flashback to 2021 when i made neural network for 4bit inputs from scratch
this projec wast test if i could use my knowlage about neural nets and training algorythms to make the simplest neural network and train it
the idea : make neural net that detects logic in examples and repeats it:
these are the training data:
for 4 bit inputs we have total of 16 possible states curently we are training on dataset with size of 8
this is our structure of the network:
firs we will make random weights and bias:
then we train our network by adjusting our weights and bias by smaller and smaller amounts based on number of epochs and our learning rate:
and then we chose weights and biases that are better by good gueseses and error:
at the end we graph it and do lil demo for testing:
