/Handwritten-Digits-Classifier

Uses the MNIST data set and the LeNet-5 architecture to build a handwritten digits classifier.

Primary LanguageJupyter Notebook

Handwritten-Digits-Classifier

This classifier uses the MNIST dataset and the LeNet-5 architecture.

As this is usually a beginner's project, I have another "Explained" version of the main notebook that has copious comments and additional code to help understand the basics of PyTorch better. This also helps those who have started their Deep Learning journey.

Results

  • Test set accuracy: ~98.7%
  • Train set accuracy: ~99%+

References