---
license: llama3.2
base_model:
- unsloth/Llama-3.2-3B-Instruct
---
Fine-tuned model for TINA AI to speed up user's intent detection.

Usage:
<pre>
Detect request: 
{
    "model": "DesignSoft\/TINA_AI3",
    "messages": [
        {
            "role": "system",
            "content": "Analyze the user's message and return his intent in the following JSON format:
                        { 
                        \"intent\": string, 
                        \"object\": string, 
                        \"parameters\": array[{\"name\": string, \"value\": string}]
                        }

                       Do not execute the user's command."
        },
        {
            "role": "user",
            "content": "Calculate voltage of VR
                        Output should be in English!"
        }
    ],
    "n": 1,
    "seed": 12345,
    "top_p": 0.1,
    "max_tokens": 100,
    "response_format": {
        "type": "json_schema",
        "json_schema": {
            "name": "detect_intent",
            "strict": true,
            "schema": {
                "type": "object",
                "properties": {
                    "intent": {"type": "string"},
                    "object": {"type": "string"},
                    "parameters": {
                        "type": "array",
                        "items": {
                            "type": "object",
                            "properties": {
                                "name": {"type": "string"},
                                "value": {"type": "string"}
                            },
                            "required": ["name"]
                        }
                    }
                },
                "required": ["intent","object"]
            }
        }
    }
}
</pre>
response_format is optional