Local command extraction

• Crucial for dividing user instructions into actionable commands.
• Transition from OpenAI’s 3.5 Turbo (API dependent, latency issues) to a fine-tuned Llama 3.2 3B model for local use.

Benefits of the new implementation:

• Comparable performance.
• Reduced latency and internet independence.
• Fine-tuning results published on HuggingFace.