Update README.md

README.md

@@ -9,7 +9,7 @@ datasets:
 
 Day 2 RP finetune of Apriel 15B, with several iterative improvements from the first version. In particular, coherence at good temperatures (~.7) should be much higher.
 
-Compared to the first version I merged 20% of the instruct checkpoint back in to mitigate forgetting and to preserve more of the base model's style.
+Compared to the first version I merged 20% of the original instruct checkpoint back in to mitigate forgetting and to preserve more of the original model's style.
 
 I also fully converted the model to use the Phi 3 format; this comes at the slight tradeoff of the `<|end|>` tag not always tokenizing exactly the same way in a few niche scenarios.
 
@@ -31,6 +31,8 @@ Similar to the Qwen 3 model line, Apriel R1P can be used with or without thinkin
 
 The chat template has been converted to a Phi 3 template as the model seemed to respond best to this format.
 
+This model does prefer having character cards placed in user messages, not the system prompt.
+
 ## Special Thanks:
 
 Undi95 for portions of their dataset and inspiration.