Update README.md
Browse files
README.md
CHANGED
@@ -31,18 +31,21 @@ Full disclosure, this is an experimental model that needs some more finetuning d
|
|
31 |
|
32 |
(Still Testing; details might change)
|
33 |
|
34 |
-
Blackfall Summanus is a good balance between the chaos and literacy. Can get very explicit once in eRP.
|
35 |
-
|
|
|
|
|
|
|
36 |
|
37 |
For a merge made from some of the most temperature intolerant models I've seen, Summanus is fine with high temps. Looses coherency past 2, but much better than Blackened Sunfall. Also, this model works best of default template setting. Llama 3 is okay, doesn't have the same 'sparkle' as default but ChatML straight up can break it.
|
38 |
-
I find I get better results with Top P & K, but min-p is great too; lean
|
39 |
|
40 |
Rec. Settings:
|
41 |
```
|
42 |
Template: Model Default
|
43 |
Temperature: 1.4
|
44 |
Top P: 0.91
|
45 |
-
Top K:
|
46 |
Repeat Penelty: 1.05
|
47 |
Repeat Penelty Tokens: 256
|
48 |
```
|
|
|
31 |
|
32 |
(Still Testing; details might change)
|
33 |
|
34 |
+
Blackfall Summanus is a good balance between the chaos and literacy. Can get very explicit once in eRP.
|
35 |
+
Still has the Jamet tendency to speak for the user, but is much better at listening to isntructions so but a note in the model instrustions like 'avoid speaking on behalf of {user}' and that'll curb it.
|
36 |
+
Dialogue is much more natural and connotative compared to Aethora v2 but doesn't ramble on like Blackened Sunfall.
|
37 |
+
Understands nuance and context better from being a larger(15B) model and from the additional training in Aethora v2, yet generates more vivid and human text.
|
38 |
+
Likes sticking close to the example text. Leans romantic? The artistic movement, not the genre.
|
39 |
|
40 |
For a merge made from some of the most temperature intolerant models I've seen, Summanus is fine with high temps. Looses coherency past 2, but much better than Blackened Sunfall. Also, this model works best of default template setting. Llama 3 is okay, doesn't have the same 'sparkle' as default but ChatML straight up can break it.
|
41 |
+
I find I get better results with Top P & K, but min-p is great too; lean around 0.8.
|
42 |
|
43 |
Rec. Settings:
|
44 |
```
|
45 |
Template: Model Default
|
46 |
Temperature: 1.4
|
47 |
Top P: 0.91
|
48 |
+
Top K: 30
|
49 |
Repeat Penelty: 1.05
|
50 |
Repeat Penelty Tokens: 256
|
51 |
```
|