Skip to content

Fully connected layer from scratch, with training on MNIST dataset

Notifications You must be signed in to change notification settings

YasinShafiei/FullyConnected-from-scratch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 

Repository files navigation

FullyConnected neural network layer from scratch

Fully connected layer from scratch, with training on MNIST dataset.
The scripts are all for creating a Fully-Connected Neural Network layer and making a custom model.
Also the model trains on fully custom training loop. All the optimizer and loss function are written from scratch.

Results

I implemented a MLP from scratch only using NumPy for model creation. Here are the results of the training:
Screenshot 2024-01-05 161138 Figure_1 Figure_2

Article:

Please go and checkout my article about this project for full explanations and detils: https://medium.com/@YasinShafiei/making-a-neural-network-fully-connected-layer-from-scratch-only-numpy-49bd7958b6f3