# Download model
huggingface-cli download gameversellc/hermes_3_tests_gv Hermes-3-Llama-3.2-3B_q4_0.gguf

# Run inference
./llama-cli -m Hermes-3-Llama-3.2-3B_q4_0.gguf -p "Your prompt here" -n 100

With llama-cpp-python

python

from llama_cpp import Llama

llm = Llama(model_path="Hermes-3-Llama-3.2-3B_q4_0.gguf")
output = llm("Your prompt here", max_tokens=100)
print(output)

Performance

MMLU Accuracy: ~40% (quantized)
Inference Speed: Fast on CPU
Memory Usage: ~2 GB RAM

License

Apache 2.0 (same as base model)