Add model card

README.md

@@ -0,0 +1,63 @@
+---
+license: mit
+language: es
+tags:
+  - biomedical
+  - spanish
+  - entity-linking
+  - sapbert
+  - bi-encoder
+  - umls
+  - knowledge-graph
+  - clinical
+---
+
+# ClinLinker-KB-P
+
+**ClinLinker-KB-P** enhances ClinLinker by leveraging hierarchical relations from the UMLS knowledge graph. It includes not only synonyms but also their **parent concepts** as positive training pairs, enabling more robust semantic generalization.
+
+## 🧠 Training Details
+
+- Base model: `PlanTL-GOB-ES/roberta-base-biomedical-clinical-es`
+- Data: UMLS Spanish concepts + `parent_of` relations
+- Strategy: Contrastive training with concept parents as positives
+
+## 📚 Citation
+
+> Gallego, Fernando, et al. "Clinlinker-Kb: Clinical Entity Linking in Spanish with Knowledge-Graph Enhanced Biencoders." Available at SSRN 4939986.
+
+## 💡 Recommended Usage
+
+We recommend using this model together with:
+
+- [Faiss](https://github.com/facebookresearch/faiss)
+- Or the `FaissEncoder` from [ICB-UMA/KnowledgeGraph](https://github.com/ICB-UMA/KnowledgeGraph)
+
+## 📚 Citation
+
+> Gallego, Fernando and López-García, Guillermo and Gasco, Luis and Krallinger, Martin and Veredas, Francisco J., Clinlinker-Kb: Clinical Entity Linking in Spanish with Knowledge-Graph Enhanced Biencoders. Available at SSRN:http://dx.doi.org/10.2139/ssrn.4939986
+
+## 💡 Recommended Usage
+
+We recommend using this model together with:
+
+- [Faiss](https://github.com/facebookresearch/faiss) for similarity search
+- Or the `FaissEncoder` utility available at [ICB-UMA/KnowledgeGraph](https://github.com/ICB-UMA/KnowledgeGraph)
+
+## 🧪 Example: Encoding a Spanish Mention
+
+```python
+from transformers import AutoTokenizer, AutoModel
+import torch
+
+tokenizer = AutoTokenizer.from_pretrained("ICB-UMA/ClinLinker-KB-P")
+model = AutoModel.from_pretrained("ICB-UMA/ClinLinker-KB-P")
+
+mention = "insuficiencia renal aguda"
+inputs = tokenizer(mention, return_tensors="pt", padding=True, truncation=True)
+with torch.no_grad():
+    outputs = model(**inputs)
+embedding = outputs.last_hidden_state[:, 0, :]  # CLS token
+
+print(embedding.shape)  
+