🚀 Llama-3.2-1B-JSON-Extractor

This model is a professional fine-tuned version of Llama 3.2 1B Instruct, specialized in converting Natural Language into structured JSON objects.

🎯 Project Goal

The system interprets human intent and extracts key entities into a machine-readable format. It is designed to be the bridge between human communication and database systems.

Extracted Fields:

• product: Name or description of the item.
• price: Numerical value (currency independent).
• category: Market segment or classification.

🛠️ Technical Specifications

• Architecture: Llama 3.2 1B
• Optimization: QLoRA (4-bit)
• Rank (r): 16
• Alpha: 32
• Learning Rate: 2e-4
• Final Training Loss: ~0.013 (High precision)

💡 Usage Example

Input:

"I want to sell a Sony PlayStation 5 for 500 dollars in the gaming category."

Output:

{
  "product": "Sony PlayStation 5",
  "price": 500,
  "category": "gaming"
}

Developed by ivanasp using Unsloth.