Create README.md

README.md

@@ -0,0 +1,8 @@
+---
+base_model:
+- google/medgemma-27b-text-it
+- google/gemma-3-27b-it
+---
+This is [google/medgemma-27b-text-it](https://huggingface.co/google/medgemma-27b-text-it) with the vision tower and embeddings of [google/gemma-3-27b-it](https://huggingface.co/google/gemma-3-27b-it) attached to restore multimodality. 
+
+**I do not expect the vision function to work with the med data to the same degree as the multimodal 4B medgemma.** It has not been trained beyond the instruct model.