Description
Standard scaled-down Stockfish plays perfectly, then makes random obvious blunders to lower its ELO. This isn't how humans play. We need realistic Maia-like neural networks.
What Needs to be Done
- Download and integrate the Maia Chess weights (predicting human moves at specific ELO tiers: 1100, 1500, 1900).
- Configure the backend to spin up workers running these specific weights.
- Create distinct 'Bot Personalities' in the UI mapped to these models.
Acceptance Criteria
Description
Standard scaled-down Stockfish plays perfectly, then makes random obvious blunders to lower its ELO. This isn't how humans play. We need realistic Maia-like neural networks.
What Needs to be Done
Acceptance Criteria