Instructions to use lazy-guy12/chess-llama with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use lazy-guy12/chess-llama with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="lazy-guy12/chess-llama")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("lazy-guy12/chess-llama")
model = AutoModelForCausalLM.from_pretrained("lazy-guy12/chess-llama")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use lazy-guy12/chess-llama with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "lazy-guy12/chess-llama"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lazy-guy12/chess-llama",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/lazy-guy12/chess-llama

SGLang

How to use lazy-guy12/chess-llama with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "lazy-guy12/chess-llama" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lazy-guy12/chess-llama",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "lazy-guy12/chess-llama" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "lazy-guy12/chess-llama",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use lazy-guy12/chess-llama with Docker Model Runner:
```
docker model run hf.co/lazy-guy12/chess-llama
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

We trained a tiny Llama-based decoder-only transformer model for chess play, consisting of 23M parameters. The model is trained on a 3 million high-quality chess games from the Lichess Elite Database, on a single Nvidia L4 GPU for 18 hours, using the Google Cloud’s Vertex AI platform.

It uses the UCI format for input and output. It has been trained with the token indicating result appended to the beginning of the games, hoping it would improve performance during actual chess play. The model achieves an estimated Elo rating of 1400, and easily outperforms Skill-level 0 Stockfish.

You can try it out here! It runs directly on your device, within the browser, thanks to transformers.js.

Todo:

Upload model to Hugging Face
Add inference scripts for proper chess play
Integrate it into a webui, possibly by using transformers.js

Blogpost - https://lazy-guy.github.io/blog/chessllama/

Downloads last month: 145

Safetensors

Model size

23M params

Tensor type

F32

Model tree for lazy-guy12/chess-llama

Quantizations

1 model

lazy-guy12
/

chess-llama

Model tree for lazy-guy12/chess-llama

Dataset used to train lazy-guy12/chess-llama