Github

Descent
Visualisers.

Interactive labs designed to visualize the behaviour of various Machine Learning algorithms.

Gradient Descent Visualiser

A fast, interactive tool to visualize how different gradient descent algorithms navigate complex loss surfaces in real time.

Exploring standard optimization terrains: Saddle Points, Bowls, and Plateaus.

Vizit now

Real-time visualization of convergence paths initiated from user-defined starting coordinates.

Vizit now

Rendering custom 3D loss surfaces generated instantly via mathematical function input.

Vizit now

RL visualiser

Exploring how various Reinforcement learning algorithms like QL, DQL and PPO learn and act upon custom simulated environments.

Visualizing tabular Q-Learning in a grid world. Watch as an autonomous agent explores the maze.

Vizit now

Visualizing agents mastering the CartPole physics challenge.

Vizit now