rogeriobayer/focus-patrol-qwen2.5-0.5b-v7
The rogeriobayer/focus-patrol-qwen2.5-0.5b-v7 is a 0.5 billion parameter Qwen2.5-based causal language model, fine-tuned by rogeriobayer for browser tab classification. This model specializes in categorizing browser tabs into 'focus', 'neutral', or 'distraction' with an overall test accuracy of 90%. It is optimized for on-device inference, leveraging its small size and specific fine-tuning for real-time application in browser environments.
Loading preview...
Model Overview
The rogeriobayer/focus-patrol-qwen2.5-0.5b-v7 is a compact 0.5 billion parameter language model based on the Qwen2.5 architecture. It has been specifically fine-tuned by rogeriobayer using the PEFT LoRA method (r=16, alpha=32) for a unique application: 3-label browser tab classification.
Key Capabilities
- Browser Tab Classification: Accurately categorizes browser tabs into three distinct labels: 'focus', 'neutral', and 'distraction'.
- High Accuracy: Achieves an overall test accuracy of 90% and a validation accuracy of 80% for its classification task.
- Efficient Design: Built upon the Qwen2.5-0.5B-Instruct base model, making it suitable for resource-constrained environments.
- Optimized Training: Trained on 156 labeled browsing examples over 5 epochs (with early stopping at epoch 3) using Apple Silicon MPS hardware.
Ideal Use Cases
- Productivity Tools: Integrating into browser extensions or applications to help users manage their focus by identifying distracting tabs.
- On-Device Inference: Its small parameter count and specialized fine-tuning make it well-suited for running directly in web browsers via libraries like Transformers.js, minimizing latency and server reliance.
- Personalized Browsing Experience: Enabling custom rules or notifications based on the classification of active browser content.