Generative AI

Analytical AI has few output neurons, GAI has many

Encoder-Decoder CNN

DC-GANs

Two deep neural-nets, compete against each other.

• The generator tries to produce fake images.

• The discriminator will try and disprove the image, it will typically have images to compair against.

Generator example

• Takes a random vector of numbers

• Transpose will sharpen the image

Discriminator

Conditional GAN

• Essentially the same as above, but will take text as input (aswell as random stuff, I think).

How to train a GPT language model

FeedForwards neural network

• The data flows through the net work in one direction, from the input to output layers

Recurrent Neural Networks (RNN)

• Frame - a time step

• Some layers point backwards

• Each cycle has a delay → processes in different framees → this is good for processing a sequence

• Each cycle has a state → the system has a memory → content can be retained between cycles
  • The memory essentially emerges from it being recurrent

• The output is based on
  • Current input
  • Previous output

Nerdy way of showing it

• A Cell, is a group of neurons that can perserve its state

Types of RNN

Illustrated Guide to Transformers Neural Network: A step by step explanation

Transformers are the rage nowadays, but how do they work? This video demystifies the novel neural network architecture with step by step explanation and illustrations on how transformers work. CORRECTIONS: The sine and cosine functions are actually applied to the embedding dimensions and time steps! ⭐ Play and Experiment With the Latest AI Technologies at https://grandline.ai ⭐ Hugging Face Write with Transformers https://transformer.huggingface.co/

https://www.youtube.com/watch?v=4Bdc55j80l8

How to train RNN

LSTM Cells