Update README.md

README.md

@@ -56,13 +56,18 @@ models into one massive powerhouse at 18.4B parameters (equal to 24B - 8 X 3B).
 
 This model's instruction following, and output generation for creative writing, prose, fiction and role play are exceptional.
 
-This model is also "gated", contains a master reasoning model (this can be turned on/off) and was built at float32 precision.
+This model is also "gated", contains a master reasoning model (this can be turned on/off) and was built at float32 (32 bit) precision.
 
-This is the "reasoning / gated version" of this model:
+The "gated" strucure means the "reasoning model" is re-inforced by the other 7 models in the MOE during reasoning, and then during
+output generation / non-reasoning the non-reasoning model(s) take control.
+
+Also, with "gating" you can directly access/control the model(s) you want to use during instruction following and generation.
+
+This model is the "reasoning / gated version" of this model:
 
 [ https://huggingface.co/DavidAU/Llama-3.2-8X3B-MOE-Dark-Champion-Instruct-uncensored-abliterated-18.4B-GGUF ] 
 
-(193 likes // 447,000+ downloads to date. )
+(193 likes // 447,000+ downloads to date. (Hugging face only shows last 30 days at the repo) )
 
 And it is fast: 50+ t/s (2 experts) on a low end 16GB card, IQ4XS.