Anshrajsingh/qwen2.5-1.5b-ticket-classifier

TEXT GENERATIONConcurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 7, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Anshrajsingh/qwen2.5-1.5b-ticket-classifier is a 1.5 billion parameter Small Language Model (SLM) fine-tuned by Anshrajsingh from the Qwen2.5-1.5B-Instruct base model. Optimized for production-grade structured data extraction, this model excels at schema-bound classification of unstructured text into strict JSON formats. It achieves high accuracy and 100% pure JSON output without conversational filler, making it ideal for cost-conscious, low-latency enterprise applications like customer ticket routing.

Loading preview...

Overview

This model, Anshrajsingh/qwen2.5-1.5b-ticket-classifier, is a 1.5 billion parameter Small Language Model (SLM) fine-tuned from the Qwen/Qwen2.5-1.5B-Instruct base. Developed by Anshrajsingh, its primary purpose is to address critical bottlenecks in enterprise environments when mapping unstructured text (like customer support tickets) to structured databases. It focuses on providing a cost-effective, low-latency, and privacy-compliant solution for deterministic structured data extraction.

Key Capabilities

  • Strict JSON Schema Compliance: Achieves 95.6% schema adherence and 100% pure JSON output, eliminating conversational filler ("yapping").
  • High Accuracy: Demonstrates 91.3% exact match accuracy on a hidden test dataset for ticket classification.
  • Cost-Efficient: Offers significant cost reduction (estimated ~96.5% savings) compared to large commercial LLM APIs for high-volume tasks.
  • Low Latency & Data Privacy: Suitable for edge deployment or low-cost commodity GPUs, enabling real-time processing and maintaining data compliance.
  • Fine-Tuning: Utilizes QLoRA (4-bit quantization) for parameter-efficient fine-tuning on a consumer-grade T4 GPU.

Ideal Use Cases

This model is specifically engineered for production-grade applications requiring:

  • Automated classification and routing of customer support tickets.
  • Extracting structured data from invoices, logs, or other unstructured text.
  • Scenarios where strict JSON output, low operational costs, and data privacy are paramount.
  • Deployment on resource-constrained environments or for high-throughput, narrow tasks.

Finetunes
1 models