DavidAU commited on
Commit
bcb2283
1 Parent(s): 1f6f411

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -4
README.md CHANGED
@@ -97,7 +97,7 @@ This model is capable of horror, science fiction, romance - you name it.
97
  This model has a very strong VIDIDINESS bias. It generates extremely vivid prose, description, and dialog as well
98
  as in the moment metaphors and similes but rarely uses "cliches".
99
 
100
- It also has a STRONG horror bias, although it will generate content for almost any genre. That being said
101
  if there is a "hint" of things going wrong... they will.
102
 
103
  It will also swear (R-18) like there is no tomorrow at times and "dark" characters will be VERY dark so to speak.
@@ -118,7 +118,7 @@ This model also does not show an "GPTisms" (NO happy ever after, NO morality pol
118
 
119
  (see examples section below for prose output)
120
 
121
- Because of the nature of this merge most attributes of each of the 3 models will be in this rebuilt 16.5B model as opposed to the
122
  original 8B model where some of one or more of the model's features and/or strengths maybe reduced or overshadowed.
123
 
124
  With the triple step merge these qualities are further amplified.
@@ -159,13 +159,13 @@ Please note that this model will NOT reject any request.
159
  NOTES:
160
 
161
  - Lower temp if there is an issue with instruction following.
162
- - This model has a stronger horror BIAS. Reducing temp changes this.
163
  - Small changes to your prompt(s) have a stronger impact.
164
- - Other versions are in the works, with lower horror bias.
165
  - Temps as lower as .1 work well and are very creative, but you can use .8 or higher.
166
  - Temp range 0 to 2. Suggest starting at .2 to .5, as this model runs hots.
167
  - Rep pen of 1.1 is the default suggested, and used for examples generation.
168
  - For chat/roleplay a rep pen of 1.12-1.15 is recommended. This will lower creativity, but raise stability.
 
169
 
170
  <B>Versions of GRAND HORROR that can be used for all storytelling / all genres (in release order):</B>
171
 
 
97
  This model has a very strong VIDIDINESS bias. It generates extremely vivid prose, description, and dialog as well
98
  as in the moment metaphors and similes but rarely uses "cliches".
99
 
100
+ It also has a VERY STRONG horror bias, although it will generate content for almost any genre. That being said
101
  if there is a "hint" of things going wrong... they will.
102
 
103
  It will also swear (R-18) like there is no tomorrow at times and "dark" characters will be VERY dark so to speak.
 
118
 
119
  (see examples section below for prose output)
120
 
121
+ Because of the nature of this merge most attributes of each of the 3 models will be in this rebuilt 20.7B model as opposed to the
122
  original 8B model where some of one or more of the model's features and/or strengths maybe reduced or overshadowed.
123
 
124
  With the triple step merge these qualities are further amplified.
 
159
  NOTES:
160
 
161
  - Lower temp if there is an issue with instruction following.
162
+ - This model has a strong horror BIAS. Reducing temp/rep pen changes this.
163
  - Small changes to your prompt(s) have a stronger impact.
 
164
  - Temps as lower as .1 work well and are very creative, but you can use .8 or higher.
165
  - Temp range 0 to 2. Suggest starting at .2 to .5, as this model runs hots.
166
  - Rep pen of 1.1 is the default suggested, and used for examples generation.
167
  - For chat/roleplay a rep pen of 1.12-1.15 is recommended. This will lower creativity, but raise stability.
168
+ - Much higher rep pens (1.3+) will reduce vividness / horror levels.
169
 
170
  <B>Versions of GRAND HORROR that can be used for all storytelling / all genres (in release order):</B>
171