Update README.md
Browse files
README.md
CHANGED
@@ -11,12 +11,13 @@ tags:
|
|
11 |
- story telling
|
12 |
- 32 bit upscale
|
13 |
- ultra high precision
|
|
|
14 |
---
|
15 |
|
16 |
-
<B> Ultra High Quality - 20B Dark Forest Version 1 - 32 bit upscale </b>
|
17 |
|
18 |
Fully rebuilt from master files, including full merge(s) to maintain full 32 bit precision right
|
19 |
-
up until it is compressed into GGUF files.
|
20 |
|
21 |
The result is far superior performance in reasoning, depth, nuence and emotion.
|
22 |
|
@@ -24,22 +25,36 @@ The result is far superior performance in reasoning, depth, nuence and emotion.
|
|
24 |
|
25 |
Better prose, better instruction following and stronger creative brainpower.
|
26 |
|
27 |
-
|
28 |
-
|
29 |
-
All GGUF quants are improved using 32 bit transfer.
|
30 |
-
|
31 |
-
On average this means a q4km operates at Q6 levels and Q6 and Q8 exceed orginal model full precision performance.
|
32 |
|
33 |
Perplexity drop (lower is better) is close to 10% (over 600 points for q4km) for all quants.
|
34 |
|
35 |
-
That means precision has been enhanced for all 20 billion parameters
|
|
|
36 |
|
37 |
-
Quants uploading now.
|
|
|
|
|
38 |
|
39 |
For more details, including a list of enhancements see our other 32 bit upscale rebuild here:
|
40 |
|
41 |
[ https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF ]
|
42 |
|
43 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
|
|
|
45 |
[ https://huggingface.co/TeeZee/DarkForest-20B-v1.0 ]
|
|
|
11 |
- story telling
|
12 |
- 32 bit upscale
|
13 |
- ultra high precision
|
14 |
+
- nsfw
|
15 |
---
|
16 |
|
17 |
+
<B> Ultra High Quality - 20B Dark Forest Version 1.0 - 32 bit upscale </b>
|
18 |
|
19 |
Fully rebuilt from master files, including full merge(s) to maintain full 32 bit precision right
|
20 |
+
up until it is compressed into GGUF files which results on a top to bottom upgrade.
|
21 |
|
22 |
The result is far superior performance in reasoning, depth, nuence and emotion.
|
23 |
|
|
|
25 |
|
26 |
Better prose, better instruction following and stronger creative brainpower.
|
27 |
|
28 |
+
On average this means a q4km operates at Q6 levels and Q6 and Q8 exceeds original model full precision performance.
|
|
|
|
|
|
|
|
|
29 |
|
30 |
Perplexity drop (lower is better) is close to 10% (over 600 points for q4km) for all quants.
|
31 |
|
32 |
+
That means precision has been enhanced for all 20 billion parameters which affects "brain density" / "function",
|
33 |
+
instruction following and output quality.
|
34 |
|
35 |
+
Quants uploading now.
|
36 |
+
|
37 |
+
Imatrix quants to follow shortly.
|
38 |
|
39 |
For more details, including a list of enhancements see our other 32 bit upscale rebuild here:
|
40 |
|
41 |
[ https://huggingface.co/DavidAU/Psyonic-Cetacean-Ultra-Quality-20b-GGUF ]
|
42 |
|
43 |
+
Special thanks to "TEEZEE" for making a fantasic model.
|
44 |
+
|
45 |
+
<b> Info from the original model card: </B>
|
46 |
+
|
47 |
+
Warning: This model can produce NSFW content!
|
48 |
+
|
49 |
+
Results
|
50 |
+
|
51 |
+
- produces SFW nad NSFW content without issues, switches context seamlessly.
|
52 |
+
- good at following instructions.
|
53 |
+
- good at tracking multiple characters in one scene.
|
54 |
+
- very creative, scenarios produced are mature and complicated, model doesn't shy from writing about PTSD, menatal issues or complicated relationships.
|
55 |
+
- NSFW output is more creative and suprising than typical limaRP output.
|
56 |
+
- definitely for mature audiences, not only because of vivid NSFW content but also because of overall maturity of stories it produces.
|
57 |
+
- This is NOT Harry Potter level storytelling.
|
58 |
|
59 |
+
For original model spec and information please visit:
|
60 |
[ https://huggingface.co/TeeZee/DarkForest-20B-v1.0 ]
|