CJC Chess RL — Interactive Platform

Deterministic Play | SplitMix64 RNG | Actor-Critic [288→128→128→P+V] | A2C + GAE | Self-Improving
Seed: 42 vs Agent idle Replay Stable ✓
Quick start:
Play
Review
Openings
Style Profile
Training
Settings
Click "New Game" to start.
Material
Move History
#WhiteBlack
Trace
No trace yet.
Post-Game Review
Play a game first, then review it here.
Opening Explorer (from stored traces)
Play some games to populate the opening tree.
Your Style Profile
Play some games to generate your style profile.
Training Controls
Idle
Curriculum: Phase A (vs Random)
Episodes: 0
Learning Rate: 0.0003
Replay Buffer: 0 / 100
Evaluation vs Baselines (last eval)
No evaluation run yet. Click "Evaluate Now".
Reward Curve
Win Rate vs Random
Policy Loss & Value Loss
Entropy & Gradient Norm
Model Checkpoints
No checkpoints saved yet.
Platform Settings



Demo Presets
Session Export