Update README.md

README.md

@@ -41,7 +41,9 @@ creating a 19B model with the "Abliterated" (Uncensored) version of Deepseek Qwe
 
 The model is just over 19B because of the unqiue "shared expert" (roughly 2.5 models here) used in Qwen MOEs.
 
-The oddball configuration yields interesting "thinking/reasoning" which is stronger than either 7B model on its own.
+This "oddball" configuration yields interesting "thinking/reasoning" which is stronger than either 7B model on its own.
+
+And you can use any temp settings you want (rather than a narrow range of .4 to .8), and the model will still "think/reason".
 
 Five example generations at the bottom of this page.
 
@@ -112,6 +114,17 @@ SOFTWARE patch (by me) for Silly Tavern (front end to connect to multiple AI app
 
 ---
 
+Known Issues:
+
+---
+
+From time to time model will generate some Chinese symbols/characters, especially at higher temps. This is normal
+for DeepSeek Distill models.
+
+Reasoning/Thinking may be a little "odd" at temps 1.5+ ; you may need to regen to get a better response.
+
+---
+
 <h2>Example Generation:</h2>
 
 IQ4XS Quant, Temp 1.5, rep pen 1.06, topp: .95, minp: .05, topk: 40