---
license: other
tags:
- chess
- transformer
- text-generation
language:
- en
pipeline_tag: text-generation
---

# GofAlpha/GoA_Chess_Small

A decoder-only transformer language model trained on Lichess standard chess games.
Generates chess moves in standard algebraic notation (PGN).

## Model Details

| Property | Value |
|----------|-------|
| Parameters | 51,544,049 |
| Hidden size | 512 |
| Layers | 8 |
| Attention heads | 8 |
| Context length | 1024 tokens |
| Tokenizer | GPT-2 |

## Files

- `model.safetensors` — Model weights
- `config.json` — Architecture configuration
- `generation_config.json` — Default generation parameters
- `tokenizer.json`, `vocab.json`, `merges.txt` — GPT-2 tokenizer files

## Usage

```python
import torch
from safetensors.torch import load_file
from transformers import GPT2TokenizerFast

# Load tokenizer
tokenizer = GPT2TokenizerFast.from_pretrained("GofAlpha/GoA_Chess_Small")

# Load weights
state_dict = load_file("model.safetensors")

# Encode a chess opening
prompt = "1. e4 e5 2. Nf3"
input_ids = tokenizer.encode(prompt, return_tensors="pt")
```

## Generation Parameters

| Parameter | Default |
|-----------|---------|
| temperature | 0.6 |
| top_k | 10 |
| top_p | 0.95 |
| max_new_tokens | 1024 |

## License

Copyright (c) G of Alpha LLC. All rights reserved.