Your Stats
0
Total XP
0/24
Badges
0
Day Streak
0
Total Score
0 /5 levels
0 pts
Games
Optimization
Gradient Descent
Adjust the learning rate to reach the minimum loss without overshooting or crawling.
+50 XP Active
Your task
Easy · 1× ptsFind the optimal learning rate
How to play: Drag the slider to explore, then press Lock In.
Levels
0 pts
You are training a neural network. The learning rate controls how big each step is when descending the loss landscape.
0.50
0.00 ← range → 1.00
Balancing between the extremes…
What this means
Training an AI model is like rolling a ball down a hill — you want it to reach the bottom (the best solution). The learning rate controls step size. Too small, and it crawls forever. Too large, and it bounces past the bottom. Your job: find the sweet spot.
Leaderboard
Loading…