Update README.md

README.md

@@ -32,6 +32,8 @@ Splits are of the form `ratio{negative_to_positive_ratio}_v{1 or 2}`. For exampl
 
 `ratio6_v2` was used to train https://huggingface.co/noanabeshima/forecastability-classifier-v1.
 
-Prompt can be found in `prompt.txt`. It was iterated on slightly using a small set of human labels. Claude considers ~2% of fineweb documents to be forecastable.
+Prompt can be found in `prompt.txt`. It was iterated on slightly using a small set of human labels. GPT-4.1 performed slightly better, but Claude had a more favorable TOS for open-sourcing data/models.
+
+Claude considers ~2% of fineweb documents to be forecastable.
 
 Made with the help of Collin Gray.