Make GitHub evaluation code prominent in dataset card

README.md

@@ -13,6 +13,8 @@ configs:
 
 # BEANS-Next
 
+**[GitHub: evaluation code and instructions](https://github.com/earthspecies/beans-next)**
+
 BEANS-Next evaluates audio-language models on **45 bioacoustic tasks**, with **37,718 examples** and all evaluation audio included.
 
 | Tier | Focus | Examples |
@@ -28,6 +30,9 @@ The Gibbon support order and answer letters match the evaluated paper subset. Th
 
 ## Quick start
 
+To evaluate a model, follow the [installation and evaluation instructions on GitHub](https://github.com/earthspecies/beans-next#evaluate).
+The repository provides model adapters, benchmark suites, and scoring tools.
+
 Load the `test` split, then download the audio for an example:
 
 ```python