0 - 0
DEBUG INFO
Left Paddle Y:
0
Predicted Y:
0
Difference:
0
Ball X:
0
Direction:
-
Ball Speed:
0
Hidden Size:
24
⚙️
⏸️
❔
🏠
🐛
Settings
Ball Speed:
10-50
Speed Increment:
0.1
AI Speed:
8
Game Speed:
60 TPS
Learning Rate:
0.0005
Scoring Reward
OFF
Training
ON
Load Model
Load Model Files
Load Checkpoint
Select a checkpoint...
Load Checkpoint
Download Latest Model
Hidden Size:
Reset Model
Paddle Segments
Random: OFF
🎲 Randomize
Top:
Upper Mid:
Center:
Lower Mid:
Bottom:
W:
0.00
Pong AI Help
About the Agents
Left Paddle: Deep Q-Learning agent that learns from experience
Right Paddle: Perfect AI with configurable difficulty
The learning agent improves by playing against the perfect AI or you!
Controls
Space: Pause/Resume game
Arrow Up/Down: Take control of right paddle
Mouse: Drag right paddle
Touch: Drag right paddle on mobile devices
Settings
Ball Speed Range: Controls both starting and maximum ball speed (5-200)
Speed Increment: How quickly ball speeds up during rallies (0.05-0.5)
AI Speed: Controls the perfect AI's movement speed (right paddle)
Game Speed: Controls the ticks per second (1-600 TPS)
Scoring Reward: Toggles additional rewards for scoring/being scored on:
ON: AI gets +1.5 reward for scoring, -1.5 for being scored on
OFF: AI only learns from paddle positioning
Training: Controls whether the AI is learning:
ON: AI continues to learn and adapt its strategy
OFF: AI uses current knowledge without learning
Model Management
Load Model: Import previously saved model files (.json and .bin)
Load Checkpoint: Load pre-trained models from server:
Download Latest: Save current model to files for later use
Debug Tools
🐛 Debug Panel: Shows real-time paddle positions, ball info, and predictions
📈 Performance Graph: Displays training loss and reward history
Paddle Segments
Random: ON - Generate new random weights each game
Random: OFF - Use manually specified weights
Randomize - Generate new random weights immediately
Manual inputs - Set specific probabilities for each segment
Close