Commit History
-
Upload 11 files
Verified signature# Duplicate Ticket Detection Model (SBERT) This repository contains a Sentence-BERT (SBERT) model fine-tuned for detecting duplicate or semantically similar support tickets. The model generates dense vector embeddings for ticket descriptions, enabling efficient similarity comparison using cosine distance. ## 🔍 Purpose The model is designed to identify duplicate or closely related tickets in IT service management systems. It helps reduce redundancy, improves resolution time, and enhances overall ticket handling efficiency. ## ⚙️ Model Details * **Base Model**: microsoft/mpnet-base * **Architecture**: Transformer (MPNet) + Mean Pooling + Normalization * **Embedding Dimension**: 768 * **Max Sequence Length**: 384 * **Output**: Dense semantic embeddings for similarity comparison ## 📂 Repository Structure * `modules.json` → Defines model pipeline (Transformer → Pooling → Normalize) * `0_Transformer/` → Core transformer model, weights, and tokenizer files * `1_Pooling/` → Mean pooli…f7eebff