Hyperparameters
Learning Rate (α):
0.20
Discount Factor (γ):
0.90
Exploration (ε):
0.15
Episode Stats
Episodes Completed
0
Total Steps Taken
0
Last Episode Return
0.0
Live Bellman Update
Q-Learning Rule:
Q(s,a) ← Q + α [r + γ max Q(s',a') - Q]
Waiting for step...
Learning Curve (Return / Ep)
🟩 Goal: +10 | 🟥 Pit: -10 | Step cost: -0.1
Blue arrows indicate greedy policy π(s).
Blue arrows indicate greedy policy π(s).