Update README.md

README.md

@@ -31,7 +31,7 @@ tags:
 Too moralizing, really simple as that. I did not find much upside to this model in general, and the downsides are pretty clear. 
 
 # Add PKU-Alignment/ProgressGym-HistLlama3-70B-C013-pretrain-v0.1
-I wanted another base model to regularize, typically the inclusion of base models is purely for regularization and that's the case here.
+I wanted another base model to regularize, typically the inclusion of base models is purely for regularization and that's the case here. (This model is already somewhat represented in the histmerge, so this should act more as weak regularization)
 
 # Add NousResearch/Hermes-3-Llama-3.1-70B
 The inclusion of this seemed fine, potentially helped fix some of the issues with early-ctx instruct following. I could not find any real negatives to this add.