# Chess Reasoning Quality Evaluation on Lichess Puzzle Database (held-out positions) Source: [Chess Reasoning Quality Evaluation on Lichess Puzzle Database (held-out positions)]() Collection index: [llms.txt]() Data updated: 2026-05-14T06:55:00.646Z Evaluation of model performance on held-out positions from the Lichess Puzzle Database, assessing general Chess Reasoning Quality using metrics such as WR MAE, PV Overlap, and Consistency. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Chess Reasoning Quality Evaluation | | Dataset (including any stated split) | Lichess Puzzle Database (held-out positions) | | Metric | WR MAE | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 6 of 6 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [WR MAE](): Lower is better; unit: SCALAR. - [PV Overlap](): Higher is better; unit: PERCENT. - [Consistency](): Higher is better; unit: PERCENT. ## Recorded results | Method | WR MAE | Result date | Report paper | | --- | --- | --- | --- | | VPS | 0.218 | Not recorded | 6a05570cdbc836b87d828c09 | | VPS | 0.286 | Not recorded | 6a05570cdbc836b87d828c09 | | SFT only | 0.31 | Not recorded | 6a05570cdbc836b87d828c09 | | SFT only | 0.346 | Not recorded | 6a05570cdbc836b87d828c09 | | SFT + GRPO | 0.452 | Not recorded | 6a05570cdbc836b87d828c09 | | SFT + GRPO | 0.732 | Not recorded | 6a05570cdbc836b87d828c09 | ### VPS WR MAE: 0.218. Result ID: 6a05701347f34ce924ebd039. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### VPS WR MAE: 0.286. Result ID: 6a05701347f34ce924ebd03c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Llama-8B"}. This record does not assert independent reproduction. ### SFT only WR MAE: 0.31. Result ID: 6a05701347f34ce924ebd037. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### SFT only WR MAE: 0.346. Result ID: 6a05701347f34ce924ebd03a. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Llama-8B"}. This record does not assert independent reproduction. ### SFT + GRPO WR MAE: 0.452. Result ID: 6a05701347f34ce924ebd038. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### SFT + GRPO WR MAE: 0.732. Result ID: 6a05701347f34ce924ebd03b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DeepSeek-R1-Distill-Llama-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.