ium_464914/sacred_model.py
Alicja Szulecka 7ff2f9711e sacred
2024-05-05 14:07:27 +02:00

126 lines
4.3 KiB
Python

import torch
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader, Dataset
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
import torch.nn.functional as F
from sacred import Experiment
from sacred.observers import FileStorageObserver, MongoObserver
device = (
"cuda"
if torch.cuda.is_available()
else "cpu"
)
ex = Experiment("464914", interactive=True, save_git_info=False)
ex.observers.append(FileStorageObserver('experiments'))
ex.observers.append(MongoObserver(url='mongodb://admin:IUM_2021@tzietkiewicz.vm.wmi.amu.edu.pl:27017',
db_name='sacred'))
class Model(nn.Module):
def __init__(self, input_features=54, hidden_layer1=25, hidden_layer2=30, output_features=8):
super().__init__()
self.fc1 = nn.Linear(input_features,output_features)
self.bn1 = nn.BatchNorm1d(hidden_layer1) # Add batch normalization
self.fc2 = nn.Linear(hidden_layer1, hidden_layer2)
self.bn2 = nn.BatchNorm1d(hidden_layer2) # Add batch normalization
self.out = nn.Linear(hidden_layer2, output_features)
def forward(self, x):
x = F.relu(self.fc1(x)) # Apply batch normalization after first linear layer
#x = F.relu(self.bn2(self.fc2(x))) # Apply batch normalization after second linear layer
#x = self.out(x)
return x
@ex.capture
def capture_params(epochs):
print(f"epochs: {epochs}")
@ex.main
def main(_run):
forest_train_ex = ex.open_resource('forest_train.csv')
forest_val_ex = ex.open_resource('forest_val.csv')
forest_val = pd.read_csv('forest_val.csv')
forest_train = pd.read_csv('forest_train.csv')
X_train = forest_train.drop(columns=['Cover_Type']).values
y_train = forest_train['Cover_Type'].values
X_val = forest_val.drop(columns=['Cover_Type']).values
y_val = forest_val['Cover_Type'].values
# Initialize model, loss function, and optimizer
model = Model().to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)
# Convert to PyTorch tensors
X_train = torch.tensor(X_train, dtype=torch.float32).to(device)
y_train = torch.tensor(y_train, dtype=torch.long).to(device)
X_val = torch.tensor(X_val, dtype=torch.float32).to(device)
y_val = torch.tensor(y_val, dtype=torch.long).to(device)
# Create DataLoader
train_loader = DataLoader(list(zip(X_train, y_train)), batch_size=64, shuffle=True)
val_loader = DataLoader(list(zip(X_val, y_val)), batch_size=64)
# Training loop
epochs = 10
for epoch in range(epochs):
model.train() # Set model to training mode
running_loss = 0.0
for inputs, labels in train_loader:
inputs, labels = inputs.to(device), labels.to(device)
optimizer.zero_grad()
outputs = model(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
running_loss += loss.item() * inputs.size(0)
# Calculate training loss
epoch_loss = running_loss / len(train_loader.dataset)
# Validation
model.eval() # Set model to evaluation mode
val_running_loss = 0.0
correct = 0
total = 0
with torch.no_grad():
for inputs, labels in val_loader:
inputs, labels = inputs.to(device), labels.to(device)
outputs = model(inputs)
val_loss = criterion(outputs, labels)
val_running_loss += val_loss.item() * inputs.size(0)
_, predicted = torch.max(outputs, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
# Calculate validation loss and accuracy
val_epoch_loss = val_running_loss / len(val_loader.dataset)
val_accuracy = correct / total
print(f"Epoch {epoch+1}/{epochs}, "
f"Train Loss: {epoch_loss:.4f}, "
f"Val Loss: {val_epoch_loss:.4f}, "
f"Val Accuracy: {val_accuracy:.4f}")
_run.log_scalar("train loss", epoch_loss)
_run.log_scalar("val loss", val_epoch_loss)
capture_params(epochs)
torch.save(model.state_dict(), 'model.pth')
ex.add_artifact("model.pth")
ex.run()