diff --git "a/docs/source/BestPractices/Embedding\350\256\255\347\273\203.md" "b/docs/source/BestPractices/Embedding\350\256\255\347\273\203.md"
index 5d4ce3a350..cde99fbd8a 100644
--- "a/docs/source/BestPractices/Embedding\350\256\255\347\273\203.md"
+++ "b/docs/source/BestPractices/Embedding\350\256\255\347\273\203.md"
@@ -73,5 +73,5 @@ loss的源代码可以在[这里](https://github.com/modelscope/ms-swift/blob/ma
 
 SWIFT提供了两个脚手架训练脚本：
 
-- [gte模型](../../../examples/train/embedding/train_gte.sh)
-- [gme模型](../../../examples/train/embedding/train_gme.sh)
+- [gte模型](https://github.com/tastelikefeet/swift/blob/main/examples/train/embedding/train_gte.sh)
+- [gme模型](https://github.com/tastelikefeet/swift/blob/main/examples/train/embedding/train_gme.sh)
diff --git a/docs/source_en/BestPractices/Embedding.md b/docs/source_en/BestPractices/Embedding.md
index 7062a34906..8df6497b29 100644
--- a/docs/source_en/BestPractices/Embedding.md
+++ b/docs/source_en/BestPractices/Embedding.md
@@ -72,5 +72,5 @@ When using a dataset in `infonce` format, please set the values of `per_device_t
 
 SWIFT provides two scaffold training scripts:
 
-- [GTE Model](../../../examples/train/embedding/train_gte.sh)
-- [GME Model](../../../examples/train/embedding/train_gme.sh)
+- [GTE Model](https://github.com/tastelikefeet/swift/blob/main/examples/train/embedding/train_gte.sh)
+- [GME Model](https://github.com/tastelikefeet/swift/blob/main/examples/train/embedding/train_gme.sh)