Commit
•
bfa47d2
1
Parent(s):
54d8378
Update README.md
Browse files
README.md
CHANGED
@@ -11,9 +11,9 @@ language:
|
|
11 |
|
12 |
This model is a frankenmerge of [Midnight-Rose-70B-v1.0](https://huggingface.co/sophosympatheia/Midnight-Rose-70B-v1.0) with itself. (See that model card for details on what's in the blend.) It features 120 layers and should weigh in at 103b parameters.
|
13 |
|
14 |
-
|
15 |
|
16 |
-
This model
|
17 |
|
18 |
This model was designed for roleplaying and storytelling and I think it does well at both. It *should* perform well at other tasks, but I haven't tested its capabilities in other areas.
|
19 |
|
@@ -24,7 +24,7 @@ I recommend using the new Min-P sampler method with this model. The creator has
|
|
24 |
I find this model performs reasonably well at 8192 context but you will likely get better results at 4096.
|
25 |
|
26 |
Experiment with any and all of the settings below, but trust me on a few points:
|
27 |
-
*
|
28 |
* Frequency Penalty set to 0.01 is like adding a dash of salt to the dish. Go higher at your own peril. 0 is fine too, but gosh I like 0.01.
|
29 |
|
30 |
If you save the below settings as a .json file, you can import them directly into Silly Tavern.
|
|
|
11 |
|
12 |
This model is a frankenmerge of [Midnight-Rose-70B-v1.0](https://huggingface.co/sophosympatheia/Midnight-Rose-70B-v1.0) with itself. (See that model card for details on what's in the blend.) It features 120 layers and should weigh in at 103b parameters.
|
13 |
|
14 |
+
Midnight Rose is a successor to Rogue Rose and Aurora Nights and improves upon them both. It wants to produce lengthy output by default and is the best creative writing merge I have produced so far.
|
15 |
|
16 |
+
This model is uncensored. *You are responsible for whatever you do with it.*
|
17 |
|
18 |
This model was designed for roleplaying and storytelling and I think it does well at both. It *should* perform well at other tasks, but I haven't tested its capabilities in other areas.
|
19 |
|
|
|
24 |
I find this model performs reasonably well at 8192 context but you will likely get better results at 4096.
|
25 |
|
26 |
Experiment with any and all of the settings below, but trust me on a few points:
|
27 |
+
* I think this model performs best with Min-P in a range of 0.6 - 0.8 with temperature around 1.0 - 1.2.
|
28 |
* Frequency Penalty set to 0.01 is like adding a dash of salt to the dish. Go higher at your own peril. 0 is fine too, but gosh I like 0.01.
|
29 |
|
30 |
If you save the below settings as a .json file, you can import them directly into Silly Tavern.
|