Summary  
Shows how to evaluate a trained model by switching to evaluation mode, disabling gradient tracking, performing inference to obtain logits, converting logits into class labels, and computing accuracy.

General domain of usage  
Machine learning model evaluation

## Preparing for Evaluation

Before starting the evaluation process on the **test set**, you need to ensure the following:

1. **Set the model to evaluation mode**: use `model.eval()` to turn off features like dropout and batch normalization, ensuring consistent behavior during evaluation;

2. **Disable gradient tracking**: use `torch.no_grad()` to save memory and speed up computations, as gradients are not required during evaluation.

```python
# Set the model to evaluation mode
model.eval()
# Disable gradient computation for evaluation
with torch.no_grad():
    # Forward pass on the test data
    test_predictions = model(X_test)
```

## Converting Predictions

As we've already mentioned previously, the output from the model will be **logits** (raw scores). To get the predicted class labels, we use `torch.argmax` to extract the **index of the maximum value** along the class dimension.

```python
# Convert logits to predicted class labels
predicted_labels = torch.argmax(test_predictions, dim=1)
```

## Calculating Metrics

For classification problems, **accuracy** is a useful starting metric, provided the dataset is balanced.

```python
# Calculate accuracy
correct_predictions = (predicted_labels == y_test).sum().item()
accuracy = correct_predictions / len(y_test) * 100
print(f"Test accuracy: {accuracy:.2f}%")
```

To gain deeper insights into model performance, you can calculate additional metrics such as **precision**, **recall**, and **F1-score**. You can learn more about these metrics and their formulas in <a href="https://codefinity.com/blog/Understanding-Classification-Metrics" target="_blank">this article</a>, using their respective formulas.

import torch
import os
os.system('wget https://staging-content-media-cdn.codefinity.com/courses/1dd2b0f6-6ec0-40e6-a570-ed0ac2209666/section_3/model_training.py 2>/dev/null')
from model_training import model, X_test, y_test

# Set model to evaluation mode
model.eval() 
# Disable gradient tracking
with torch.no_grad():
    # Forward pass
    test_predictions = model(X_test)
    # Get predicted classes
    predicted_labels = torch.argmax(test_predictions, dim=1)

# Calculate accuracy
correct_predictions = (predicted_labels == y_test).sum().item()
accuracy = correct_predictions / len(y_test) * 100
print(f"Test accuracy: {accuracy:.2f}%")

Which of the following steps is necessary when evaluating a trained PyTorch model?

Learn the fundamental and advanced concepts needed to work with PyTorch efficiently. Gain a solid understanding tensors, including creation, operations, and reshaping. Explore the essentials of gradients, backpropagation, and linear regression before moving on to handling datasets. Master the skill needed build, train, and evaluate neural networks.

Explore the fundamentals of PyTorch, focusing on tensors—the core data structure used for computations. You will learn about tensor creation, random initialization, mathematical operations, and shape manipulation.

Explore key concepts for training models in PyTorch, including computing gradients and performing multi-step backpropagation. Master linear regression as a foundational machine learning model and introduces handling datasets efficiently.

Discover how to build, train, and evaluate neural networks using PyTorch. You will learn how to define a simple feedforward neural network, optimize its parameters through training, and assess its performance.

Evaluation of the Model

Preparing for Evaluation

Converting Predictions

Calculating Metrics

Full Implementation