Overview

Harmj0y/qwen3-4b-instruct-phishing-classifier is a 4 billion parameter Qwen3-based instruction-tuned model developed by Harmj0y. It was specifically fine-tuned for phishing classification, building upon the unsloth/qwen3-4b-instruct-2507-unsloth-bnb-4bit base model. The training process utilized Unsloth and Huggingface's TRL library, enabling a 2x faster fine-tuning compared to standard methods.

Key Capabilities

• Phishing Classification: The primary capability of this model is to identify and classify phishing attempts within text data.
• Efficient Fine-tuning: Benefits from accelerated training via Unsloth, making it a resource-efficient option for specialized tasks.
• Qwen3 Architecture: Built on the Qwen3 model family, providing a robust foundation for language understanding.

Good For

• Security Applications: Ideal for integrating into systems that require automated detection of phishing content.
• Content Moderation: Can be used to filter out malicious or deceptive text in various platforms.
• Research and Development: Provides a specialized model for further experimentation and development in text-based threat detection.