Add model card for pretrained backbone

README.md

@@ -0,0 +1,45 @@
+---
+library_name: pytorch
+tags:
+- contrastive-learning
+- tag-classification
+- semantic-search
+- embeddings
+- persona-conditioned
+- pretrained-backbone
+---
+
+# modernbert-base-tag-classification
+
+This is a **pretrained backbone model** (answerdotai/ModernBERT-base) used for tag classification via contrastive learning.
+
+## Model Description
+
+This model uses the `answerdotai/ModernBERT-base` backbone directly without fine-tuning. It's designed for zero-shot tag classification tasks where you want to use a pretrained embedding model for semantic similarity computation.
+
+## Usage
+
+See the README.md for detailed usage examples using our module abstractions.
+
+## Model Architecture
+
+- **Backbone**: `answerdotai/ModernBERT-base`
+- **Type**: Pretrained backbone (no fine-tuning)
+- **Embedding Dimension**: Varies by backbone model
+
+## Citation
+
+If you use this model, please cite:
+
+```bibtex
+@software{{tag_module,
+  title = {{TAG Module: Persona-Conditioned Contrastive Learning for Tag Classification}},
+  author = {{Your Name}},
+  year = {{2025}},
+  url = {{https://github.com/yourusername/tag-module}}
+}}
+```
+
+## License
+
+Please refer to the original model license for the backbone model.