llama3-8b-pokerbench-sft-gguf/README.md

license, base_model, tags, language, pipeline_tag

license	base_model	tags	language	pipeline_tag
llama3	YiPz/llama3-8b-pokerbench-sft	poker gguf llama-cpp ollama quantized	en	text-generation

Llama 3 8B - PokerBench SFT (GGUF)

GGUF quantized versions of YiPz/llama3-8b-pokerbench-sft.

Available Files

File	Size	Description
llama3-8b-pokerbench-sft-q4_k_m.gguf	~4.5 GB	Recommended - good quality/size balance
llama3-8b-pokerbench-sft-q8_0.gguf	~8.5 GB	Higher quality

Usage with Ollama

# Download
huggingface-cli download YiPz/llama3-8b-pokerbench-sft-gguf llama3-8b-pokerbench-sft-q4_k_m.gguf --local-dir ./

# Create Modelfile
cat > Modelfile << 'EOF'
FROM ./llama3-8b-pokerbench-sft-q4_k_m.gguf
PARAMETER temperature 0.1
SYSTEM "You are an expert poker player. Respond with your action in <action></action> tags."
EOF

# Create and run
ollama create pokerbench -f Modelfile
ollama run pokerbench "Your scenario..."

License

Subject to Llama 3 license.