Update README.md

README.md

@@ -10,6 +10,11 @@ MedSAM is a fine-tuned version of [SAM](https://huggingface.co/docs/transformers
 
 ## Model Description
 
+MedSAM was trained on a large-scale medical image segmentation dataset with 1,090,486 medical image-mask pairs collected from different publicly available sources including 15 imaging modalities and over 30 cancer types.
+
+MedSAM was initialized with the pre-trained SAM model with the ViT-Base backbone. The prompt encoder's weights were frozen, while all trainable parameters in the image encoder and mask decoder were updated during training. 
+The training was performed for 100 epochs with a batch size of 160 using the AdamW optimizer with a learning rate of 10−4 and a weight decay of 0.01. 
+
 - **Repository:** [MedSAM Official GitHub Repository](https://github.com/bowang-lab/medsam)
 - **Paper:** [Segment Anything in Medical Images](https://arxiv.org/abs/2304.12306v1)