fix urls

app.py

@@ -165,8 +165,8 @@ Researchers use [photographic identification](https://whalescientists.com/photo-
 decades to study their migration, population, and behavior. While this is a 
 tedious and costly process, it is tempting to leverage the huge amount of 
 image data collected by the whale-watching community and private encounters around 
-the globe. Organizations like [WildMe](www.wildme.org) or 
-[happywhale](www.happywhale.com) develop AI models for automated identification at
+the globe. Organizations like [WildMe](https://www.wildme.org) or 
+[happywhale](https://www.happywhale.com) develop AI models for automated identification at
 scale. To push the state-of-the-art, happywhale hosted two competitions on kaggle, 
 the 2018 [Humpback Whale Identification](https://www.kaggle.com/c/humpback-whale-identification) 
 and the 2022 [Happywhale](https://www.kaggle.com/competitions/happy-whale-and-dolphin) 
@@ -180,7 +180,9 @@ competition images with auto- or manually generated labels.
 
 Below you can test my solution (down-cut version) on your own images. 
 The detector is an ensemble of five YOLOv5 models, the identifier ensembles three
-models with EfficientNet-B7, EfficientNetV2-XL, and ConvNext-base backbone.
+models with EfficientNet-B7, EfficientNetV2-XL, and ConvNext-base backbone. 
+You can find model code and training pipelines in the 
+[DeepTrane](https://github.com/yellowdolphin/deeptrane) repository.
 """  # appears between title and input/output
 
 article = """