anandshivam44/jarvis-ticket-classifier
The anandshivam44/jarvis-ticket-classifier is a 0.5 billion parameter instruction-tuned causal language model, based on Qwen2.5-0.5B-Instruct, fine-tuned by anandshivam44. This model is specifically designed for classifying support tickets and generating JSON-formatted responses, demonstrating a specialized application of a smaller LLM. It leverages Supervised Fine-Tuning (SFT) and LoRA for efficient adaptation, making it suitable for focused, domain-specific classification tasks.
Loading preview...
Overview
The anandshivam44/jarvis-ticket-classifier is a specialized language model, fine-tuned from the Qwen2.5-0.5B-Instruct base model. With 0.5 billion parameters and a context length of 32768 tokens, this model is designed for a very specific task: classifying support tickets and outputting structured JSON responses. It utilizes Supervised Fine-Tuning (SFT) and LoRA (Low-Rank Adaptation) for efficient training on a dataset of "fake support tickets."
Key Capabilities
- Support Ticket Classification: Accurately categorizes incoming support requests.
- JSON Output Generation: Formats classification results into structured JSON, facilitating integration with automated systems.
- Domain-Specific Fine-tuning: Optimized for the nuances of support ticket language.
- Efficient Adaptation: Leverages LoRA for effective fine-tuning on a smaller dataset.
Good For
- Demonstrating SFT + LoRA: Serves as a practical example of applying these techniques to a small language model for a specific use case.
- Automated Ticket Routing: Can be used to automatically classify and route support tickets to the correct departments or agents.
- Structured Data Extraction: Useful for extracting key information from unstructured text (support tickets) into a machine-readable format.
- Local Deployment: Its smaller size makes it suitable for deployment on consumer-grade hardware using tools like Ollama, as demonstrated in the README.