Amey9766/llama32-1b-maintenance-classifier

🛠️ Llama‑3.2‑1B — Maintenance Request Classifier

Fine‑tuned for hotel & facility repair ticket triage (urgent / routine / cosmetic)

This model is a fine‑tuned version of meta‑llama/Llama‑3.2‑1B, trained on a curated hotel maintenance dataset to classify repair tickets into:

• Urgent
• Routine
• Cosmetic

It is designed for hotel operations, engineering teams, and facility management systems that need fast, automated prioritization of maintenance requests.

---

✨ Key Features

• Classifies maintenance requests into 3 priority levels
• Optimized for short, noisy, real‑world repair descriptions
• Lightweight 1B parameter model suitable for low‑cost inference
• Trained using QLoRA (4‑bit) for efficiency
• Ideal for hotels, apartments, facility ops, and CMMS systems

---

📌 Intended Use

• Auto‑triage maintenance tickets
• Prioritize engineering workload
• Route urgent issues to on‑call staff
• Improve SLA compliance
• Power internal tools or chatbots for maintenance teams

---

🚫 Out‑of‑Scope

• Medical, legal, or safety‑critical decision making
• Predicting repair cost or time
• Any use requiring guaranteed accuracy

---

🧠 Training Details

• Base model: meta‑llama/Llama‑3.2‑1B
• Method: QLoRA (4‑bit quantization)
• Dataset: hotel_maintenance_dataset
• Labels: urgent, routine, cosmetic
• Hardware: Google Colab (L4/T4)
• Epochs / LR / Batch size: (fill in your values)

---

📊 Evaluation Metrics

Overall Performance

Accuracy: 0.9583 Precision (weighted): 0.9618 Recall (weighted): 0.9583 F1-Score (weighted): 0.9591

📚 Citations

Base Model Citation (Meta Llama 3.2)

@misc{meta2024llama3, title={The Llama 3 Herd of Models}, author={Meta AI}, year={2024}, url={https://ai.meta.com} }

Citation Fine Tuned Model

@misc{tillu2026maintenanceclassifier, title={Llama-3.2-1B Maintenance Request Classifier}, author={Amey Tillu}, year={2026}, howpublished={\url{https://huggingface.co/Amey9766/llama32-1b-maintenance-classifier}} }