# Load model directly
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("Belnadino/SmartHome-AI-IoT-Generator")
model = AutoModelForSequenceClassification.from_pretrained("Belnadino/SmartHome-AI-IoT-Generator", device_map="auto")SmartHome-AI-IoT Generator Model
Overview
The SmartHome-AI-IoT Generator Model generates executable smart-home device actions from natural language commands and contextual information.
The model performs multi-label classification, allowing a single user command to control multiple IoT devices simultaneously.
Example:
Input: Prepare the bedroom for sleeping Location: Bedroom Time: Night Temperature: 18°C
Output: AC_OFF Curtains_CLOSE BedLamp_OFF TV_OFF Radio_OFF AirPurifier_ON
Authors
Belnadino Mgimba
Department of Computer Science, Durham University, Durham, United Kingdom
Anish Jindal
Department of Computer Science, Durham University, Durham, United Kingdom
Project
This model is part of the:
SmartHome-AI-IoT Framework
"A Natural Language Understanding, Low-Latency, Context-Aware, Multi-Device Control and Self-Learning SmartHome IoT Framework"
Developed for research on:
- Natural language understanding for IoT
- Context-aware smart-home automation
- Edge AI deployment
- Human-in-the-loop adaptive systems
Intended Use
Applications include:
- Smart-home automation
- Natural language IoT control
- Multi-device orchestration
- Edge AI systems
Model Architecture
- Base model: DistilBERT
- Task: Multi-label classification
- Framework: PyTorch + Transformers
The model predicts device-action labels:
Examples: Light_ON Light_OFF AC_ADJUST Curtains_OPEN Curtains_CLOSE
Dataset
Trained using:
SmartHomeIoTNLU Dataset
Dataset includes:
- 200,320 smart-home events
- 19 IoT devices
- 5 rooms
- 23 scenarios
- Natural language commands
- Environmental context
Training Configuration
Base model: distilbert-base-uncased
Maximum sequence length: 64
Batch size: 16
Learning rate: 2e-5
Optimizer: AdamW
Epochs: 3
Loss: Binary Cross Entropy
Performance
| Metric | Value |
|---|---|
| Accuracy | 0.9839 |
| F1 samples | 0.9946 |
| F1 micro | 0.9978 |
| Jaccard Score | 0.9920 |
| Hamming Loss | 0.0004 |
Deployment
The model is designed for deployment in edge-based smart-home systems.
Deployment components:
- Raspberry Pi gateway
- Smartphone controller
- ESP32 IoT devices
Limitations
- Dataset is synthetically generated.
- Performance may vary with real household commands.
- Additional adaptation may be required for new devices.
Citation
SmartHome-AI-IoT Framework
License
apache-2.0
- Downloads last month
- 7
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="Belnadino/SmartHome-AI-IoT-Generator")