| """ |
| Support Ticket Classifier -Inference Script |
| Author: Adedayo Adebayo |
| Date: 2026-03-14 |
| """ |
|
|
| import torch |
| import logging |
| from typing import List, Dict, Any |
| from transformers import AutoTokenizer, AutoModelForSequenceClassification |
|
|
| logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s') |
| logger = logging.getLogger(__name__) |
|
|
| def load_inference_tools(model_path: str): |
| logger.info(f"Loading resources from: {model_path}") |
| device = torch.device("cuda" if torch.cuda.is_available() else "cpu") |
| |
| tokenizer = AutoTokenizer.from_pretrained(model_path) |
| model = AutoModelForSequenceClassification.from_pretrained(model_path) |
| model.to(device) |
| model.eval() |
| return tokenizer, model, device |
|
|
| def get_predictions(texts: List[str], tokenizer, model, device) -> List[Dict[str, Any]]: |
| id2label = model.config.id2label |
| |
| inputs = tokenizer( |
| texts, |
| padding=True, |
| truncation=True, |
| max_length=128, |
| return_tensors="pt" |
| ).to(device) |
|
|
| with torch.no_grad(): |
| outputs = model(**inputs) |
| probs = torch.nn.functional.softmax(outputs.logits, dim=-1) |
| confidences, predictions = torch.max(probs, dim=-1) |
|
|
| results = [] |
| for text, pred_id, conf in zip(texts, predictions, confidences): |
| results.append({ |
| "text": text, |
| "category": id2label[pred_id.item()], |
| "confidence": conf.item() |
| }) |
| return results |
|
|
| def main(): |
| MODEL_PATH = "Adedayo2000/BERTweet-Support-Ticket-Classifier" |
| |
| input_ticket = [ |
| "My account is locked,help!.", |
| "The latest update broke the export feature on the dashboard.", |
| "The app is crashing when I try to upload a file.", |
| "How do I change the notification settings for my team?", |
| "I need to reset my password but the reset link isn't working.", |
| "The new UI is confusing and hard to navigate.", |
| "I was double charged for my subscription this month, please assist." |
| ] |
| |
| tokenizer, model, device = load_inference_tools(MODEL_PATH) |
| |
| logger.info("Classifying tickets...") |
| predictions = get_predictions(input_ticket, tokenizer, model, device) |
|
|
| print(f"{'TICKET PREVIEW':<60} | {'CATEGORY':<15} | {'CONF'}") |
| |
| for res in predictions: |
| preview = (res['text'][:40] + '...') if len(res['text']) > 40 else res['text'] |
| print(f"{preview:<60} | {res['category']:<15} | {res['confidence']:.2%}") |
|
|
| if __name__ == "__main__": |
| main() |
