# Released ChessLFM baseline

**Observed match:** 17.5/32 points, W/D/L 12–11–9, against Stockfish set to Elo 2000.
**Evaluation block:** Later opening block (openings 16–31).
**Descriptive logistic estimate:** ~2033 Elo; this small match is not a precise rating.

## Method

Original public ChessLFM checkpoint. Included as the reference model; it is not a checkpoint trained by this project.

## Interpretation

This is the highest observed score among the saved matches. Its estimate is specific to this 32-game block.

**Lineage:** LiquidAI/LFM2.5-230M-Base → this public ChessLFM checkpoint. Its `surgery_meta.json` identifies the LiquidAI base. This is an upstream reference model, not trained by this project. For legal move masking, board prompt construction, and search, see https://github.com/0xrushi/ChessLFM.