Update README.md

README.md

@@ -9,19 +9,17 @@ app_file: app.py
 pinned: false
 ---
 
-# 🩻 MedSigLIP-448 Demo (FP16 + Pruned)
+# 🩻 MedSigLIP → MedGemma Radiology Fusion
 
-This Space hosts a lightweight demo of the model  
-[`fokan/medsiglip-448-fp16-pruned20`](https://huggingface.co/fokan/medsiglip-448-fp16-pruned20).
+This demo chains two models together:
 
-### ⚙️ Features
-- Vision encoder specialized for **radiology embeddings**
-- **20% pruned + FP16** for faster inference
-- Works with **X-ray / CT / MRI images**
+- **Encoder:** `fokan/medsiglip-448-fp16-pruned20` (extracts radiology embeddings)
+- **Decoder:** `fokan/medgemma-4b-it-fp16-pruned20` (generates radiology-style reports)
 
-### 🧪 Try It
-Upload any medical image and view its embedding summary instantly.
+### ⚙️ Workflow
+1. Upload an X-ray (or any medical image)
+2. Model encodes it into semantic embedding
+3. Decoder interprets embedding → generates medical report
 
 Built by [Karrar Alhdrawi](https://huggingface.co/fokan)
 
-