DavidAU commited on
Commit
b5ab98b
1 Parent(s): 3217d13

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +44 -0
README.md ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language:
4
+ - en
5
+ tags:
6
+ - story
7
+ - general usage
8
+ - roleplay
9
+ - creative
10
+ - rp
11
+ - fantasy
12
+ - story telling
13
+ - ultra high precision
14
+ ---
15
+ <B>NEO CLASS Ultra Quant for : L3-8B-Stheno-v3.2</B>
16
+
17
+ Additional quants are uploading...
18
+
19
+ The NEO Class tech was created after countless investigations and over 120 lab experiments backed by
20
+ real world testing and qualitative results.
21
+
22
+ NEO Class results:
23
+
24
+ Better overall function, instruction following, output quality and stronger connections to ideas, concepts and the world in general.
25
+
26
+ In addition quants now operate above their "grade" so to speak :
27
+
28
+ IE: Q4 / IQ4 operate at Q5KM/Q6 levels.
29
+
30
+ Likewise for Q3/IQ3 operate at Q4KM/Q5 levels.
31
+
32
+ The examples below illustrate the least amount of improvement using some of the lowest quants.
33
+
34
+ Perplexity drop of 1191 points VS regular quant of IQ4XS.
35
+
36
+ (lower is better)
37
+
38
+ <B> Model Notes: </B>
39
+
40
+ Maximum context is 8k. Please see original model maker's page for details, and usage information for this model.
41
+
42
+ Special thanks to the model creators at SAO10K for making such a fantastic model:
43
+
44
+ [ https://huggingface.co/Sao10K/L3-8B-Stheno-v3.2 ]