llama
Installation
SKILL.md
Llama
Meta Llama is the foundation standard for open-weight language models, providing dense and Mixture-of-Experts architectures optimized for fine-tuning, quantization, and local deployment.
When to Use
- Local & On-Premise LLM Deployment: Running Meta Llama 3.1/3.2/3.3 locally without data leaving the private perimeter.
- High-Efficiency Quantized Inference: Utilizing GGUF (via llama.cpp) and AWQ/GPTQ on consumer and enterprise GPUs.
- Fine-Tuning on Custom Domain Data: Training LoRA/QLoRA adapters with Unsloth or Axolotl.
- Embedded & Mobile Edge AI: Running lightweight Llama 3.2 1B and 3B models directly on laptops and mobile devices.