Text Generation
English
chess
puzzles
chess-games
stockfish
fen
best-move
uci
san
text-generation-inference
File size: 2,124 Bytes
c15b6b0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5fa4515
 
f0c113d
5d8e6d3
5fa4515
 
 
 
 
5d8e6d3
5fa4515
5d8e6d3
5fa4515
 
 
5d8e6d3
5fa4515
5d8e6d3
5fa4515
 
 
5d8e6d3
5fa4515
5d8e6d3
5fa4515
 
 
 
 
8bb2e18
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
---
license: mit
language:
- en
pipeline_tag: text-generation
tags:
- chess
- puzzles
- chess-games
- stockfish
- fen
- best-move
- uci
- san
- text-generation-inference
datasets:
  - ethanjtang/GAMBIT-stockfish18-selfplay
  - ethanjtang/GAMBIT-lichess-puzzle-positions
---

# GAMBIT: <ins>G</ins>ener<ins>a</ins>lization or <ins>M</ins>emorization? <ins>B</ins>r<ins>i</ins>ttleness <ins>T</ins>esting for Chess-Trained Language Models

[![arXiv](https://img.shields.io/badge/arXiv-2605.17565-b31b1b.svg?style=for-the-badge)](https://arxiv.org/abs/2605.17565) <br>
[![GitHub](https://img.shields.io/badge/GitHub-KinGPT-black.svg?style=for-the-badge)](https://github.com/ethanjtang/KinGPT) <br>
[![HuggingFace](https://img.shields.io/badge/🤗_HuggingFace-Puzzles-yellow?style=for-the-badge)](https://huggingface.co/datasets/ethanjtang/GAMBIT-lichess-puzzle-positions) <br>
[![HuggingFace](https://img.shields.io/badge/🤗_HuggingFace-SF18%20Selfplay-yellow?style=for-the-badge)](https://huggingface.co/datasets/ethanjtang/GAMBIT-stockfish18-selfplay) <br>

## Variants

### KinGPT-Woodpecker

KinGPT variant trained on 13,341,057 unique puzzle positions (FEN + best move pairs).

Achieved `train loss 0.3590, val loss 0.3704` on puzzles corpus after training for ~500B tokens.

### KinGPT-Beaver

KinGPT variant trained on 54,681 unique positions generated from 1050 Stockfish 18 self-play games.

Achieved `train loss 0.0974, val loss 1.7554` (overfitting due to small dataset size) on selfplay corpus after training for ~25B tokens.

### KinGPT-Chimera

KinGPT variant trained on combined dataset of 13,395,738 Woodpecker and Beaver variant positions.

Achieved `train loss 0.3594, val loss 0.3710` on combined corpus after training for ~500B tokens.

## Citation

```bibtex
@misc{tang2026generalizationmemorizationbrittlenesstesting,
      title={Generalization or Memorization? Brittleness Testing for Chess-Trained Language Models}, 
      author={Ethan Tang},
      year={2026},
      eprint={2605.17565},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2605.17565}, 
}
```