You can deploy this model and start earning money today!

mrigaanksh/
priority-classification-distilbert

text-embeddings-inference

endpoints_compatible

Views

No views yet

🚦 Civic Issue Priority Classification Model

This model classifies civic issue reports (text descriptions) into three priority levels — High, Medium, and Low — to assist municipal systems in automatically routing and prioritizing citizen complaints.

🧠 Model Details

Model Description

This model fine-tunes DistilBERT (distilbert-base-uncased) for text classification to predict the priority of civic issues reported by citizens through an app or web portal.

Model Type: DistilBERT For Sequence Classification
Language: English
Fine-tuned On: Custom civic issue dataset
Labels:
- 0 → Low (Minor issues, e.g., paint fade, broken bench)
- 1 → Medium (Moderate issues, e.g., streetlight not working, drainage blockage)
- 2 → High (Critical issues, e.g., gas leak, transformer fire, road flood)

📊 Training Details

Dataset

A custom dataset of 30 handcrafted civic issue reports divided evenly into three categories:

10 High priority examples
10 Medium priority examples
10 Low priority examples

Each example represents a real-world scenario commonly reported in urban complaint systems.

Training Configuration

Hyperparameter	Value
Base model	distilbert-base-uncased
Batch size	4
Epochs	15
Learning rate	3e-5
Weight decay	0.02
Max sequence length	64
Optimizer	AdamW
Evaluation strategy	per epoch

Hardware

Trained on: Google Colab (T4 GPU)
Framework: PyTorch + Hugging Face Transformers

⚙️ How to Use

python

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import torch.nn.functional as F

# Load model
tokenizer = AutoTokenizer.from_pretrained("mrigaanksharma/priority-classifier")
model = AutoModelForSequenceClassification.from_pretrained("mrigaanksharma/priority-classifier")

text = "Transformer caught fire near main road"
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
outputs = model(**inputs)
probs = F.softmax(outputs.logits, dim=-1)
pred = torch.argmax(probs).item()
confidence = torch.max(probs).item()

label_map = {0: "Low", 1: "Medium", 2: "High"}
print(f"Text: {text}")
print(f"Predicted Label: {label_map[pred]} (Confidence: {confidence:.2f})")

mrigaanksh/priority-classification-distilbert