File size: 2,637 Bytes
08c038b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 | import torch
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader
from sklearn.metrics import roc_auc_score, f1_score, classification_report
import joblib
import os
# Components
from model import HybridTabTransformer
from dataset import HeartDiseaseDataset
def train_model():
# Setup & Hyperparameters
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
metadata = joblib.load('assets/model_metadata.joblib')
batch_size = 32
epochs = 20
lr = 0.001
# Data Loaders
train_ds = HeartDiseaseDataset('data/processed/train.csv')
test_ds = HeartDiseaseDataset('data/processed/test.csv')
train_loader = DataLoader(train_ds, batch_size=batch_size, shuffle=True)
test_loader = DataLoader(test_ds, batch_size=batch_size)
# Instantiate Model
model = HybridTabTransformer(
cat_dims=metadata['cat_dims'],
num_continuous=len(metadata['num_cols'])
).to(device)
criterion = nn.BCELoss()
optimizer = optim.Adam(model.parameters(), lr=lr)
print(f"Starting training on {device}...")
# Train the model
for epoch in range(epochs):
model.train()
total_loss = 0
for x_cat, x_num, y in train_loader:
x_cat, x_num, y = x_cat.to(device), x_num.to(device), y.to(device)
optimizer.zero_grad()
outputs = model(x_cat, x_num)
loss = criterion(outputs, y)
loss.backward()
optimizer.step()
total_loss += loss.item()
if (epoch + 1) % 5 == 0:
print(f"Epoch [{epoch+1}/{epochs}], Loss: {total_loss/len(train_loader):.4f}")
# Evaluate the model
model.eval()
all_preds = []
all_targets = []
with torch.no_grad():
for x_cat, x_num, y in test_loader:
x_cat, x_num, y = x_cat.to(device), x_num.to(device), y.to(device)
outputs = model(x_cat, x_num)
all_preds.extend(outputs.cpu().numpy())
all_targets.extend(y.cpu().numpy())
# Convert to binary for F1/Classification Report
binary_preds = [1 if p >= 0.5 else 0 for p in all_preds]
print("\n--- Final Model Evaluation ---")
print(f"AUROC Score: {roc_auc_score(all_targets, all_preds):.4f}")
print(f"F1 Score: {f1_score(all_targets, binary_preds):.4f}")
print("\nClassification Report:")
print(classification_report(all_targets, binary_preds))
# Save Model
torch.save(model.state_dict(), 'assets/model.pth')
print("Model saved to assets/model.pth")
if __name__ == "__main__":
train_model() |