PatentSBERTa – Final Green Patent Detection Model
This model is fine-tuned for binary classification of green patents.
Training Data
- Silver labels (automatic LLM labeling)
- Gold labels (QLoRA-powered MAS + Targeted Human-in-the-loop)
Architecture
- Base model: AI-Growth-Lab/PatentSBERTa
- Task: Binary sequence classification
- Labels:
Final Pipeline
Silver + Gold (QLoRA MAS + Targeted HITL)
Evaluation
Metric: F1-score on test split
