Update README.md

README.md

@@ -14,6 +14,26 @@ tags:
 - medical
 ---
 
+## 📚 Dataset
+
+This model was trained on the full [RAGBench](https://huggingface.co/datasets/galileo-ai/ragbench) dataset:
+
+> Galileo AI. *RAGBench: A Benchmark Suite for Evaluating Retrieval-Augmented Generation.*  
+> Available at: [https://huggingface.co/datasets/galileo-ai/ragbench](https://huggingface.co/datasets/galileo-ai/ragbench)
+
+RAGBench includes multiple domains:
+- ✅ Biomedical (e.g. PubMedQA, emrQA)
+- ✅ Clinical (e.g. MedQA)
+- ✅ General open-domain
+
+Each sample in the dataset includes:
+- A user or clinician-formulated question
+- A multi-sentence context
+- Binary labels indicating whether each sentence is relevant
+
+The model was trained using sentence-level supervision across all available splits (`train`, `dev`, `test` where applicable) with document examples drawn from **all included datasets** in the benchmark.
+
+
 ## Installation
 
 ```bash