diff --git a/README.md b/README.md
index 0a31338..5fd4c39 100644
--- a/README.md
+++ b/README.md
@@ -84,7 +84,7 @@ TexTeller also supports **formula detection and recognition** on full images, al
 
 ### Download Weights
 
-Chinese and English document formula detection [[link](https://huggingface.co/TonyLee1256/texteller_det/resolve/main/rtdetr_r50vd_6x_coco.onnx?download=true)]: Trained on a total of 11,867 images, consisting of 3,415 images from Chinese textbooks (130+ layouts) and 8,272 images from the IBEM dataset.
+Chinese and English document formula detection [[link](https://huggingface.co/TonyLee1256/texteller_det/resolve/main/rtdetr_r50vd_6x_coco.onnx?download=true)]: Trained on a total of 11,867 images, consisting of 3,415 images from Chinese textbooks (130+ layouts) and 8,272 images from the [IBEM dataset](https://zenodo.org/records/4757865).
 
 ### Formula Detection