๐0.80032|Spaceship Titanic with a FCNN in PyTorch
Original Notebook : https://www.kaggle.com/code/marcokurepa/0-80032-spaceship-titanic-with-a-fcnn-in-pytorch
Data Dictionary
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
from scipy import stats
sns.set_style("whitegrid")
plt.style.use("fivethirtyeight")
import os
train = pd.read_csv(os.path.join(os.getcwd(), "data", "train.csv"))
test = pd.read_csv(os.path.join(os.getcwd(), "data", "test.csv"))
train.head(); test.head()
#> PassengerId HomePlanet CryoSleep ... VRDeck Name Transported
#> 0 0001_01 Europa False ... 0.0 Maham Ofracculy False
#> 1 0002_01 Earth False ... 44.0 Juanna Vines True
#> 2 0003_01 Europa False ... 49.0 Altark Susent False
#> 3 0003_02 Europa False ... 193.0 Solam Susent False
#> 4 0004_01 Earth False ... 2.0 Willy Santantines True
#>
#> [5 rows x 14 columns]
#> PassengerId HomePlanet CryoSleep ... Spa VRDeck Name
#> 0 0013_01 Earth True ... 0.0 0.0 Nelly Carsoning
#> 1 0018_01 Earth False ... 2823.0 0.0 Lerome Peckers
#> 2 0019_01 Europa True ... 0.0 0.0 Sabih Unhearfus
#> 3 0021_01 Europa False ... 181.0 585.0 Meratz Caltilter
#> 4 0023_01 Earth False ... 0.0 0.0 Brence Harperez
#>
#> [5 rows x 13 columns]
Exploratory Data Analysis (EDA)
train.info()
#> <class 'pandas.core.frame.DataFrame'>
#> RangeIndex: 8693 entries, 0 to 8692
#> Data columns (total 14 columns):
#> # Column Non-Null Count Dtype
#> --- ------ -------------- -----
#> 0 PassengerId 8693 non-null object
#> 1 HomePlanet 8492 non-null object
#> 2 CryoSleep 8476 non-null object
#> 3 Cabin 8494 non-null object
#> 4 Destination 8511 non-null object
#> 5 Age 8514 non-null float64
#> 6 VIP 8490 non-null object
#> 7 RoomService 8512 non-null float64
#> 8 FoodCourt 8510 non-null float64
#> 9 ShoppingMall 8485 non-null float64
#> 10 Spa 8510 non-null float64
#> 11 VRDeck 8505 non-null float64
#> 12 Name 8493 non-null object
#> 13 Transported 8693 non-null bool
#> dtypes: bool(1), float64(6), object(7)
#> memory usage: 891.5+ KB
Looking for NAs
If a feature has >= 30% nulls, we disregard it.
pd.Series(data = [train[col].isna().sum() / train[col].size * 100 for col in train.columns],
index = [col for col in train.columns],
name = "Percentage Missing").sort_values(ascending = False)
#> CryoSleep 2.496261
#> ShoppingMall 2.392730
#> VIP 2.335212
#> HomePlanet 2.312205
#> Name 2.300702
#> Cabin 2.289198
#> VRDeck 2.162660
#> FoodCourt 2.105142
#> Spa 2.105142
#> Destination 2.093639
#> RoomService 2.082135
#> Age 2.059128
#> PassengerId 0.000000
#> Transported 0.000000
#> Name: Percentage Missing, dtype: float64
With rangeIndex : 8693, it would seem that all our features are filled to a satisfactory level.
Checking Duplicates
duplicates = pd.DataFrame(train.loc[train.duplicated(subset=["Name"])].query("not Name.isnull()"))
print("Duplicate samples:\n"); duplicates[:5]
#> Duplicate samples:
#>
#> PassengerId HomePlanet CryoSleep ... VRDeck Name Transported
#> 956 1018_01 Earth True ... 0.0 Elaney Webstephrey True
#> 2700 2892_03 Earth True ... 0.0 Sharie Gallenry True
#> 2852 3081_01 Europa False ... 1.0 Gollux Reedall True
#> 2930 3176_01 Europa False ... 1464.0 Ankalik Nateansive True
#> 3291 3535_02 Mars True ... 0.0 Grake Porki True
#>
#> [5 rows x 14 columns]
print(f"Total amount of duplicate names: {len(duplicates.index)}")
#> Total amount of duplicate names: 20
train.query('Name == "Elaney Webstephrey"')
#> PassengerId HomePlanet CryoSleep ... VRDeck Name Transported
#> 156 0179_01 Earth False ... 11.0 Elaney Webstephrey True
#> 956 1018_01 Earth True ... 0.0 Elaney Webstephrey True
#>
#> [2 rows x 14 columns]
train.query('Name == "Sharie Gallenry"')
#> PassengerId HomePlanet CryoSleep ... VRDeck Name Transported
#> 1812 1935_01 Earth True ... 0.0 Sharie Gallenry False
#> 2700 2892_03 Earth True ... 0.0 Sharie Gallenry True
#>
#> [2 rows x 14 columns]
train.query('Name == "Gollux Reedall"')
#> PassengerId HomePlanet CryoSleep ... VRDeck Name Transported
#> 827 0881_01 Europa False ... 26.0 Gollux Reedall False
#> 2852 3081_01 Europa False ... 1.0 Gollux Reedall True
#>
#> [2 rows x 14 columns]
As the names are not a feature which our model will consider whilst learning, and the data for these duplicates differs, we’ll leave them in our model. Odds are they just duplicated by chance while the model was being generated.
Splitting the Cabin Feature
train.Cabin
#> 0 B/0/P
#> 1 F/0/S
#> 2 A/0/S
#> 3 A/0/S
#> 4 F/1/S
#> ...
#> 8688 A/98/P
#> 8689 G/1499/S
#> 8690 G/1500/S
#> 8691 E/608/S
#> 8692 E/608/S
#> Name: Cabin, Length: 8693, dtype: object
train[["CabinDeck", "CabinNum", "CabinSide"]] = train.Cabin.apply(
lambda cabin : pd.Series(
cabin.split("/") if not pd.isnull(cabin) else [float("NaN"), float("NaN"), float("NaN")]
)
)
test[["CabinDeck", "CabinNum", "CabinSide"]] = test.Cabin.apply(
lambda cabin : pd.Series(
cabin.split("/") if not pd.isnull(cabin) else [float("NaN"), float("NaN"), float("NaN")]
)
)
Univariate Analysis
Let’s separate numerical and categorical features so we can visualize them separately.
# Seperating categorical feature names and numerical feature names
categorical_features = [col for col in train.columns if train[col].dtype == 'object']
# We can remove the following three features from the list since we won't be dealing with those
categorical_features.remove("Name")
categorical_features.remove("Cabin")
categorical_features.remove("CabinNum")
numerical_features = [
col for col in train.columns if train[col].dtype == 'int64' or train[col].dtype == "float64"
]
print(f"categorical_features : {categorical_features}")
#> categorical_features : ['PassengerId', 'HomePlanet', 'CryoSleep', 'Destination', 'VIP', 'CabinDeck', 'CabinSide']
print(f"numerical_features : {numerical_features}")
#> numerical_features : ['Age', 'RoomService', 'FoodCourt', 'ShoppingMall', 'Spa', 'VRDeck']
# Seperating the data
train_cat = train[categorical_features]
train_num = train[numerical_features]
Numerical Data
We’ll be plotting the numerical data with Kernel Density Estimation (KDE).
train_num.columns.to_list()
#> ['Age', 'RoomService', 'FoodCourt', 'ShoppingMall', 'Spa', 'VRDeck']
for i in train_num.columns:
fig, (ax1, ax2) = plt.subplots(1, 2, figsize = (12, 4))
train_num[i].plot(kind = "kde", title = f"{i} KDE All Data", ax = ax1, )
if i == "Age":
ax1.set_xlabel("Age (Years)")
else:
ax2.set_xlabel("Money Spent (Space Currency)")
Q1 = train_num[i].quantile(0.25)
Q3 = train_num[i].quantile(0.75)
IQR = Q3-Q1
train_num.query(f'(@Q1 - 1.5 * @IQR) <= {i} <= (@Q3 + 1.5 * @IQR)')[i] \
.plot(kind = 'kde', color = 'm', title = f"{i} KDE IQR", ax = ax2)
if i == "Age":
ax2.set_xlabel("Age (Years)")
else:
ax2.set_xlabel("Money Spent (Space Currency)")
plt.tight_layout()
plt.show()
# Get X-axis ranges
range_ax1 = ax1.get_xlim()
range_ax2 = ax2.get_xlim()
# Print X-axis ranges
print(f"X-axis range for {i} Distribution All Data : {range_ax1}")
print(f"X-axis range for {i} Distribution IQR : {range_ax2}")
-
Agre Distribution: As the distrubution is already somewhat normalized, there was no major change between the “all data” distribution and the IQR distrubution. We’ll leave this this data as is without doing any transformations.
-
Rest of the Distrubutions: This distrubution is spiked highly at 0, presenting what looks like an incredibly rapid exponential decay. No transformation is needed when using decision or boost tree based models like XGBoost or CatBoost as they do not prefer Gaussian or standard probability distributions.
Feature Engineering
First we’ll clear the dataset to get it ready for training the model with the following:
- Imputation
- Categorical Encoding
First, let’s split the training data into features and target variables
X = train.drop(['Transported'], axis = 1).copy()
y = train.Transported.copy()
We’ll also pop off PassengerID as we’ll need it for making our submission.
test_IDs = test["PassengerId"]
X = X.drop("PassengerId", axis = 1)
test = test.drop("PassengerId", axis = 1)
categorical_features.remove("PassengerId")
Imputation
Imputation is a technique for handling missing data. If greater tahn 80% of our samples, it would be wise to drop it entirely. However, in our case we were only missing 2 to 2.5% of values.
Imputing Numerical Data
for i in numerical_features:
X[i] = X[i].fillna(X[i].median())
test[i] = test[i].fillna(X[i].median())
Now we check to make sure there aren’t still any missing values.
X[numerical_features].isnull().sum()
#> Age 0
#> RoomService 0
#> FoodCourt 0
#> ShoppingMall 0
#> Spa 0
#> VRDeck 0
#> dtype: int64
print()
test[numerical_features].isnull().sum()
#> Age 0
#> RoomService 0
#> FoodCourt 0
#> ShoppingMall 0
#> Spa 0
#> VRDeck 0
#> dtype: int64
Imputing Categorical Data
for i in categorical_features:
X[i] = X[i].fillna(X[i].mode().iloc[0])
test[i] = test[i].fillna(X[i].mode().iloc[0])
X[categorical_features].isnull().sum()
#> HomePlanet 0
#> CryoSleep 0
#> Destination 0
#> VIP 0
#> CabinDeck 0
#> CabinSide 0
#> dtype: int64
print()
test[categorical_features].isnull().sum()
#> HomePlanet 0
#> CryoSleep 0
#> Destination 0
#> VIP 0
#> CabinDeck 0
#> CabinSide 0
#> dtype: int64
Categorical Encoding
Categorical encoding is the process of converting categorical data to numerical values for training models.
First off, let’s check the heads of our training dataset and our test dataset.
X.head(5)
#> HomePlanet CryoSleep Cabin ... CabinDeck CabinNum CabinSide
#> 0 Europa False B/0/P ... B 0 P
#> 1 Earth False F/0/S ... F 0 S
#> 2 Europa False A/0/S ... A 0 S
#> 3 Europa False A/0/S ... A 0 S
#> 4 Earth False F/1/S ... F 1 S
#>
#> [5 rows x 15 columns]
test.head(5)
#> HomePlanet CryoSleep Cabin ... CabinDeck CabinNum CabinSide
#> 0 Earth True G/3/S ... G 3 S
#> 1 Earth False F/4/S ... F 4 S
#> 2 Europa True C/0/S ... C 0 S
#> 3 Europa False C/1/S ... C 1 S
#> 4 Earth False F/5/S ... F 5 S
#>
#> [5 rows x 15 columns]
Before categorical encoding, let’s remove the Name and CabinNum columns from the dataframes, since they have way too many unique values or provide way to little information for the model.
print(numerical_features + categorical_features)
#> ['Age', 'RoomService', 'FoodCourt', 'ShoppingMall', 'Spa', 'VRDeck', 'HomePlanet', 'CryoSleep', 'Destination', 'VIP', 'CabinDeck', 'CabinSide']
X = X[numerical_features + categorical_features]
test = test[numerical_features + categorical_features]
Now, let’s have a look at the different categorical features and how many unique values they each have (if using one-hot encoding, they should preferably have less than 15 unique values):
for i in categorical_features:
print(f"Feature name: {i}. \nNo. of unique values: {X[i].nunique()} ({X[i].unique()})\n\n")
#> Feature name: HomePlanet.
#> No. of unique values: 3 (['Europa' 'Earth' 'Mars'])
#>
#>
#> Feature name: CryoSleep.
#> No. of unique values: 2 ([False True])
#>
#>
#> Feature name: Destination.
#> No. of unique values: 3 (['TRAPPIST-1e' 'PSO J318.5-22' '55 Cancri e'])
#>
#>
#> Feature name: VIP.
#> No. of unique values: 2 ([False True])
#>
#>
#> Feature name: CabinDeck.
#> No. of unique values: 8 (['B' 'F' 'A' 'G' 'E' 'D' 'C' 'T'])
#>
#>
#> Feature name: CabinSide.
#> No. of unique values: 2 (['P' 'S'])
All the features are well whitin the threshold, so we can use one-hot encoding for all of them. However, we’ll use ordinal encoding for CabinDeck as it will group related values.
Before we start encoding let’s take a quick look at the datataypes again:
X.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> HomePlanet object
#> CryoSleep bool
#> Destination object
#> VIP bool
#> CabinDeck object
#> CabinSide object
#> dtype: object
CryoSleep and VIP are of type boll, not object. We can turn those columns into 1s and 0s as so:
X[["CryoSleep", "VIP"]] = X[["CryoSleep", "VIP"]].astype(int)
test[["CryoSleep", "VIP"]] = test[["CryoSleep", "VIP"]].astype(int)
X.head(5)
#> Age RoomService FoodCourt ... VIP CabinDeck CabinSide
#> 0 39.0 0.0 0.0 ... 0 B P
#> 1 24.0 109.0 9.0 ... 0 F S
#> 2 58.0 43.0 3576.0 ... 1 A S
#> 3 33.0 0.0 1283.0 ... 0 A S
#> 4 16.0 303.0 70.0 ... 0 F S
#>
#> [5 rows x 12 columns]
test.head(5)
#> Age RoomService FoodCourt ... VIP CabinDeck CabinSide
#> 0 27.0 0.0 0.0 ... 0 G S
#> 1 19.0 0.0 9.0 ... 0 F S
#> 2 31.0 0.0 0.0 ... 0 C S
#> 3 38.0 0.0 6652.0 ... 0 C S
#> 4 20.0 10.0 0.0 ... 0 F S
#>
#> [5 rows x 12 columns]
X.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> HomePlanet object
#> CryoSleep int64
#> Destination object
#> VIP int64
#> CabinDeck object
#> CabinSide object
#> dtype: object
Now to separate CabinDeck from the other categorical features:
ordinal_column_names = ["CabinDeck"]
one_hot_column_names = ["HomePlanet", "Destination", "CabinSide"]
Ordianl Encoding
We’ll just use a mapping dictionary so we can ordinal encode using just Pandas.
deck_mapping = {"A": 1, "B": 2, "C": 3, "D": 4, "E": 5, "F": 6, "G": 7, "T": 8}
X[ordinal_column_names] = X[ordinal_column_names].replace(deck_mapping)
test[ordinal_column_names] = test[ordinal_column_names].replace(deck_mapping)
X.head(5); test.head(5)
#> Age RoomService FoodCourt ... VIP CabinDeck CabinSide
#> 0 39.0 0.0 0.0 ... 0 2 P
#> 1 24.0 109.0 9.0 ... 0 6 S
#> 2 58.0 43.0 3576.0 ... 1 1 S
#> 3 33.0 0.0 1283.0 ... 0 1 S
#> 4 16.0 303.0 70.0 ... 0 6 S
#>
#> [5 rows x 12 columns]
#> Age RoomService FoodCourt ... VIP CabinDeck CabinSide
#> 0 27.0 0.0 0.0 ... 0 7 S
#> 1 19.0 0.0 9.0 ... 0 6 S
#> 2 31.0 0.0 0.0 ... 0 3 S
#> 3 38.0 0.0 6652.0 ... 0 3 S
#> 4 20.0 10.0 0.0 ... 0 6 S
#>
#> [5 rows x 12 columns]
As T is so small, I’m not really worried about it hampering the model.
Categorical Encoding
X = pd.get_dummies(X).copy()
test = pd.get_dummies(test).copy()
X.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> CryoSleep int64
#> VIP int64
#> CabinDeck int64
#> HomePlanet_Earth bool
#> HomePlanet_Europa bool
#> HomePlanet_Mars bool
#> Destination_55 Cancri e bool
#> Destination_PSO J318.5-22 bool
#> Destination_TRAPPIST-1e bool
#> CabinSide_P bool
#> CabinSide_S bool
#> dtype: object
print()
test.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> CryoSleep int64
#> VIP int64
#> CabinDeck int64
#> HomePlanet_Earth bool
#> HomePlanet_Europa bool
#> HomePlanet_Mars bool
#> Destination_55 Cancri e bool
#> Destination_PSO J318.5-22 bool
#> Destination_TRAPPIST-1e bool
#> CabinSide_P bool
#> CabinSide_S bool
#> dtype: object
# bool column convert to unit8
bool_columns = X.select_dtypes(include = 'bool').columns
X[bool_columns] = X[bool_columns].astype('uint8')
test[bool_columns] = test[bool_columns].astype('uint8')
X.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> CryoSleep int64
#> VIP int64
#> CabinDeck int64
#> HomePlanet_Earth uint8
#> HomePlanet_Europa uint8
#> HomePlanet_Mars uint8
#> Destination_55 Cancri e uint8
#> Destination_PSO J318.5-22 uint8
#> Destination_TRAPPIST-1e uint8
#> CabinSide_P uint8
#> CabinSide_S uint8
#> dtype: object
print()
test.dtypes
#> Age float64
#> RoomService float64
#> FoodCourt float64
#> ShoppingMall float64
#> Spa float64
#> VRDeck float64
#> CryoSleep int64
#> VIP int64
#> CabinDeck int64
#> HomePlanet_Earth uint8
#> HomePlanet_Europa uint8
#> HomePlanet_Mars uint8
#> Destination_55 Cancri e uint8
#> Destination_PSO J318.5-22 uint8
#> Destination_TRAPPIST-1e uint8
#> CabinSide_P uint8
#> CabinSide_S uint8
#> dtype: object
Model Training
Imports
import torch
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader, TensorDataset
from sklearn.metrics import roc_auc_score, confusion_matrix, roc_curve
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
Preparing Data for Training
Now we need to split the data into the training set, validation set, and test set. Furthermore we need to convert the data which is currently stored in dataframes, into Pytorch Tensors.
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
X_train, X_valid, y_train, y_valid = train_test_split(X_train, y_train, test_size=0.2, random_state=42)
# Convert the data to Numpy arrays and then to Pytorch tensors
X_train = torch.tensor(np.array(X_train), dtype=torch.float32)
X_valid = torch.tensor(np.array(X_valid), dtype=torch.float32)
X_test = torch.tensor(np.array(X_test), dtype=torch.float32)
y_train = torch.tensor(np.array(y_train), dtype = torch.float32)
y_valid = torch.tensor(np.array(y_valid), dtype = torch.float32)
y_test = torch.tensor(np.array(y_test), dtype = torch.float32)
We also need to create a DataLoader for each dataset.
train_dataset = TensorDataset(X_train, y_train)
valid_dataset = TensorDataset(X_valid, y_valid)
test_dataset = TensorDataset(X_test, y_test)
train_loader = DataLoader(train_dataset, batch_size = 64, shuffle = True)
valid_loader = DataLoader(valid_dataset, batch_size = 64)
test_loader = DataLoader(test_dataset, batch_size = 64)
Define the FCNN
Here we define the FCNN. Note that we’re also using th dropout regularization technique to minimize overfitting.
class FCNN(nn.Module):
def __init__(self, input_size):
super(FCNN, self).__init__()
self.fc1 = nn.Linear(input_size, 128)
self.fc2 = nn.Linear(128, 64)
self.fc3 = nn.Linear(64, 32)
self.fc4 = nn.Linear(32, 1)
self.dropout = nn.Dropout(0.5)
self.sigmoid = nn.Sigmoid()
def forward(self, x):
x = torch.relu(self.fc1(x))
x = self.dropout(x)
x = torch.relu(self.fc2(x))
x = self.dropout(x)
x = torch.relu(self.fc3(x))
x = self.dropout(x)
x = self.fc4(x)
x = self.sigmoid(x)
return(x)
Initialize the Model
Here we do a few things:
- Set the Device : Here we set the device to the CUDA enabled GPU (if present) or the CPU.
- Initialize the Model
- Define the Loss Function: Here we use Binary Crossentropy(BCE) to compare the value predicted by the model and the target value.
- Define the Optimizer and Scheduler: Here we set the attributes of the model. We use Root mean Square Propagation (RMSprop) to set the minibatches, and we add a weight decay for L2 regularization to minimize overfitting.
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
input_size = X_train.shape[1]
model = FCNN(input_size).to(device)
criterion = nn.BCELoss()
optimizer = optim.RMSprop(model.parameters(), lr = 0.001, weight_decay = 1e-5) # Added weight_decay for L2 regularization
scheduler = optim.lr_scheduler.StepLR(optimizer, step_size = 30, gamma = 0.1)
Training & Validation Loop
We implement early stops to once agine, minimie the chances of overfitting.
num_epochs = 100
best_loss = float('inf')
train_loss_values = []
valid_loss_values = []
for epoch in range(num_epochs):
model.train()
train_loss = 0.0
for batch_x, batch_y in train_loader:
batch_x = batch_x.to(device)
batch_y = batch_y.to(device)
optimizer.zero_grad()
outputs = model(batch_x)
loss = criterion(outputs.squeeze(), batch_y)
loss.backward()
optimizer.step()
train_loss += loss.item() * batch_x.size(0)
scheduler.step() # Decay learning rate
# Validation
model.eval()
valid_loss = 0.0
with torch.no_grad():
for batch_x, batch_y in valid_loader:
batch_x = batch_x.to(device)
batch_y = batch_y.to(device)
outputs = model(batch_x)
loss = criterion(outputs.squeeze(), batch_y)
valid_loss += loss.item() * batch_x.size(0)
train_loss /= len(train_loader.dataset)
valid_loss /= len(valid_loader.dataset)
train_loss_values.append(train_loss)
valid_loss_values.append(valid_loss)
print(f"Epoch {epoch+1}/{num_epochs}, Train Loss: {train_loss:.4f}, Valid Loss: {valid_loss:.4f}")
# Save the model if validation loss improves
if valid_loss < best_loss:
best_loss = valid_loss
torch.save(model.state_dict(), os.path.join(os.getcwd(), "model", "fcnn_model.pt"))
else:
print("No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.")
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 1/100, Train Loss: 5.4054, Valid Loss: 0.5135
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 2/100, Train Loss: 1.3786, Valid Loss: 0.5932
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 3/100, Train Loss: 0.7172, Valid Loss: 0.6003
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 4/100, Train Loss: 0.6188, Valid Loss: 0.5659
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 5/100, Train Loss: 0.6165, Valid Loss: 0.5404
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 6/100, Train Loss: 0.5798, Valid Loss: 0.5351
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 7/100, Train Loss: 0.5659, Valid Loss: 0.5017
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 8/100, Train Loss: 0.5823, Valid Loss: 0.4927
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 9/100, Train Loss: 0.5620, Valid Loss: 0.4810
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 10/100, Train Loss: 0.5318, Valid Loss: 0.4694
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 11/100, Train Loss: 0.5406, Valid Loss: 0.4685
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 12/100, Train Loss: 0.5076, Valid Loss: 0.4495
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 13/100, Train Loss: 0.5133, Valid Loss: 0.4459
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 14/100, Train Loss: 0.5051, Valid Loss: 0.4596
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 15/100, Train Loss: 0.5047, Valid Loss: 0.4547
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 16/100, Train Loss: 0.5070, Valid Loss: 0.4480
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 17/100, Train Loss: 0.5005, Valid Loss: 0.4441
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 18/100, Train Loss: 0.4926, Valid Loss: 0.4462
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 19/100, Train Loss: 0.5040, Valid Loss: 0.4507
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 20/100, Train Loss: 0.4832, Valid Loss: 0.4430
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 21/100, Train Loss: 0.4780, Valid Loss: 0.4379
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 22/100, Train Loss: 0.5010, Valid Loss: 0.4574
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 23/100, Train Loss: 0.4925, Valid Loss: 0.4411
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 24/100, Train Loss: 0.4823, Valid Loss: 0.4444
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 25/100, Train Loss: 0.4729, Valid Loss: 0.4399
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 26/100, Train Loss: 0.4786, Valid Loss: 0.4413
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 27/100, Train Loss: 0.4755, Valid Loss: 0.4402
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 28/100, Train Loss: 0.4640, Valid Loss: 0.4549
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 29/100, Train Loss: 0.4676, Valid Loss: 0.4393
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 30/100, Train Loss: 0.4782, Valid Loss: 0.4372
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 31/100, Train Loss: 0.4668, Valid Loss: 0.4409
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 32/100, Train Loss: 0.4766, Valid Loss: 0.4474
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 33/100, Train Loss: 0.4667, Valid Loss: 0.4434
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 34/100, Train Loss: 0.4635, Valid Loss: 0.4460
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 35/100, Train Loss: 0.4619, Valid Loss: 0.4439
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 36/100, Train Loss: 0.4687, Valid Loss: 0.4499
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 37/100, Train Loss: 0.4682, Valid Loss: 0.4447
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 38/100, Train Loss: 0.4776, Valid Loss: 0.4407
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 39/100, Train Loss: 0.4614, Valid Loss: 0.4407
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 40/100, Train Loss: 0.4552, Valid Loss: 0.4394
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 41/100, Train Loss: 0.4589, Valid Loss: 0.4428
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 42/100, Train Loss: 0.4662, Valid Loss: 0.4409
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 43/100, Train Loss: 0.4645, Valid Loss: 0.4446
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 44/100, Train Loss: 0.4563, Valid Loss: 0.4377
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 45/100, Train Loss: 0.4598, Valid Loss: 0.4412
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 46/100, Train Loss: 0.4566, Valid Loss: 0.4376
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 47/100, Train Loss: 0.4696, Valid Loss: 0.4396
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 48/100, Train Loss: 0.4630, Valid Loss: 0.4455
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 49/100, Train Loss: 0.4554, Valid Loss: 0.4485
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 50/100, Train Loss: 0.4566, Valid Loss: 0.4418
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 51/100, Train Loss: 0.4584, Valid Loss: 0.4401
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 52/100, Train Loss: 0.4601, Valid Loss: 0.4420
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 53/100, Train Loss: 0.4635, Valid Loss: 0.4373
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 54/100, Train Loss: 0.4572, Valid Loss: 0.4360
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 55/100, Train Loss: 0.4510, Valid Loss: 0.4331
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 56/100, Train Loss: 0.4581, Valid Loss: 0.4382
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 57/100, Train Loss: 0.4607, Valid Loss: 0.4362
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 58/100, Train Loss: 0.4532, Valid Loss: 0.4415
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 59/100, Train Loss: 0.4585, Valid Loss: 0.4390
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 60/100, Train Loss: 0.4549, Valid Loss: 0.4392
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 61/100, Train Loss: 0.4553, Valid Loss: 0.4390
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 62/100, Train Loss: 0.4573, Valid Loss: 0.4385
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 63/100, Train Loss: 0.4585, Valid Loss: 0.4388
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 64/100, Train Loss: 0.4532, Valid Loss: 0.4379
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 65/100, Train Loss: 0.4627, Valid Loss: 0.4380
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 66/100, Train Loss: 0.4511, Valid Loss: 0.4376
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 67/100, Train Loss: 0.4567, Valid Loss: 0.4374
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 68/100, Train Loss: 0.4576, Valid Loss: 0.4373
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 69/100, Train Loss: 0.4722, Valid Loss: 0.4368
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 70/100, Train Loss: 0.4534, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 71/100, Train Loss: 0.4522, Valid Loss: 0.4360
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 72/100, Train Loss: 0.4558, Valid Loss: 0.4357
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 73/100, Train Loss: 0.4590, Valid Loss: 0.4363
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 74/100, Train Loss: 0.4539, Valid Loss: 0.4360
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 75/100, Train Loss: 0.4572, Valid Loss: 0.4362
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 76/100, Train Loss: 0.4549, Valid Loss: 0.4361
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 77/100, Train Loss: 0.4536, Valid Loss: 0.4362
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 78/100, Train Loss: 0.4603, Valid Loss: 0.4363
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 79/100, Train Loss: 0.4514, Valid Loss: 0.4364
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 80/100, Train Loss: 0.4593, Valid Loss: 0.4366
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 81/100, Train Loss: 0.4577, Valid Loss: 0.4366
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 82/100, Train Loss: 0.4543, Valid Loss: 0.4362
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 83/100, Train Loss: 0.4574, Valid Loss: 0.4365
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 84/100, Train Loss: 0.4553, Valid Loss: 0.4364
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 85/100, Train Loss: 0.4610, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 86/100, Train Loss: 0.4551, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 87/100, Train Loss: 0.4544, Valid Loss: 0.4371
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 88/100, Train Loss: 0.4610, Valid Loss: 0.4372
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 89/100, Train Loss: 0.4566, Valid Loss: 0.4373
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 90/100, Train Loss: 0.4549, Valid Loss: 0.4368
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 91/100, Train Loss: 0.4535, Valid Loss: 0.4368
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 92/100, Train Loss: 0.4696, Valid Loss: 0.4369
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 93/100, Train Loss: 0.4527, Valid Loss: 0.4369
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 94/100, Train Loss: 0.4558, Valid Loss: 0.4369
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 95/100, Train Loss: 0.4515, Valid Loss: 0.4368
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 96/100, Train Loss: 0.4535, Valid Loss: 0.4368
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 97/100, Train Loss: 0.4522, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 98/100, Train Loss: 0.4590, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 99/100, Train Loss: 0.4568, Valid Loss: 0.4367
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
#> Epoch 100/100, Train Loss: 0.4557, Valid Loss: 0.4366
#> No improvement in validation loss for this epoch. Model parameters from last epoch with best validation loss were saved.
Evalute the Model on Test set
First, let’s load the best model.
model.load_state_dict(torch.load(os.path.join(os.getcwd(), "model", "fcnn_model.pt")))
#> <All keys matched successfully>
Now we run the evaluation loop in the test set.
model.eval()
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
test_loss = 0.0
test_predictions = []
with torch.no_grad():
for batch_x, batch_y in test_loader:
batch_x = batch_x.to(device)
batch_y = batch_y.to(device)
outputs = model(batch_x)
loss = criterion(outputs.squeeze(), batch_y)
test_loss += loss.item() * batch_x.size(0)
test_predictions.extend(outputs.cpu().numpy())
test_loss /= len(test_loader.dataset)
test_predictions = np.array(test_predictions)
# Calculate AUC-ROC for the test set
test_auc_roc = roc_auc_score(y_test, test_predictions)
print(f"Test Loss : {test_loss:.4f}")
#> Test Loss : 0.4451
print(f"Test AUC-ROC: {test_auc_roc:.4f}")
#> Test AUC-ROC: 0.8710
print(y_test)
#> tensor([1., 0., 0., ..., 0., 1., 0.])
print(test_predictions)
#> [[0.2430638 ]
#> [0.5769563 ]
#> [0.60590726]
#> ...
#> [0.41472107]
#> [0.7937017 ]
#> [0.03511113]]
Visualization
Plot Training & Validation Loss
plt.figure(figsize = (8, 6))
plt.plot(train_loss_values, label = "Training Loss")
plt.plot(valid_loss_values, label = "Validation Loss")
plt.xlabel("Epoch")
plt.ylabel("Loss")
plt.title("Training and Validation Loss over Epochs")
plt.legend()
plt.grid(True)
plt.show()
ROC Curve
fpr, tpr, _ = roc_curve(y_test, test_predictions)
plt.figure(figsize = (8, 6))
#> <Figure size 800x600 with 0 Axes>
plt.plot(fpr, tpr, label = f"AUC-ROC = {test_auc_roc:.4f}")
#> [<matplotlib.lines.Line2D object at 0x7fd020384040>]
plt.plot([0, 1], [0, 1], linestyle = "--", color = "r")
#> [<matplotlib.lines.Line2D object at 0x7fd0203aef20>]
plt.xlabel("False Positive Rate")
#> Text(0.5, 0, 'False Positive Rate')
plt.ylabel("True Positive Rate")
#> Text(0, 0.5, 'True Positive Rate')
plt.title("ROC Curve")
#> Text(0.5, 1.0, 'ROC Curve')
plt.legend()
#> <matplotlib.legend.Legend object at 0x7fd02059fbb0>
plt.grid(True)
plt.show()
Confusion Matrix
plt.figure()
#> <Figure size 700x500 with 0 Axes>
cm = confusion_matrix(y_test.numpy(), test_predictions.round())
sns.heatmap(cm, annot = True, fmt = "d")
#> <Axes: >
plt.title("Confusion Matrix")
#> Text(0.5, 1.0, 'Confusion Matrix')
plt.xlabel("Predicted Label")
#> Text(0.5, 5.583333333333313, 'Predicted Label')
plt.ylabel("True Label")
#> Text(26.58333333333333, 0.5, 'True Label')
plt.show()
Classification Accuracy
binary_predictions = [1 if p>= 0.5 else 0 for p in test_predictions]
test_accuracy = accuracy_score(y_test.numpy(), binary_predictions)
print(f"Test Accuracy : {test_accuracy:.4f}")
#> Test Accuracy : 0.7849
Submission
train = pd.read_csv(os.path.join(os.getcwd(), "data", "train.csv"))
submit = pd.read_csv(os.path.join(os.getcwd(), "data", "sample_submission.csv"))
test_IDs = submit['PassengerId']
test_competition = torch.tensor(np.array(test), dtype = torch.float32).to(device)
model.eval()
#> FCNN(
#> (fc1): Linear(in_features=17, out_features=128, bias=True)
#> (fc2): Linear(in_features=128, out_features=64, bias=True)
#> (fc3): Linear(in_features=64, out_features=32, bias=True)
#> (fc4): Linear(in_features=32, out_features=1, bias=True)
#> (dropout): Dropout(p=0.5, inplace=False)
#> (sigmoid): Sigmoid()
#> )
test_competition_predictions = []
with torch.no_grad():
outputs = model(test_competition)
test_competition_predictions.extend(outputs.cpu().numpy())
test_competition_predictions = np.array(test_competition_predictions).flatten()
binary_competition_predictions = [True if p >= 0.5 else False for p in test_competition_predictions]
submission = pd.DataFrame({
"PassengerId" : test_IDs,
"Transported" : binary_competition_predictions
})
submission.to_csv(os.path.join(os.getcwd(), "model", "submission.csv"), index = False)