485e-27, 4.7688e-27, 6.2073e-27, 5.6358e-27],
[ 7.7098e-29, 1.1938e-28, 8.8871e-29, 9.3849e-29, 1.1785e-28,
9.6479e-29, 1.0543e-28, 7.9196e-29, 1.0309e-28, 9.3596e-29],
[ 1.5379e-26, 2.3813e-26, 1.7728e-26, 1.8721e-26, 2.3509e-26,
1.9246e-26, 2.1031e-26, 1.5798e-26, 2.0564e-26, 1.8670e-26],
[-9.0731e-28, -1.4049e-27, -1.0459e-27, -1.1044e-27, -1.3869e-27,
-1.1354e-27, -1.2407e-27, -9.3200e-28, -1.2132e-27, -1.1015e-27],
[-6.0908e-27, -9.4310e-27, -7.0209e-27, -7.4142e-27, -9.3102e-27,
-7.6220e-27, -8.3292e-27, -6.2566e-27, -8.1439e-27, -7.3942e-27],
[-1.5623e-27, -2.4190e-27, -1.8008e-27, -1.9017e-27, -2.3880e-27,
-1.9550e-27, -2.1364e-27, -1.6048e-27, -2.0889e-27, -1.8966e-27],
[ 6.3394e-28, 9.8159e-28, 7.3074e-28, 7.7168e-28, 9.6902e-28,
7.9331e-28, 8.6691e-28, 6.5119e-28, 8.4763e-28, 7.6959e-28]])
74-th layer: tensor([[-5.8296e-26, -5.6433e-26, -4.4268e-26, -4.8356e-26, -3.5382e-26,
-5.2760e-26, -4.1313e-26, -5.5997e-26, -3.8039e-26, -4.2175e-26],
[ 6.4583e-26, 6.2519e-26, 4.9042e-26, 5.3571e-26, 3.9198e-26,
5.8450e-26, 4.5769e-26, 6.2036e-26, 4.2141e-26, 4.6723e-26],
[ 7.1047e-26, 6.8776e-26, 5.3950e-26, 5.8932e-26, 4.3121e-26,
6.4300e-26, 5.0349e-26, 6.8245e-26, 4.6359e-26, 5.1399e-26],
[-1.8563e-25, -1.7970e-25, -1.4096e-25, -1.5398e-25, -1.1267e-25,
-1.6800e-25, -1.3155e-25, -1.7831e-25, -1.2113e-25, -1.3430e-25],
[ 9.1204e-26, 8.8289e-26, 6.9257e-26, 7.5653e-26, 5.5355e-26,
8.2543e-26, 6.4635e-26, 8.7607e-26, 5.9512e-26, 6.5983e-26],
[ 7.1547e-26, 6.9260e-26, 5.4330e-26, 5.9347e-26, 4.3424e-26,
6.4752e-26, 5.0704e-26, 6.8725e-26, 4.6685e-26, 5.1761e-26],
[ 2.2147e-25, 2.1440e-25, 1.6818e-25, 1.8371e-25, 1.3442e-25,
2.0044e-25, 1.5695e-25, 2.1274e-25, 1.4451e-25, 1.6023e-25],
[-1.1721e-25, -1.1346e-25, -8.9002e-26, -9.7221e-26, -7.1137e-26,
-1.0608e-25, -8.3062e-26, -1.1258e-25, -7.6479e-26, -8.4794e-26],
[-7.4820e-27, -7.2429e-27, -5.6816e-27, -6.2063e-27, -4.5411e-27,
-6.7715e-27, -5.3024e-27, -7.1870e-27, -4.8821e-27, -5.4130e-27],
[ 1.1505e-26, 1.1137e-26, 8.7365e-27, 9.5432e-27, 6.9828e-27,
1.0412e-26, 8.1534e-27, 1.1051e-26, 7.5072e-27, 8.3234e-27]])
75-th layer: tensor([[-4.8955e-25, -4.1808e-25, -6.2285e-25, -5.8004e-25, -5.2754e-25,
-6.0802e-25, -4.1492e-25, -4.4193e-25, -4.4673e-25, -4.0057e-25],
[ 8.7515e-25, 7.4738e-25, 1.1134e-24, 1.0369e-24, 9.4305e-25,
1.0869e-24, 7.4173e-25, 7.9002e-25, 7.9860e-25, 7.1609e-25],
[ 2.8280e-25, 2.4151e-25, 3.5980e-25, 3.3507e-25, 3.0474e-25,
3.5124e-25, 2.3969e-25, 2.5529e-25, 2.5807e-25, 2.3140e-25],
[ 1.0082e-24, 8.6096e-25, 1.2826e-24, 1.1945e-24, 1.0864e-24,
1.2521e-24, 8.5446e-25, 9.1008e-25, 9.1997e-25, 8.2492e-25],
[-1.3169e-25, -1.1246e-25, -1.6755e-25, -1.5603e-25, -1.4191e-25,
-1.6356e-25, -1.1161e-25, -1.1888e-25, -1.2017e-25, -1.0776e-25],
[-4.5740e-26, -3.9062e-26, -5.8194e-26, -5.4194e-26, -4.9289e-26,
-5.6808e-26, -3.8767e-26, -4.1291e-26, -4.1739e-26, -3.7427e-26],
[ 2.7616e-25, 2.3584e-25, 3.5135e-25, 3.2720e-25, 2.9758e-25,
3.4298e-25, 2.3406e-25, 2.4929e-25, 2.5200e-25, 2.2596e-25],
[ 7.1648e-25, 6.1187e-25, 9.1156e-25, 8.4891e-25, 7.7207e-25,
8.8985e-25, 6.0725e-25, 6.4678e-25, 6.5381e-25, 5.8626e-25],
[-3.7395e-25, -3.1935e-25, -4.7577e-25, -4.4307e-25, -4.0296e-25,
-4.6444e-25, -3.1694e-25, -3.3757e-25, -3.4124e-25, -3.0598e-25],
[ 5.8042e-25, 4.9568e-25, 7.3846e-25, 6.8770e-25, 6.2546e-25,
7.2088e-25, 4.9194e-25, 5.2396e-25, 5.2965e-25, 4.7493e-25]])
76-th layer: tensor([[-4.5859e-24, -5.9994e-24, -5.7456e-24, -4.7470e-24, -6.4025e-24,
-5.1995e-24, -5.1117e-24, -4.7575e-24, -4.5596e-24, -5.5471e-24],
[ 7.8347e-25, 1.0249e-24, 9.8160e-25, 8.1099e-25, 1.0938e-24,
8.8830e-25, 8.7330e-25, 8.1279e-25, 7.7898e-25, 9.4767e-25],
[-1.3024e-24, -1.7039e-24, -1.6318e-24, -1.3482e-24, -1.8184e-24,
-1.4767e-24, -1.4518e-24, -1.3512e-24, -1.2950e-24, -1.5754e-24],
[-2.9678e-25, -3.8826e-25, -3.7184e-25, -3.0721e-25, -4.1435e-25,
-3.3650e-25, -3.3081e-25, -3.0789e-25, -2.9508e-25, -3.5899e-25],
[ 7.7096e-24, 1.0086e-23, 9.6592e-24, 7.9804e-24, 1.0763e-23,
8.7412e-24, 8.5935e-24, 7.9981e-24, 7.6654e-24, 9.3254e-24],
[-3.6695e-24, -4.8005e-24, -4.5974e-24, -3.7984e-24, -5.1230e-24,
-4.1605e-24, -4.0902e-24, -3.8068e-24, -3.6484e-24, -4.4385e-24],
[-5.3489e-24, -6.9975e-24, -6.7016e-24, -5.5368e-24, -7.4677e-24,
-6.0646e-24, -5.9622e-24, -5.5491e-24, -5.3182e-24, -6.4700e-24],
[ 2.2351e-24, 2.9240e-24, 2.8003e-24, 2.3136e-24, 3.1204e-24,
2.5342e-24, 2.4914e-24, 2.3187e-24, 2.2223e-24, 2.7035e-24],
[-5.7483e-24, -7.5201e-24, -7.2020e-24, -5.9502e-24, -8.0253e-24,
-6.5175e-24, -6.4074e-24, -5.9635e-24, -5.7154e-24, -6.9531e-24],
[ 4.1379e-24, 5.4133e-24, 5.1844e-24, 4.2833e-24, 5.7770e-24,
4.6916e-24, 4.6124e-24, 4.2928e-24, 4.1142e-24, 5.0052e-24]])
77-th layer: tensor([[ 8.4080e-24, 8.5656e-24, 6.9463e-24, 6.2248e-24, 7.7204e-24,
6.7694e-24, 6.1721e-24, 6.0989e-24, 5.9535e-24, 7.7240e-24],
[ 6.5304e-23, 6.6529e-23, 5.3951e-23, 4.8348e-23, 5.9964e-23,
5.2578e-23, 4.7939e-23, 4.7370e-23, 4.6241e-23, 5.9991e-23],
[-3.5473e-23, -3.6138e-23, -2.9306e-23, -2.6262e-23, -3.2572e-23,
-2.8560e-23, -2.6040e-23, -2.5731e-23, -2.5118e-23, -3.2587e-23],
[-1.8701e-23, -1.9052e-23, -1.5450e-23, -1.3845e-23, -1.7172e-23,
-1.5056e-23, -1.3728e-23, -1.3565e-23, -1.3242e-23, -1.7180e-23],
[ 8.1550e-24, 8.3080e-24, 6.7373e-24, 6.0375e-24, 7.4882e-24,
6.5658e-24, 5.9864e-24, 5.9154e-24, 5.7744e-24, 7.4916e-24],
[ 6.1519e-23, 6.2673e-23, 5.0824e-23, 4.5546e-23, 5.6489e-23,
4.9530e-23, 4.5160e-23, 4.4624e-23, 4.3561e-23, 5.6515e-23],
[-8.5203e-23, -8.6801e-23, -7.0391e-23, -6.3080e-23, -7.8236e-23,
-6.8599e-23, -6.2546e-23, -6.1804e-23, -6.0331e-23, -7.8272e-23],
[-9.8839e-24, -1.0069e-23, -8.1656e-24, -7.3175e-24, -9.0757e-24,
-7.9577e-24, -7.2556e-24, -7.1695e-24, -6.9987e-24, -9.0799e-24],
[-5.5105e-24, -5.6139e-24, -4.5525e-24, -4.0797e-24, -5.0599e-24,
-4.4366e-24, -4.0452e-24, -3.9972e-24, -3.9019e-24, -5.0622e-24],
[-6.5156e-23, -6.6378e-23, -5.3829e-23, -4.8238e-23, -5.9828e-23,
-5.2458e-23, -4.7830e-23, -4.7262e-23, -4.6136e-23, -5.9855e-23]])
78-th layer: tensor([[ 2.3416e-22, 2.1715e-22, 2.4422e-22, 3.0941e-22, 2.6079e-22,
2.9132e-22, 2.6808e-22, 2.8636e-22, 2.6490e-22, 2.0250e-22],
[-5.2670e-22, -4.8845e-22, -5.4934e-22, -6.9597e-22, -5.8661e-22,
-6.5528e-22, -6.0300e-22, -6.4412e-22, -5.9585e-22, -4.5549e-22],
[ 4.5084e-22, 4.1810e-22, 4.7022e-22, 5.9573e-22, 5.0212e-22,
5.6090e-22, 5.1615e-22, 5.5135e-22, 5.1003e-22, 3.8989e-22],
[ 1.0601e-22, 9.8316e-23, 1.1057e-22, 1.4009e-22, 1.1807e-22,
1.3190e-22, 1.2137e-22, 1.2965e-22, 1.1993e-22, 9.1682e-23],
[-1.6783e-22, -1.5565e-22, -1.7505e-22, -2.2177e-22, -1.8692e-22,
-2.0881e-22, -1.9215e-22, -2.0525e-22, -1.8987e-22, -1.4514e-22],
[ 2.0271e-23, 1.8799e-23, 2.1143e-23, 2.6786e-23, 2.2577e-23,
2.5220e-23, 2.3208e-23, 2.4790e-23, 2.2933e-23, 1.7531e-23],
[-2.7209e-22, -2.5234e-22, -2.8379e-22, -3.5954e-22, -3.0304e-22,
-3.3852e-22, -3.1151e-22, -3.3275e-22, -3.0782e-22, -2.3531e-22],
[-2.0700e-22, -1.9197e-22, -2.1590e-22, -2.7353e-22, -2.3055e-22,
-2.5754e-22, -2.3699e-22, -2.5315e-22, -2.3418e-22, -1.7902e-22],
[ 1.9337e-22, 1.7933e-22, 2.0169e-22, 2.5552e-22, 2.1537e-22,
2.4058e-22, 2.2139e-22, 2.3648e-22, 2.1876e-22, 1.6723e-22],
[-3.6357e-22, -3.3717e-22, -3.7920e-22, -4.8042e-22, -4.0492e-22,
-4.5233e-22, -4.1624e-22, -4.4462e-22, -4.1130e-22, -3.1442e-22]])
79-th layer: tensor([[ 1.7265e-21, 1.2819e-21, 1.1340e-21, 9.6982e-22, 1.3040e-21,
1.6448e-21, 1.2898e-21, 1.3113e-21, 1.4313e-21, 1.6456e-21],
[-1.4295e-21, -1.0614e-21, -9.3896e-22, -8.0300e-22, -1.0797e-21,
-1.3618e-21, -1.0679e-21, -1.0858e-21, -1.1851e-21, -1.3625e-21],
[ 3.7064e-21, 2.7520e-21, 2.4345e-21, 2.0820e-21, 2.7993e-21,
3.5310e-21, 2.7689e-21, 2.8151e-21, 3.0727e-21, 3.5327e-21],
[-3.0247e-21, -2.2459e-21, -1.9867e-21, -1.6991e-21, -2.2845e-21,
-2.8815e-21, -2.2596e-21, -2.2974e-21, -2.5076e-21, -2.8829e-21],
[-4.1719e-21, -3.0976e-21, -2.7402e-21, -2.3435e-21, -3.1509e-21,
-3.9744e-21, -3.1166e-21, -3.1686e-21, -3.4586e-21, -3.9763e-21],
[-8.7316e-23, -6.4832e-23, -5.7352e-23, -4.9048e-23, -6.5946e-23,
-8.3182e-23, -6.5229e-23, -6.6318e-23, -7.2387e-23, -8.3222e-23],
[-7.9089e-22, -5.8724e-22, -5.1948e-22, -4.4426e-22, -5.9733e-22,
-7.5345e-22, -5.9083e-22, -6.0070e-22, -6.5567e-22, -7.5381e-22],
[ 4.1752e-21, 3.1001e-21, 2.7424e-21, 2.3453e-21, 3.1534e-21,
3.9775e-21, 3.1191e-21, 3.1712e-21, 3.4614e-21, 3.9794e-21],
[-3.7913e-21, -2.8150e-21, -2.4902e-21, -2.1296e-21, -2.8634e-21,
-3.6118e-21, -2.8322e-21, -2.8795e-21, -3.1430e-21, -3.6135e-21],
[ 9.3294e-22, 6.9271e-22, 6.1278e-22, 5.2406e-22, 7.0461e-22,
8.8877e-22, 6.9695e-22, 7.0859e-22, 7.7343e-22, 8.8920e-22]])
80-th layer: tensor([[-4.1043e-20, -2.9506e-20, -3.0757e-20, -4.0604e-20, -3.9914e-20,
-3.7653e-20, -3.4275e-20, -3.0576e-20, -3.7857e-20, -3.9084e-20],
[-4.1227e-21, -2.9638e-21, -3.0895e-21, -4.0786e-21, -4.0093e-21,
-3.7821e-21, -3.4428e-21, -3.0713e-21, -3.8026e-21, -3.9259e-21],
[ 1.1002e-20, 7.9092e-21, 8.2445e-21, 1.0884e-20, 1.0699e-20,
1.0093e-20, 9.1875e-21, 8.1959e-21, 1.0148e-20, 1.0477e-20],
[ 1.1084e-20, 7.9682e-21, 8.3061e-21, 1.0965e-20, 1.0779e-20,
1.0168e-20, 9.2561e-21, 8.2571e-21, 1.0223e-20, 1.0555e-20],
[ 1.0260e-20, 7.3760e-21, 7.6887e-21, 1.0150e-20, 9.9778e-21,
9.4125e-21, 8.5681e-21, 7.6434e-21, 9.4636e-21, 9.7703e-21],
[ 1.8964e-20, 1.3633e-20, 1.4212e-20, 1.8761e-20, 1.8443e-20,
1.7398e-20, 1.5837e-20, 1.4128e-20, 1.7492e-20, 1.8059e-20],
[ 1.1401e-20, 8.1962e-21, 8.5437e-21, 1.1279e-20, 1.1087e-20,
1.0459e-20, 9.5209e-21, 8.4934e-21, 1.0516e-20, 1.0857e-20],
[ 1.2127e-20, 8.7180e-21, 9.0877e-21, 1.1997e-20, 1.1793e-20,
1.1125e-20, 1.0127e-20, 9.0341e-21, 1.1185e-20, 1.1548e-20],
[ 2.6682e-20, 1.9182e-20, 1.9995e-20, 2.6397e-20, 2.5948e-20,
2.4478e-20, 2.2282e-20, 1.9877e-20, 2.4611e-20, 2.5408e-20],
[-1.5980e-20, -1.1488e-20, -1.1976e-20, -1.5810e-20, -1.5541e-20,
-1.4660e-20, -1.3345e-20, -1.1905e-20, -1.4740e-20, -1.5218e-20]])
81-th layer: tensor([[ 4.7821e-20, 6.0434e-20, 5.3424e-20, 6.1115e-20, 5.3167e-20,
4.7603e-20, 4.8888e-20, 6.9342e-20, 6.5714e-20, 5.5123e-20],
[ 4.1677e-20, 5.2670e-20, 4.6561e-20, 5.3264e-20, 4.6337e-20,
4.1488e-20, 4.2607e-20, 6.0434e-20, 5.7272e-20, 4.8041e-20],
[ 3.1525e-20, 3.9840e-20, 3.5219e-20, 4.0289e-20, 3.5049e-20,
3.1382e-20, 3.2228e-20, 4.5712e-20, 4.3321e-20, 3.6339e-20],
[ 1.1904e-19, 1.5044e-19, 1.3299e-19, 1.5213e-19, 1.3235e-19,
1.1850e-19, 1.2170e-19, 1.7261e-19, 1.6358e-19, 1.3722e-19],
[-8.3253e-20, -1.0521e-19, -9.3008e-20, -1.0640e-19, -9.2561e-20,
-8.2874e-20, -8.5111e-20, -1.2072e-19, -1.1440e-19, -9.5966e-20],
[-1.3391e-21, -1.6923e-21, -1.4960e-21, -1.7114e-21, -1.4888e-21,
-1.3330e-21, -1.3690e-21, -1.9417e-21, -1.8402e-21, -1.5436e-21],
[ 8.4952e-20, 1.0736e-19, 9.4906e-20, 1.0857e-19, 9.4450e-20,
8.4566e-20, 8.6848e-20, 1.2318e-19, 1.1674e-19, 9.7924e-20],
[ 1.5599e-19, 1.9713e-19, 1.7427e-19, 1.9935e-19, 1.7343e-19,
1.5528e-19, 1.5947e-19, 2.2619e-19, 2.1436e-19, 1.7981e-19],
[ 6.8005e-20, 8.5941e-20, 7.5973e-20, 8.6910e-20, 7.5608e-20,
6.7695e-20, 6.9522e-20, 9.8610e-20, 9.3450e-20, 7.8389e-20],
[-2.1326e-20, -2.6951e-20, -2.3825e-20, -2.7255e-20, -2.3711e-20,
-2.1229e-20, -2.1802e-20, -3.0924e-20, -2.9306e-20, -2.4583e-20]])
82-th layer: tensor([[-1.5469e-19, -1.7588e-19, -1.7523e-19, -1.3091e-19, -1.3477e-19,
-1.6132e-19, -1.6693e-19, -1.5453e-19, -1.8874e-19, -1.6217e-19],
[ 3.4967e-20, 3.9756e-20, 3.9609e-20, 2.9591e-20, 3.0463e-20,
3.6465e-20, 3.7734e-20, 3.4930e-20, 4.2663e-20, 3.6656e-20],
[ 5.8605e-19, 6.6631e-19, 6.6385e-19, 4.9594e-19, 5.1057e-19,
6.1116e-19, 6.3242e-19, 5.8543e-19, 7.1503e-19, 6.1436e-19],
[-7.5469e-19, -8.5804e-19, -8.5488e-19, -6.3865e-19, -6.5749e-19,
-7.8703e-19, -8.1440e-19, -7.5389e-19, -9.2078e-19, -7.9115e-19],
[ 5.3017e-19, 6.0277e-19, 6.0055e-19, 4.4865e-19, 4.6189e-19,
5.5289e-19, 5.7212e-19, 5.2961e-19, 6.4685e-19, 5.5578e-19],
[-9.4980e-19, -1.0799e-18, -1.0759e-18, -8.0377e-19, -8.2747e-19,
-9.9050e-19, -1.0249e-18, -9.4880e-19, -1.1588e-18, -9.9569e-19],
[ 1.6655e-19, 1.8936e-19, 1.8866e-19, 1.4094e-19, 1.4510e-19,
1.7369e-19, 1.7973e-19, 1.6637e-19, 2.0320e-19, 1.7460e-19],
[ 1.8550e-19, 2.1090e-19, 2.1013e-19, 1.5698e-19, 1.6161e-19,
1.9345e-19, 2.0017e-19, 1.8530e-19, 2.2632e-19, 1.9446e-19],
[-6.0381e-19, -6.8650e-19, -6.8397e-19, -5.1098e-19, -5.2604e-19,
-6.2969e-19, -6.5159e-19, -6.0318e-19, -7.3670e-19, -6.3298e-19],
[-7.6338e-19, -8.6792e-19, -8.6472e-19, -6.4601e-19, -6.6506e-19,
-7.9609e-19, -8.2378e-19, -7.6258e-19, -9.3139e-19, -8.0026e-19]])
83-th layer: tensor([[ 1.2754e-18, 1.1136e-18, 1.2728e-18, 1.0934e-18, 1.2504e-18,
9.5868e-19, 1.5573e-18, 1.2665e-18, 1.0406e-18, 1.0224e-18],
[ 4.0898e-18, 3.5708e-18, 4.0815e-18, 3.5060e-18, 4.0095e-18,
3.0742e-18, 4.9936e-18, 4.0611e-18, 3.3368e-18, 3.2786e-18],
[-1.9987e-18, -1.7450e-18, -1.9946e-18, -1.7134e-18, -1.9594e-18,
-1.5023e-18, -2.4404e-18, -1.9847e-18, -1.6307e-18, -1.6022e-18],
[ 2.5889e-18, 2.2604e-18, 2.5837e-18, 2.2194e-18, 2.5381e-18,
1.9460e-18, 3.1611e-18, 2.5708e-18, 2.1123e-18, 2.0754e-18],
[-3.8799e-18, -3.3875e-18, -3.8720e-18, -3.3261e-18, -3.8037e-18,
-2.9164e-18, -4.7373e-18, -3.8527e-18, -3.1655e-18, -3.1103e-18],
[-7.0400e-18, -6.1465e-18, -7.0256e-18, -6.0350e-18, -6.9017e-18,
-5.2917e-18, -8.5957e-18, -6.9905e-18, -5.7437e-18, -5.6435e-18],
[ 1.3197e-18, 1.1522e-18, 1.3170e-18, 1.1313e-18, 1.2938e-18,
9.9197e-19, 1.6113e-18, 1.3104e-18, 1.0767e-18, 1.0579e-18],
[ 2.8644e-18, 2.5008e-18, 2.8585e-18, 2.4555e-18, 2.8081e-18,
2.1530e-18, 3.4973e-18, 2.8442e-18, 2.3370e-18, 2.2962e-18],
[ 1.7817e-18, 1.5556e-18, 1.7780e-18, 1.5273e-18, 1.7467e-18,
1.3392e-18, 2.1754e-18, 1.7692e-18, 1.4536e-18, 1.4283e-18],
[-1.2262e-18, -1.0706e-18, -1.2237e-18, -1.0512e-18, -1.2021e-18,
-9.2169e-19, -1.4972e-18, -1.2176e-18, -1.0004e-18, -9.8297e-19]])
84-th layer: tensor([[-1.1870e-17, -1.6434e-17, -1.3918e-17, -1.3213e-17, -1.2434e-17,
-1.7723e-17, -8.1046e-18, -1.6568e-17, -1.5314e-17, -1.4789e-17],
[ 4.1472e-17, 5.7420e-17, 4.8628e-17, 4.6166e-17, 4.3444e-17,
6.1922e-17, 2.8317e-17, 5.7886e-17, 5.3504e-17, 5.1672e-17],
[-2.0856e-17, -2.8877e-17, -2.4455e-17, -2.3217e-17, -2.1848e-17,
-3.1141e-17, -1.4240e-17, -2.9111e-17, -2.6907e-17, -2.5986e-17],
[ 2.5042e-17, 3.4671e-17, 2.9362e-17, 2.7876e-17, 2.6232e-17,
3.7390e-17, 1.7098e-17, 3.4953e-17, 3.2307e-17, 3.1200e-17],
[ 1.0178e-17, 1.4092e-17, 1.1934e-17, 1.1330e-17, 1.0662e-17,
1.5196e-17, 6.9492e-18, 1.4206e-17, 1.3131e-17, 1.2681e-17],
[-1.4983e-19, -2.0745e-19, -1.7568e-19, -1.6679e-19, -1.5695e-19,
-2.2371e-19, -1.0230e-19, -2.0913e-19, -1.9330e-19, -1.8668e-19],
[ 1.6186e-18, 2.2411e-18, 1.8979e-18, 1.8018e-18, 1.6956e-18,
2.4168e-18, 1.1052e-18, 2.2592e-18, 2.0882e-18, 2.0167e-18],
[-8.7205e-18, -1.2074e-17, -1.0225e-17, -9.7075e-18, -9.1351e-18,
-1.3021e-17, -5.9543e-18, -1.2172e-17, -1.1251e-17, -1.0865e-17],
[ 1.5554e-17, 2.1535e-17, 1.8238e-17, 1.7314e-17, 1.6293e-17,
2.3223e-17, 1.0620e-17, 2.1710e-17, 2.0066e-17, 1.9379e-17],
[-6.4332e-19, -8.9071e-19, -7.5432e-19, -7.1613e-19, -6.7390e-19,
-9.6054e-19, -4.3925e-19, -8.9793e-19, -8.2996e-19, -8.0153e-19]])
85-th layer: tensor([[ 5.2797e-17, 4.4464e-17, 4.0163e-17, 4.9752e-17, 3.6746e-17,
4.2566e-17, 3.0932e-17, 3.5457e-17, 3.9707e-17, 4.4567e-17],
[-1.5027e-16, -1.2655e-16, -1.1431e-16, -1.4160e-16, -1.0458e-16,
-1.2115e-16, -8.8037e-17, -1.0091e-16, -1.1301e-16, -1.2684e-16],
[-2.1613e-16, -1.8202e-16, -1.6442e-16, -2.0367e-16, -1.5043e-16,
-1.7425e-16, -1.2663e-16, -1.4515e-16, -1.6255e-16, -1.8245e-16],
[-2.8538e-16, -2.4034e-16, -2.1710e-16, -2.6893e-16, -1.9863e-16,
-2.3009e-16, -1.6720e-16, -1.9166e-16, -2.1463e-16, -2.4090e-16],
[-2.4561e-16, -2.0684e-16, -1.8684e-16, -2.3144e-16, -1.7094e-16,
-1.9801e-16, -1.4389e-16, -1.6494e-16, -1.8471e-16, -2.0732e-16],
[ 2.6979e-16, 2.2720e-16, 2.0523e-16, 2.5423e-16, 1.8777e-16,
2.1751e-16, 1.5806e-16, 1.8118e-16, 2.0290e-16, 2.2773e-16],
[ 4.6013e-17, 3.8750e-17, 3.5003e-17, 4.3359e-17, 3.2025e-17,
3.7097e-17, 2.6958e-17, 3.0901e-17, 3.4605e-17, 3.8841e-17],
[-2.5488e-16, -2.1465e-16, -1.9389e-16, -2.4018e-16, -1.7739e-16,
-2.0549e-16, -1.4933e-16, -1.7117e-16, -1.9168e-16, -2.1515e-16],
[ 1.8061e-16, 1.5210e-16, 1.3739e-16, 1.7019e-16, 1.2570e-16,
1.4561e-16, 1.0581e-16, 1.2129e-16, 1.3583e-16, 1.5246e-16],
[-3.0551e-16, -2.5729e-16, -2.3240e-16, -2.8789e-16, -2.1263e-16,
-2.4631e-16, -1.7899e-16, -2.0517e-16, -2.2976e-16, -2.5789e-16]])
86-th layer: tensor([[-2.5303e-15, -2.3104e-15, -1.8218e-15, -2.1710e-15, -2.2115e-15,
-2.1623e-15, -2.6635e-15, -3.1376e-15, -2.3057e-15, -2.6494e-15],
[ 6.1341e-16, 5.6009e-16, 4.4164e-16, 5.2630e-16, 5.3611e-16,
5.2418e-16, 6.4569e-16, 7.6062e-16, 5.5895e-16, 6.4228e-16],
[-1.7874e-15, -1.6320e-15, -1.2869e-15, -1.5336e-15, -1.5622e-15,
-1.5274e-15, -1.8815e-15, -2.2164e-15, -1.6287e-15, -1.8716e-15],
[ 5.5465e-16, 5.0644e-16, 3.9934e-16, 4.7589e-16, 4.8476e-16,
4.7397e-16, 5.8384e-16, 6.8777e-16, 5.0541e-16, 5.8076e-16],
[ 1.7201e-15, 1.5706e-15, 1.2385e-15, 1.4759e-15, 1.5034e-15,
1.4699e-15, 1.8107e-15, 2.1330e-15, 1.5674e-15, 1.8011e-15],
[ 3.3212e-17, 3.0325e-17, 2.3912e-17, 2.8496e-17, 2.9027e-17,
2.8381e-17, 3.4960e-17, 4.1182e-17, 3.0263e-17, 3.4775e-17],
[-2.0982e-15, -1.9158e-15, -1.5107e-15, -1.8003e-15, -1.8338e-15,
-1.7930e-15, -2.2086e-15, -2.6018e-15, -1.9119e-15, -2.1970e-15],
[-5.1897e-16, -4.7386e-16, -3.7365e-16, -4.4528e-16, -4.5357e-16,
-4.4348e-16, -5.4628e-16, -6.4352e-16, -4.7289e-16, -5.4340e-16],
[-8.8119e-17, -8.0459e-17, -6.3444e-17, -7.5606e-17, -7.7014e-17,
-7.5300e-17, -9.2756e-17, -1.0927e-16, -8.0295e-17, -9.2266e-17],
[ 1.9271e-15, 1.7596e-15, 1.3875e-15, 1.6535e-15, 1.6843e-15,
1.6468e-15, 2.0285e-15, 2.3896e-15, 1.7560e-15, 2.0178e-15]])
87-th layer: tensor([[-9.5870e-15, -7.6038e-15, -1.1123e-14, -7.8136e-15, -9.6522e-15,
-9.6031e-15, -8.2616e-15, -1.0083e-14, -1.1478e-14, -7.2935e-15],
[ 9.7464e-15, 7.7303e-15, 1.1308e-14, 7.9435e-15, 9.8128e-15,
9.7629e-15, 8.3990e-15, 1.0250e-14, 1.1669e-14, 7.4148e-15],
[-1.0273e-15, -8.1482e-16, -1.1920e-15, -8.3729e-16, -1.0343e-15,
-1.0291e-15, -8.8531e-16, -1.0804e-15, -1.2299e-15, -7.8156e-16],
[-1.3193e-15, -1.0464e-15, -1.5308e-15, -1.0753e-15, -1.3283e-15,
-1.3216e-15, -1.1369e-15, -1.3875e-15, -1.5795e-15, -1.0037e-15],
[-1.1560e-14, -9.1685e-15, -1.3412e-14, -9.4214e-15, -1.1638e-14,
-1.1579e-14, -9.9617e-15, -1.2157e-14, -1.3839e-14, -8.7943e-15],
[-1.8847e-14, -1.4948e-14, -2.1867e-14, -1.5361e-14, -1.8975e-14,
-1.8879e-14, -1.6241e-14, -1.9821e-14, -2.2564e-14, -1.4338e-14],
[ 3.4696e-15, 2.7519e-15, 4.0257e-15, 2.8278e-15, 3.4932e-15,
3.4755e-15, 2.9900e-15, 3.6490e-15, 4.1539e-15, 2.6396e-15],
[-8.1469e-15, -6.4617e-15, -9.4526e-15, -6.6399e-15, -8.2024e-15,
-8.1607e-15, -7.0206e-15, -8.5681e-15, -9.7536e-15, -6.1979e-15],
[ 1.1499e-14, 9.1205e-15, 1.3342e-14, 9.3721e-15, 1.1577e-14,
1.1519e-14, 9.9094e-15, 1.2094e-14, 1.3767e-14, 8.7482e-15],
[ 1.2630e-14, 1.0017e-14, 1.4654e-14, 1.0294e-14, 1.2716e-14,
1.2651e-14, 1.0884e-14, 1.3283e-14, 1.5120e-14, 9.6083e-15]])
88-th layer: tensor([[-1.2615e-13, -1.2762e-13, -1.3994e-13, -1.1541e-13, -1.2459e-13,
-1.5992e-13, -1.1083e-13, -1.5067e-13, -1.5654e-13, -1.2885e-13],
[-6.3635e-14, -6.4375e-14, -7.0589e-14, -5.8218e-14, -6.2848e-14,
-8.0667e-14, -5.5904e-14, -7.6003e-14, -7.8963e-14, -6.4993e-14],
[-6.2015e-15, -6.2737e-15, -6.8793e-15, -5.6737e-15, -6.1249e-15,
-7.8614e-15, -5.4482e-15, -7.4069e-15, -7.6954e-15, -6.3339e-15],
[ 3.2944e-15, 3.3327e-15, 3.6545e-15, 3.0140e-15, 3.2537e-15,
4.1762e-15, 2.8942e-15, 3.9348e-15, 4.0880e-15, 3.3647e-15],
[ 1.7408e-14, 1.7611e-14, 1.9311e-14, 1.5926e-14, 1.7193e-14,
2.2067e-14, 1.5293e-14, 2.0792e-14, 2.1601e-14, 1.7780e-14],
[ 1.0281e-13, 1.0401e-13, 1.1405e-13, 9.4060e-14, 1.0154e-13,
1.3033e-13, 9.0321e-14, 1.2279e-13, 1.2758e-13, 1.0501e-13],
[ 1.8484e-14, 1.8699e-14, 2.0504e-14, 1.6911e-14, 1.8256e-14,
2.3432e-14, 1.6239e-14, 2.2077e-14, 2.2937e-14, 1.8879e-14],
[-4.7841e-14, -4.8398e-14, -5.3070e-14, -4.3769e-14, -4.7250e-14,
-6.0646e-14, -4.2029e-14, -5.7140e-14, -5.9365e-14, -4.8863e-14],
[ 3.4224e-14, 3.4623e-14, 3.7965e-14, 3.1311e-14, 3.3802e-14,
4.3385e-14, 3.0067e-14, 4.0877e-14, 4.2469e-14, 3.4955e-14],
[-5.4824e-14, -5.5462e-14, -6.0816e-14, -5.0158e-14, -5.4147e-14,
-6.9498e-14, -4.8164e-14, -6.5480e-14, -6.8030e-14, -5.5994e-14]])
89-th layer: tensor([[-1.2783e-12, -9.4581e-13, -9.1814e-13, -6.6650e-13, -8.1448e-13,
-1.1862e-12, -1.2478e-12, -1.3186e-12, -1.0982e-12, -1.1811e-12],
[-1.0016e-13, -7.4112e-14, -7.1944e-14, -5.2226e-14, -6.3822e-14,
-9.2947e-14, -9.7775e-14, -1.0333e-13, -8.6056e-14, -9.2552e-14],
[-3.9292e-13, -2.9073e-13, -2.8223e-13, -2.0487e-13, -2.5036e-13,
-3.6462e-13, -3.8356e-13, -4.0533e-13, -3.3758e-13, -3.6307e-13],
[-5.5482e-13, -4.1052e-13, -3.9851e-13, -2.8929e-13, -3.5352e-13,
-5.1485e-13, -5.4160e-13, -5.7234e-13, -4.7668e-13, -5.1267e-13],
[ 8.6588e-13, 6.4068e-13, 6.2194e-13, 4.5148e-13, 5.5172e-13,
8.0350e-13, 8.4524e-13, 8.9322e-13, 7.4393e-13, 8.0009e-13],
[ 2.0608e-13, 1.5248e-13, 1.4802e-13, 1.0745e-13, 1.3131e-13,
1.9123e-13, 2.0117e-13, 2.1259e-13, 1.7706e-13, 1.9042e-13],
[ 8.4504e-13, 6.2526e-13, 6.0697e-13, 4.4061e-13, 5.3844e-13,
7.8416e-13, 8.2489e-13, 8.7172e-13, 7.2602e-13, 7.8083e-13],
[ 1.2895e-13, 9.5413e-14, 9.2622e-14, 6.7237e-14, 8.2165e-14,
1.1966e-13, 1.2588e-13, 1.3302e-13, 1.1079e-13, 1.1915e-13],
[ 6.3368e-13, 4.6887e-13, 4.5516e-13, 3.3041e-13, 4.0377e-13,
5.8803e-13, 6.1858e-13, 6.5369e-13, 5.4444e-13, 5.8554e-13],
[ 2.9134e-13, 2.1557e-13, 2.0926e-13, 1.5191e-13, 1.8564e-13,
2.7035e-13, 2.8440e-13, 3.0054e-13, 2.5031e-13, 2.6920e-13]])
90-th layer: tensor([[ 2.6934e-12, 2.0273e-12, 2.2823e-12, 2.7581e-12, 3.1685e-12,
2.4043e-12, 2.7023e-12, 1.8143e-12, 2.2434e-12, 3.1669e-12],
[ 5.9940e-12, 4.5116e-12, 5.0791e-12, 6.1381e-12, 7.0513e-12,
5.3508e-12, 6.0139e-12, 4.0377e-12, 4.9925e-12, 7.0478e-12],
[-1.3627e-11, -1.0257e-11, -1.1547e-11, -1.3955e-11, -1.6031e-11,
-1.2165e-11, -1.3672e-11, -9.1794e-12, -1.1350e-11, -1.6023e-11],
[ 1.3144e-12, 9.8933e-13, 1.1138e-12, 1.3460e-12, 1.5463e-12,
1.1734e-12, 1.3188e-12, 8.8541e-13, 1.0948e-12, 1.5455e-12],
[ 7.8411e-13, 5.9019e-13, 6.6442e-13, 8.0296e-13, 9.2242e-13,
6.9996e-13, 7.8670e-13, 5.2819e-13, 6.5310e-13, 9.2196e-13],
[-4.6013e-12, -3.4633e-12, -3.8989e-12, -4.7119e-12, -5.4129e-12,
-4.1075e-12, -4.6165e-12, -3.0995e-12, -3.8325e-12, -5.4102e-12],
[-3.8289e-12, -2.8820e-12, -3.2445e-12, -3.9210e-12, -4.5043e-12,
-3.4180e-12, -3.8416e-12, -2.5792e-12, -3.1892e-12, -4.5021e-12],
[-4.6420e-12, -3.4939e-12, -3.9334e-12, -4.7536e-12, -5.4608e-12,
-4.1438e-12, -4.6573e-12, -3.1269e-12, -3.8664e-12, -5.4580e-12],
[ 4.0790e-12, 3.0702e-12, 3.4564e-12, 4.1771e-12, 4.7985e-12,
3.6412e-12, 4.0925e-12, 2.7477e-12, 3.3975e-12, 4.7961e-12],
[ 5.6810e-12, 4.2760e-12, 4.8138e-12, 5.8175e-12, 6.6830e-12,
5.0713e-12, 5.6998e-12, 3.8268e-12, 4.7318e-12, 6.6797e-12]])
91-th layer: tensor([[ 3.4557e-11, 3.1728e-11, 3.3636e-11, 3.2228e-11, 2.6516e-11,
2.5643e-11, 3.3864e-11, 2.7195e-11, 3.5224e-11, 3.3146e-11],
[-6.0189e-11, -5.5263e-11, -5.8585e-11, -5.6133e-11, -4.6185e-11,
-4.4663e-11, -5.8982e-11, -4.7366e-11, -6.1352e-11, -5.7732e-11],
[ 4.5801e-11, 4.2053e-11, 4.4581e-11, 4.2715e-11, 3.5145e-11,
3.3987e-11, 4.4883e-11, 3.6044e-11, 4.6687e-11, 4.3932e-11],
[ 4.8677e-11, 4.4693e-11, 4.7380e-11, 4.5397e-11, 3.7351e-11,
3.6121e-11, 4.7701e-11, 3.8307e-11, 4.9618e-11, 4.6690e-11],
[-7.2238e-11, -6.6326e-11, -7.0314e-11, -6.7371e-11, -5.5430e-11,
-5.3604e-11, -7.0790e-11, -5.6848e-11, -7.3634e-11, -6.9289e-11],
[ 6.0043e-11, 5.5129e-11, 5.8444e-11, 5.5998e-11, 4.6073e-11,
4.4555e-11, 5.8840e-11, 4.7252e-11, 6.1204e-11, 5.7592e-11],
[ 1.3029e-10, 1.1963e-10, 1.2682e-10, 1.2151e-10, 9.9976e-11,
9.6682e-11, 1.2768e-10, 1.0253e-10, 1.3281e-10, 1.2497e-10],
[-3.7852e-11, -3.4754e-11, -3.6843e-11, -3.5301e-11, -2.9045e-11,
-2.8088e-11, -3.7093e-11, -2.9788e-11, -3.8583e-11, -3.6307e-11],
[-9.3135e-11, -8.5513e-11, -9.0654e-11, -8.6860e-11, -7.1466e-11,
-6.9111e-11, -9.1268e-11, -7.3294e-11, -9.4935e-11, -8.9333e-11],
[ 3.9951e-11, 3.6681e-11, 3.8887e-11, 3.7259e-11, 3.0656e-11,
2.9646e-11, 3.9150e-11, 3.1440e-11, 4.0723e-11, 3.8320e-11]])
92-th layer: tensor([[-2.2994e-10, -2.4326e-10, -1.2835e-10, -1.8830e-10, -1.5604e-10,
-2.3003e-10, -1.7294e-10, -1.8024e-10, -1.6018e-10, -2.0094e-10],
[ 6.5995e-10, 6.9820e-10, 3.6838e-10, 5.4045e-10, 4.4786e-10,
6.6022e-10, 4.9635e-10, 5.1733e-10, 4.5973e-10, 5.7673e-10],
[ 4.0929e-10, 4.3301e-10, 2.2846e-10, 3.3518e-10, 2.7776e-10,
4.0946e-10, 3.0783e-10, 3.2084e-10, 2.8512e-10, 3.5768e-10],
[ 7.0889e-10, 7.4998e-10, 3.9570e-10, 5.8053e-10, 4.8108e-10,
7.0919e-10, 5.3316e-10, 5.5570e-10, 4.9383e-10, 6.1951e-10],
[ 5.4344e-10, 5.7494e-10, 3.0334e-10, 4.4504e-10, 3.6880e-10,
5.4367e-10, 4.0872e-10, 4.2600e-10, 3.7857e-10, 4.7492e-10],
[ 2.0679e-10, 2.1878e-10, 1.1543e-10, 1.6935e-10, 1.4033e-10,
2.0688e-10, 1.5553e-10, 1.6210e-10, 1.4405e-10, 1.8072e-10],
[ 4.0367e-10, 4.2706e-10, 2.2532e-10, 3.3057e-10, 2.7394e-10,
4.0383e-10, 3.0360e-10, 3.1643e-10, 2.8120e-10, 3.5277e-10],
[ 3.8559e-10, 4.0794e-10, 2.1523e-10, 3.1577e-10, 2.6167e-10,
3.8575e-10, 2.9001e-10, 3.0226e-10, 2.6861e-10, 3.3697e-10],
[-8.3284e-10, -8.8111e-10, -4.6488e-10, -6.8203e-10, -5.6519e-10,
-8.3318e-10, -6.2638e-10, -6.5285e-10, -5.8016e-10, -7.2782e-10],
[ 7.2955e-10, 7.7183e-10, 4.0723e-10, 5.9745e-10, 4.9509e-10,
7.2985e-10, 5.4869e-10, 5.7188e-10, 5.0821e-10, 6.3755e-10]])
93-th layer: tensor([[-6.6412e-10, -4.8910e-10, -6.5833e-10, -4.7000e-10, -7.6452e-10,
-6.8227e-10, -5.8465e-10, -5.6311e-10, -4.4868e-10, -4.9717e-10],
[-5.1250e-09, -3.7743e-09, -5.0803e-09, -3.6270e-09, -5.8997e-09,
-5.2650e-09, -4.5117e-09, -4.3455e-09, -3.4624e-09, -3.8366e-09],
[-2.1575e-09, -1.5889e-09, -2.1387e-09, -1.5269e-09, -2.4837e-09,
-2.2165e-09, -1.8993e-09, -1.8294e-09, -1.4576e-09, -1.6152e-09],
[ 3.6830e-09, 2.7124e-09, 3.6509e-09, 2.6065e-09, 4.2398e-09,
3.7836e-09, 3.2422e-09, 3.1228e-09, 2.4882e-09, 2.7571e-09],
[-1.5977e-09, -1.1767e-09, -1.5838e-09, -1.1307e-09, -1.8392e-09,
-1.6414e-09, -1.4065e-09, -1.3547e-09, -1.0794e-09, -1.1961e-09],
[ 4.5960e-09, 3.3848e-09, 4.5560e-09, 3.2526e-09, 5.2908e-09,
4.7216e-09, 4.0460e-09, 3.8970e-09, 3.1051e-09, 3.4407e-09],
[ 3.6814e-09, 2.7112e-09, 3.6493e-09, 2.6054e-09, 4.2380e-09,
3.7821e-09, 3.2409e-09, 3.1215e-09, 2.4872e-09, 2.7560e-09],
[ 7.2925e-10, 5.3706e-10, 7.2289e-10, 5.1609e-10, 8.3949e-10,
7.4918e-10, 6.4198e-10, 6.1834e-10, 4.9268e-10, 5.4593e-10],
[-5.9701e-09, -4.3968e-09, -5.9181e-09, -4.2251e-09, -6.8727e-09,
-6.1333e-09, -5.2557e-09, -5.0621e-09, -4.0334e-09, -4.4693e-09],
[ 9.4928e-10, 6.9911e-10, 9.4101e-10, 6.7181e-10, 1.0928e-09,
9.7523e-10, 8.3568e-10, 8.0490e-10, 6.4134e-10, 7.1065e-10]])
94-th layer: tensor([[ 1.7702e-09, 1.5122e-09, 1.7292e-09, 1.6010e-09, 2.2196e-09,
1.7050e-09, 1.9757e-09, 1.7905e-09, 1.3302e-09, 2.2410e-09],
[ 1.3019e-08, 1.1122e-08, 1.2717e-08, 1.1775e-08, 1.6324e-08,
1.2539e-08, 1.4531e-08, 1.3168e-08, 9.7829e-09, 1.6482e-08],
[ 2.1509e-08, 1.8375e-08, 2.1011e-08, 1.9454e-08, 2.6969e-08,
2.0716e-08, 2.4007e-08, 2.1755e-08, 1.6163e-08, 2.7230e-08],
[-9.1135e-10, -7.7852e-10, -8.9023e-10, -8.2425e-10, -1.1427e-09,
-8.7775e-10, -1.0172e-09, -9.2177e-10, -6.8481e-10, -1.1537e-09],
[-6.4190e-09, -5.4835e-09, -6.2703e-09, -5.8056e-09, -8.0484e-09,
-6.1824e-09, -7.1643e-09, -6.4925e-09, -4.8234e-09, -8.1263e-09],
[-1.1750e-08, -1.0038e-08, -1.1478e-08, -1.0627e-08, -1.4733e-08,
-1.1317e-08, -1.3114e-08, -1.1885e-08, -8.8293e-09, -1.4875e-08],
[ 2.2667e-08, 1.9363e-08, 2.2142e-08, 2.0501e-08, 2.8420e-08,
2.1831e-08, 2.5299e-08, 2.2926e-08, 1.7032e-08, 2.8695e-08],
[-3.6121e-08, -3.0856e-08, -3.5284e-08, -3.2669e-08, -4.5290e-08,
-3.4789e-08, -4.0315e-08, -3.6534e-08, -2.7142e-08, -4.5728e-08],
[-3.9715e-08, -3.3927e-08, -3.8795e-08, -3.5919e-08, -4.9796e-08,
-3.8251e-08, -4.4326e-08, -4.0169e-08, -2.9843e-08, -5.0278e-08],
[ 1.0292e-08, 8.7920e-09, 1.0054e-08, 9.3084e-09, 1.2905e-08,
9.9126e-09, 1.1487e-08, 1.0410e-08, 7.7336e-09, 1.3029e-08]])
95-th layer: tensor([[ 3.2620e-07, 3.6950e-07, 2.9170e-07, 3.9593e-07, 3.2885e-07,
3.7534e-07, 4.0026e-07, 3.7223e-07, 3.2071e-07, 3.1340e-07],
[-1.1647e-07, -1.3193e-07, -1.0415e-07, -1.4136e-07, -1.1741e-07,
-1.3401e-07, -1.4291e-07, -1.3290e-07, -1.1450e-07, -1.1189e-07],
[-6.1962e-08, -7.0186e-08, -5.5409e-08, -7.5206e-08, -6.2465e-08,
-7.1296e-08, -7.6029e-08, -7.0705e-08, -6.0918e-08, -5.9529e-08],
[-8.7815e-08, -9.9470e-08, -7.8527e-08, -1.0658e-07, -8.8527e-08,
-1.0104e-07, -1.0775e-07, -1.0021e-07, -8.6335e-08, -8.4367e-08],
[ 5.7755e-08, 6.5421e-08, 5.1647e-08, 7.0100e-08, 5.8224e-08,
6.6455e-08, 7.0867e-08, 6.5904e-08, 5.6782e-08, 5.5488e-08],
[ 1.2202e-07, 1.3821e-07, 1.0911e-07, 1.4810e-07, 1.2301e-07,
1.4040e-07, 1.4972e-07, 1.3923e-07, 1.1996e-07, 1.1723e-07],
[ 8.8777e-09, 1.0056e-08, 7.9387e-09, 1.0775e-08, 8.9497e-09,
1.0215e-08, 1.0893e-08, 1.0130e-08, 8.7280e-09, 8.5291e-09],
[ 3.2765e-08, 3.7114e-08, 2.9300e-08, 3.9769e-08, 3.3031e-08,
3.7701e-08, 4.0204e-08, 3.7389e-08, 3.2213e-08, 3.1479e-08],
[-1.1779e-07, -1.3343e-07, -1.0533e-07, -1.4297e-07, -1.1875e-07,
-1.3554e-07, -1.4453e-07, -1.3441e-07, -1.1581e-07, -1.1317e-07],
[-2.0346e-07, -2.3047e-07, -1.8194e-07, -2.4695e-07, -2.0511e-07,
-2.3411e-07, -2.4965e-07, -2.3217e-07, -2.0003e-07, -1.9547e-07]])
96-th layer: tensor([[-1.3691e-06, -2.1372e-06, -1.9141e-06, -1.4533e-06, -8.6956e-07,
-1.2537e-06, -1.2337e-06, -1.0255e-06, -1.7058e-06, -1.3026e-06],
[-9.6032e-08, -1.4991e-07, -1.3426e-07, -1.0194e-07, -6.0995e-08,
-8.7942e-08, -8.6535e-08, -7.1934e-08, -1.1965e-07, -9.1367e-08],
[ 7.4557e-07, 1.1639e-06, 1.0424e-06, 7.9145e-07, 4.7355e-07,
6.8276e-07, 6.7184e-07, 5.5848e-07, 9.2895e-07, 7.0935e-07],
[-1.2433e-06, -1.9409e-06, -1.7382e-06, -1.3198e-06, -7.8966e-07,
-1.1385e-06, -1.1203e-06, -9.3129e-07, -1.5491e-06, -1.1829e-06],
[ 5.2042e-07, 8.1242e-07, 7.2759e-07, 5.5244e-07, 3.3054e-07,
4.7658e-07, 4.6896e-07, 3.8983e-07, 6.4842e-07, 4.9514e-07],
[-6.0169e-08, -9.3929e-08, -8.4121e-08, -6.3871e-08, -3.8216e-08,
-5.5100e-08, -5.4219e-08, -4.5070e-08, -7.4968e-08, -5.7246e-08],
[ 9.2185e-07, 1.4391e-06, 1.2888e-06, 9.7857e-07, 5.8551e-07,
8.4419e-07, 8.3069e-07, 6.9052e-07, 1.1486e-06, 8.7707e-07],
[ 6.5469e-07, 1.0220e-06, 9.1531e-07, 6.9498e-07, 4.1582e-07,
5.9954e-07, 5.8995e-07, 4.9040e-07, 8.1572e-07, 6.2289e-07],
[ 8.8820e-07, 1.3866e-06, 1.2418e-06, 9.4286e-07, 5.6414e-07,
8.1338e-07, 8.0037e-07, 6.6532e-07, 1.1067e-06, 8.4506e-07],
[-3.1236e-07, -4.8763e-07, -4.3671e-07, -3.3158e-07, -1.9840e-07,
-2.8605e-07, -2.8147e-07, -2.3398e-07, -3.8919e-07, -2.9719e-07]])
97-th layer: tensor([[-1.3921e-05, -1.1897e-05, -1.0796e-05, -1.2853e-05, -1.1474e-05,
-9.3927e-06, -1.4445e-05, -1.3041e-05, -1.2036e-05, -1.4108e-05],
[-4.3005e-06, -3.6750e-06, -3.3349e-06, -3.9703e-06, -3.5444e-06,
-2.9015e-06, -4.4621e-06, -4.0286e-06, -3.7179e-06, -4.3581e-06],
[ 1.6563e-06, 1.4154e-06, 1.2844e-06, 1.5292e-06, 1.3651e-06,
1.1175e-06, 1.7186e-06, 1.5516e-06, 1.4319e-06, 1.6785e-06],
[ 1.8844e-05, 1.6103e-05, 1.4613e-05, 1.7397e-05, 1.5531e-05,
1.2714e-05, 1.9552e-05, 1.7653e-05, 1.6291e-05, 1.9096e-05],
[-3.3179e-06, -2.8353e-06, -2.5729e-06, -3.0632e-06, -2.7346e-06,
-2.2385e-06, -3.4426e-06, -3.1081e-06, -2.8684e-06, -3.3623e-06],
[ 1.5498e-05, 1.3244e-05, 1.2018e-05, 1.4308e-05, 1.2773e-05,
1.0456e-05, 1.6080e-05, 1.4518e-05, 1.3398e-05, 1.5705e-05],
[-6.4167e-06, -5.4834e-06, -4.9760e-06, -5.9240e-06, -5.2885e-06,
-4.3292e-06, -6.6578e-06, -6.0109e-06, -5.5474e-06, -6.5026e-06],
[-7.3885e-06, -6.3139e-06, -5.7296e-06, -6.8212e-06, -6.0895e-06,
-4.9849e-06, -7.6661e-06, -6.9213e-06, -6.3876e-06, -7.4874e-06],
[ 6.7090e-06, 5.7332e-06, 5.2026e-06, 6.1939e-06, 5.5294e-06,
4.5265e-06, 6.9611e-06, 6.2848e-06, 5.8001e-06, 6.7988e-06],
[-5.7382e-06, -4.9036e-06, -4.4498e-06, -5.2976e-06, -4.7293e-06,
-3.8715e-06, -5.9538e-06, -5.3753e-06, -4.9608e-06, -5.8150e-06]])
98-th layer: tensor([[-6.1899e-05, -7.0697e-05, -5.7808e-05, -5.1832e-05, -5.6604e-05,
-5.7637e-05, -5.3070e-05, -4.4301e-05, -5.6758e-05, -4.1822e-05],
[ 4.0809e-05, 4.6609e-05, 3.8112e-05, 3.4172e-05, 3.7318e-05,
3.7999e-05, 3.4989e-05, 2.9207e-05, 3.7420e-05, 2.7573e-05],
[ 2.7760e-05, 3.1705e-05, 2.5925e-05, 2.3245e-05, 2.5385e-05,
2.5848e-05, 2.3800e-05, 1.9867e-05, 2.5454e-05, 1.8756e-05],
[ 1.2785e-04, 1.4603e-04, 1.1940e-04, 1.0706e-04, 1.1692e-04,
1.1905e-04, 1.0962e-04, 9.1503e-05, 1.1723e-04, 8.6384e-05],
[ 8.5936e-05, 9.8150e-05, 8.0256e-05, 7.1959e-05, 7.8585e-05,
8.0019e-05, 7.3679e-05, 6.1504e-05, 7.8798e-05, 5.8063e-05],
[ 2.9979e-05, 3.4240e-05, 2.7997e-05, 2.5103e-05, 2.7415e-05,
2.7915e-05, 2.5703e-05, 2.1456e-05, 2.7489e-05, 2.0255e-05],
[ 5.2946e-05, 6.0471e-05, 4.9446e-05, 4.4335e-05, 4.8417e-05,
4.9300e-05, 4.5394e-05, 3.7893e-05, 4.8548e-05, 3.5773e-05],
[-1.0384e-04, -1.1860e-04, -9.6978e-05, -8.6952e-05, -9.4959e-05,
-9.6691e-05, -8.9031e-05, -7.4318e-05, -9.5216e-05, -7.0161e-05],
[ 5.0053e-05, 5.7167e-05, 4.6745e-05, 4.1912e-05, 4.5772e-05,
4.6607e-05, 4.2914e-05, 3.5823e-05, 4.5896e-05, 3.3818e-05],
[ 2.9782e-05, 3.4015e-05, 2.7813e-05, 2.4938e-05, 2.7235e-05,
2.7731e-05, 2.5534e-05, 2.1315e-05, 2.7308e-05, 2.0122e-05]])
99-th layer: tensor([[ 1.6418e-04, 2.0542e-04, 1.9810e-04, 2.2766e-04, 1.7518e-04,
2.2426e-04, 1.9689e-04, 2.1978e-04, 1.8350e-04, 2.0160e-04],
[-4.5582e-04, -5.7029e-04, -5.4997e-04, -6.3204e-04, -4.8635e-04,
-6.2261e-04, -5.4663e-04, -6.1016e-04, -5.0944e-04, -5.5970e-04],
[ 5.2332e-04, 6.5474e-04, 6.3141e-04, 7.2564e-04, 5.5837e-04,
7.1481e-04, 6.2757e-04, 7.0051e-04, 5.8488e-04, 6.4258e-04],
[ 3.2472e-04, 4.0627e-04, 3.9180e-04, 4.5027e-04, 3.4647e-04,
4.4355e-04, 3.8942e-04, 4.3468e-04, 3.6292e-04, 3.9873e-04],
[ 4.8669e-04, 6.0892e-04, 5.8723e-04, 6.7486e-04, 5.1929e-04,
6.6479e-04, 5.8366e-04, 6.5149e-04, 5.4395e-04, 5.9761e-04],
[ 8.2487e-04, 1.0320e-03, 9.9526e-04, 1.1438e-03, 8.8012e-04,
1.1267e-03, 9.8921e-04, 1.1042e-03, 9.2191e-04, 1.0129e-03],
[ 1.6304e-05, 2.0399e-05, 1.9672e-05, 2.2607e-05, 1.7396e-05,
2.2270e-05, 1.9552e-05, 2.1825e-05, 1.8222e-05, 2.0020e-05],
[-7.8433e-05, -9.8131e-05, -9.4634e-05, -1.0876e-04, -8.3686e-05,
-1.0713e-04, -9.4059e-05, -1.0499e-04, -8.7660e-05, -9.6308e-05],
[ 1.5322e-04, 1.9170e-04, 1.8487e-04, 2.1246e-04, 1.6349e-04,
2.0929e-04, 1.8375e-04, 2.0511e-04, 1.7125e-04, 1.8814e-04],
[-3.1445e-04, -3.9342e-04, -3.7940e-04, -4.3602e-04, -3.3551e-04,
-4.2951e-04, -3.7710e-04, -4.2092e-04, -3.5144e-04, -3.8611e-04]])
100-th layer: tensor([[-1.0096e-03, -1.3075e-03, -1.7615e-03, -1.1832e-03, -1.4080e-03,
-1.4563e-03, -1.1827e-03, -1.7725e-03, -8.3344e-04, -1.3619e-03],
[ 3.4639e-03, 4.4860e-03, 6.0437e-03, 4.0595e-03, 4.8310e-03,
4.9965e-03, 4.0577e-03, 6.0814e-03, 2.8595e-03, 4.6726e-03],
[ 2.2956e-03, 2.9729e-03, 4.0052e-03, 2.6903e-03, 3.2016e-03,
3.3113e-03, 2.6891e-03, 4.0302e-03, 1.8950e-03, 3.0966e-03],
[ 8.0854e-04, 1.0471e-03, 1.4107e-03, 9.4757e-04, 1.1277e-03,
1.1663e-03, 9.4715e-04, 1.4195e-03, 6.6747e-04, 1.0907e-03],
[-1.9851e-03, -2.5708e-03, -3.4634e-03, -2.3264e-03, -2.7685e-03,
-2.8634e-03, -2.3254e-03, -3.4850e-03, -1.6387e-03, -2.6777e-03],
[ 4.7600e-03, 6.1646e-03, 8.3051e-03, 5.5785e-03, 6.6387e-03,
6.8662e-03, 5.5760e-03, 8.3569e-03, 3.9295e-03, 6.4210e-03],
[ 8.8873e-04, 1.1510e-03, 1.5506e-03, 1.0415e-03, 1.2395e-03,
1.2820e-03, 1.0411e-03, 1.5603e-03, 7.3366e-04, 1.1988e-03],
[-1.0693e-03, -1.3848e-03, -1.8657e-03, -1.2532e-03, -1.4913e-03,
-1.5424e-03, -1.2526e-03, -1.8773e-03, -8.8273e-04, -1.4424e-03],
[ 2.6811e-05, 3.4722e-05, 4.6778e-05, 3.1421e-05, 3.7392e-05,
3.8673e-05, 3.1407e-05, 4.7070e-05, 2.2133e-05, 3.6166e-05],
[-8.8473e-04, -1.1458e-03, -1.5436e-03, -1.0369e-03, -1.2339e-03,
-1.2762e-03, -1.0364e-03, -1.5533e-03, -7.3037e-04, -1.1934e-03]])
[39]:
import numpy as np
[45]:
# Create Tensors to hold input and outputs.
x = torch.tensor([1.0])
y = torch.tensor([1.0])
model_layers = collections.OrderedDict()
n_layers = 100
for i in range(n_layers):
model_layers[f'linear_{i+1}']=torch.nn.Linear(1, 1)
# Use the nn package to define our model and loss function.
model = torch.nn.Sequential(
model_layers
)
loss_fn = torch.nn.MSELoss(reduction='sum')
y_pred = model(x)
loss = loss_fn(y_pred, y)
loss.backward()
weights_list=[]
for i in range(n_layers):
weights_list.append(np.abs(model[i].weight.grad[0][0].numpy()))
plt.plot(range(n_layers),weights_list)
ax = plt.gca()
ax.set_yscale('log')
plt.xlabel('n-th layer')
plt.ylabel('Gradient value')
[45]:
Text(0, 0.5, 'Gradient value')
[69]:
def vanishing_gradient(n_layers):
# Create Tensors to hold input and outputs.
x = torch.tensor([1.0])
y = torch.tensor([1.0])
model_layers = collections.OrderedDict()
#n_layers = 10
for i in range(n_layers):
model_layers[f'linear_{i+1}']=torch.nn.Linear(1, 1)
# Use the nn package to define our model and loss function.
model = torch.nn.Sequential(
model_layers
)
loss_fn = torch.nn.MSELoss(reduction='sum')
y_pred = model(x)
loss = loss_fn(y_pred, y)
loss.backward()
weights_list=[]
for i in range(n_layers):
np.abs(model[i].weight.grad.numpy())
weights_list.append()
plt.bar(range(n_layers),weights_list,label='Absolute value of gradients')
ax = plt.gca()
ax.set_yscale('log')
ax.set_title(f'{n_layers} layers with 1 nodes in each layer')
plt.xlabel('n-th layer')
plt.ylabel('Gradient value')
plt.legend()
plt.savefig(f'{n_layers}_layers_with_1_nodes.pdf',
bbox_inches='tight')
[70]:
vanishing_gradient(10)
[71]:
vanishing_gradient(20)
[72]:
vanishing_gradient(30)
[73]:
vanishing_gradient(50)
[74]:
vanishing_gradient(100)
[43]:
raw_X
[43]:
array([[5748, 5605, 5734, ..., 55, 46, 27],
[6599, 6541, 6568, ..., 54, 42, 26],
[6560, 6594, 6673, ..., 51, 40, 26],
...,
[3675, 3559, 3542, ..., 10, 9, 6],
[3840, 3648, 3610, ..., 8, 6, 4],
[3675, 3604, 3584, ..., 9, 6, 5]], dtype=int16)
[46]:
len(raw_y)
[46]:
8079
[181]:
from torch.utils.data import TensorDataset,DataLoader
[205]:
inps = torch.tensor(hyperspectral_df[['NDVI', 'MNDWI', 'NDBI']].to_numpy(),
dtype=torch.float32)
tgts = torch.tensor(raw_y)
#, dtype=torch.float32)
#.view(8079, 1)
dataset = TensorDataset(inps, tgts)
loader = DataLoader(dataset,batch_size=8079)
[206]:
tgts
[206]:
tensor([0, 0, 0, ..., 6, 6, 6])
[ ]:
[207]:
from torch import nn
[ ]:
[ ]:
[208]:
def vanishing_gradient_sigmoid(n_layers, n_epoch=100):
model_layers = collections.OrderedDict()
for i in range(n_layers):
if i == 0:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(3, 30)
model_layers[f'sigmoid_{i+1}'] = torch.nn.Sigmoid()
else:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(30, 30)
model_layers[f'sigmoid_{i+1}'] = torch.nn.Sigmoid()
model_layers[f'output_layer'] = torch.nn.Linear(30, 7)
model = torch.nn.Sequential(model_layers)
loss_fn = nn.CrossEntropyLoss()
learning_rate = 0.1
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
grad_list = []
#n_epoch = 50
for e in range(n_epoch):
for X, y in loader:
# Compute prediction and loss
y_pred = model(X)
loss = loss_fn(y_pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
grad_list.append([
np.linalg.norm(model[2 * i].weight.grad.numpy(), ord=2)
for i in range(n_layers)
])
optimizer.step()
grad = np.array(grad_list)
# plt.bar(range(n_layers),weights_list,label='Absolute value of gradients')
ax = plt.gca()
ax.set_yscale('log')
ax.set_title(f'Sigmoid with 30 neurons/layer')
for i in range(grad.shape[1]):
plt.plot(range(1, n_epoch + 1),
grad[:, i],
label=f'Hidden layer {i+1}')
plt.xlabel('Epoch')
plt.ylabel('$\Vert \delta \Vert$')
plt.legend(loc='upper left',
#fontsize=3,
frameon=False,
bbox_to_anchor=(1, 1))
plt.savefig(f'Sigmoid-with-30-neurons-each-layer.pdf',
bbox_inches='tight')
[209]:
vanishing_gradient_sigmoid(n_layers=7,n_epoch=100)
[210]:
def vanishing_gradient_relu(n_layers,loader, n_epoch=100):
model_layers = collections.OrderedDict()
for i in range(n_layers):
if i == 0:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(3, 30)
model_layers[f'relu_{i+1}'] = torch.nn.ReLU()
else:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(30, 30)
model_layers[f'relu_{i+1}'] = torch.nn.ReLU()
model_layers[f'output_layer'] = torch.nn.Linear(30, 7)
model = torch.nn.Sequential(model_layers)
loss_fn = nn.CrossEntropyLoss()
learning_rate = 0.1
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
grad_list = []
#n_epoch = 50
for e in range(n_epoch):
for X, y in loader:
# Compute prediction and loss
y_pred = model(X)
loss = loss_fn(y_pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
optimizer.step()
grad_list.append([
np.linalg.norm(model[2 * i].weight.grad.numpy(), ord=2)
for i in range(n_layers)
])
grad = np.array(grad_list)
# plt.bar(range(n_layers),weights_list,label='Absolute value of gradients')
ax = plt.gca()
ax.set_yscale('log')
ax.set_title(f'ReLU with 30 neurons/layer')
for i in range(grad.shape[1]):
plt.plot(range(1, n_epoch + 1),
grad[:, i],
label=f'Hidden layer {i+1}')
plt.xlabel('Epoch')
plt.ylabel('$\Vert \delta \Vert$')
plt.legend(loc='upper left',
#fontsize=3,
frameon=False,
bbox_to_anchor=(1, 1))
plt.savefig(f'ReLU-with-30-neurons-each-layer.pdf',
bbox_inches='tight')
[211]:
vanishing_gradient_relu(n_layers=7,n_epoch=100,loader=loader)
[212]:
def vanishing_gradient_relu_bn(n_layers,loader, n_epoch=100):
model_layers = collections.OrderedDict()
for i in range(n_layers):
if i == 0:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(3, 30)
else:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(30, 30)
model_layers[f'relu_{i+1}'] = torch.nn.ReLU()
model_layers[f'bn_{i+1}'] = torch.nn.BatchNorm1d(30)
model_layers[f'output_layer'] = torch.nn.Linear(30, 7)
model = torch.nn.Sequential(model_layers)
loss_fn = nn.CrossEntropyLoss()
learning_rate = 0.1
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
grad_list = []
#n_epoch = 50
for e in range(n_epoch):
for X, y in loader:
# Compute prediction and loss
y_pred = model(X)
loss = loss_fn(y_pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
grad_list.append([
np.linalg.norm(model[3 * i].weight.grad.numpy(), ord=2)
for i in range(n_layers)
])
optimizer.step()
grad = np.array(grad_list)
# plt.bar(range(n_layers),weights_list,label='Absolute value of gradients')
ax = plt.gca()
ax.set_yscale('log')
ax.set_title(f'ReLU+BN with 30 neurons/layer')
for i in range(grad.shape[1]):
plt.plot(range(1, n_epoch + 1),
grad[:, i],
label=f'Hidden layer {i+1}')
plt.xlabel('Epoch')
plt.ylabel('$\Vert \delta \Vert$')
plt.legend(loc='upper left',
#fontsize=3,
frameon=False,
bbox_to_anchor=(1, 1))
plt.savefig(f'ReLU-BN-with-30-neurons-each-layer.pdf',
bbox_inches='tight')
[213]:
vanishing_gradient_relu_bn(n_layers=7,n_epoch=100,loader=loader)
[214]:
def vanishing_gradient_Sigmoid_bn(n_layers,loader, n_epoch=100):
model_layers = collections.OrderedDict()
for i in range(n_layers):
if i == 0:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(3, 30)
else:
model_layers[f'linear_{i+1}'] = torch.nn.Linear(30, 30)
model_layers[f'bn_{i+1}'] = torch.nn.BatchNorm1d(30)
model_layers[f'Sigmoid_{i+1}'] = torch.nn.Sigmoid()
model_layers[f'output_layer'] = torch.nn.Linear(30, 7)
model = torch.nn.Sequential(model_layers)
loss_fn = nn.CrossEntropyLoss()
learning_rate = 0.1
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
grad_list = []
#n_epoch = 50
for e in range(n_epoch):
for X, y in loader:
# Compute prediction and loss
y_pred = model(X)
loss = loss_fn(y_pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
grad_list.append([
np.linalg.norm(model[3 * i].weight.grad.numpy(), ord=2)
for i in range(n_layers)
])
optimizer.step()
grad = np.array(grad_list)
# plt.bar(range(n_layers),weights_list,label='Absolute value of gradients')
ax = plt.gca()
ax.set_yscale('log')
ax.set_title(f'Sigmoid+BN with 30 neurons/layer')
for i in range(grad.shape[1]):
plt.plot(range(1, n_epoch + 1),
grad[:, i],
label=f'Hidden layer {i+1}')
plt.xlabel('Epoch')
plt.ylabel('$\Vert \delta \Vert$')
plt.legend(loc='upper left',
#fontsize=3,
frameon=False,
bbox_to_anchor=(1, 1))
plt.savefig(f'Sigmoid-BN-with-30-neurons-each-layer.pdf',
bbox_inches='tight')
[215]:
vanishing_gradient_Sigmoid_bn(n_layers=7,n_epoch=100,loader=loader)
[ ]: