Update README.md

README.md

@@ -591,6 +591,7 @@ Please see sections below this for advanced usage, more details, settings, notes
 
 </small>
 
+IMPORTANT: Make sure to review MIROSTAT sampler settings below, due to behaviour of this specific sampler / affect on parameters/other samplers.
 
 ---
 
@@ -614,11 +615,11 @@ Some of my newer models repo cards use a limited form of this called a "prose co
 
 Roughly a pre-prompt / system role is embedded during each prompt and can act as a guide and/or set of directives for processing the prompt and/or containing generation instructions.
 
-A prose control is a simplifed version of this, which preceeds the main prompt(s) - but the idea / effect is relatively the same (pre-prompt/system role does have a slightly higher priority however).
+A prose control is a simplified version of this, which precedes the main prompt(s) - but the idea / effect is relatively the same (pre-prompt/system role does have a slightly higher priority however).
 
 I strongly suggest you research these online, as they are a powerful addition to your generation toolbox.
 
-They are especially potent with newer model archs due to newer model types having stronger instruction following ablities AND increase context too.
+They are especially potent with newer model archs due to newer model types having stronger instruction following abilities AND increase context too.
 
 ---
 
@@ -777,6 +778,21 @@ For "text-gen-webui" and "Koboldcpp" these are directly accessible (and via Sill
 
 Use Mirostat sampling. "Top K", "Nucleus", "Tail Free" (TFS) and "Locally Typical" (TYPICAL) samplers are ignored if used.  (default: 0, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)
 
+Paper: https://arxiv.org/abs/2007.14966
+
+CRITICAL:
+
+If you activate Mirostat when using "LLAMAcpp SERVER" and/or some LLAMA_CPP based apps this will VOID/DISABLE all parameters (excluding "penalties" ) AND all other SAMPLERS except "temp" parameter plus the following:
+
+V1: n_vocab(model) (this is set internally by llamacpp), seed, mirostat_tau, mirostat_eta
+
+V2: seed, mirostat_tau, mirostat_eta
+
+It is unclear at this time if this occurs using Text Generation UI (and/or using LLAMACPP_HF loader), KoboldCPP or other customized LLM/AI app(s).
+
+("seed" is usually a random value.)
+
+
 "mirostat-lr"                       	
 
 Mirostat learning rate, parameter eta (default: 0.1)  " mirostat_tau "
@@ -1054,7 +1070,7 @@ for operation of CLASS 4 models for chat / role play and/or "smoother operation"
 
 For Class 3 models, "QUADRATIC" will have a slightly stronger effect than "DRY" relatively speaking.
 
-If you use Microstat sampler, keep in mind this will interact with these two advanced samplers too.
+If you use Mirostat sampler, keep in mind this will interact with these two advanced samplers too.
 
 And...