---
tags:
- reinforcement-learning
- iql
- fire-rescue
- emergency-response
license: mit
---

# IQL Fire Rescue Operator Model

This model uses Implicit Q-Learning (IQL) to select optimal conversation policies for emergency response scenarios.

## Model Description

- **Task**: Policy selection for fire rescue emergency conversations
- **Algorithm**: Implicit Q-Learning (IQL)
- **Input**: Conversation history (embedded using sentence-transformers)
- **Output**: Best policy and Q-values for all policies

## Policies

1. `provide_information` - Share facts, data, risk levels
2. `offer_assistance` - Provide help, resources, support
3. `express_urgency` - Convey time pressure, danger
4. `ask_question` - Gather information from resident
5. `give_direction` - Provide specific instructions
6. `acknowledge_concern` - Validate resident's feelings
7. `build_rapport` - Establish trust and connection

## Usage

```python
import requests

API_URL = "https://api-inference.huggingface.co/models/YOUR_USERNAME/iql-fire-rescue"
headers = {"Authorization": f"Bearer YOUR_HF_TOKEN"}

def query(payload):
    response = requests.post(API_URL, headers=headers, json=payload)
    return response.json()

# Example
output = query({
    "inputs": "Conversation state here",
    "parameters": {"character": "bob"}
})
```

## Training Data

Trained on emergency response conversations with multiple character profiles.

## Citation

If you use this model, please cite:
```
@misc{iql-fire-rescue,
  title={IQL-based Emergency Response Operator},
  author={Your Name},
  year={2025},
  publisher={Hugging Face}
}
```