Pradheep1647/kibitzer-sft-elo-0808-step-002000

Kibitzer checkpoint for a chess policy/value model trained on game sequences. Each board is encoded as 64 square tokens plus auxiliary state, summarized by a square-level encoder, then processed by a causal transformer over the game timeline.

Status

• checkpoint: step_002000.pt
• step: 002000
• elo_rating: 808 estimated vs stockfish-elo-1320 (-511.5 Elo diff)
• github_repo: https://github.com/Mantissagithub/kibitzer
• eval_opponent: stockfish-elo-1320
• elo_diff: -511.5
• eval_games: 20
• eval_score: 1.00
• elo_error: n/a

Architecture

• Input: 64 chess-square piece tokens plus 7 auxiliary scalars covering side to move, castling rights, en-passant file, and halfmove clock.
• Position encoder: bidirectional square-level transformer that compresses each board into one timeline token.
• Timeline trunk: causal transformer with d_model=384, sequence length 256, 12 layers, and 8 attention heads.
• Transformer blocks: pre-norm RMSNorm, RoPE on causal Q/K attention, PyTorch scaled dot-product attention, and SwiGLU feed-forward layers.
• Outputs: policy head over 4,672 AlphaZero-style moves (64 * 73) plus a tanh value head for bounded outcome prediction.

Files

• step_002000.pt: PyTorch checkpoint.
• training_metadata.yaml: training config and checkpoint metrics.
• post_eval.yaml: uploaded after local Stockfish/cutechess eval.

Usage

git clone https://github.com/Mantissagithub/kibitzer.git
cd kibitzer
uv run python scripts/uci.py --checkpoint <path-to>/step_002000.pt

This checkpoint is one artifact in a sequence. Repos with elo-pending in the name have not been evaluated yet; rated repos are renamed after scripts/eval_and_rename_hf.py --from-hf completes.