MarinaraSpaghetti
commited on
Commit
•
28486df
1
Parent(s):
7a31b05
Upload README (1).md
Browse files- README (1).md +98 -0
README (1).md
ADDED
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model: []
|
3 |
+
library_name: transformers
|
4 |
+
tags:
|
5 |
+
- mergekit
|
6 |
+
- merge
|
7 |
+
---
|
8 |
+
|
9 |
+
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6550b16f7490049d6237f200/DxZNdV33EVq6cK6_gwSqS.jpeg)
|
10 |
+
|
11 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/6550b16f7490049d6237f200/sPI_QHGXE_egmQXTsYkld.png)
|
12 |
+
|
13 |
+
# Information
|
14 |
+
## Details
|
15 |
+
New merge of NeMo based models, thankfully this time with ChatML format. My goal was to create a smart and universal roleplaying model that is stable on higher contexts. So far seems to be better than my best Nemomix attempts, especially on the 64k+ context I've been using. All credits and thanks go to the amazing Gryphe, MistralAI, Anthracite, Sao10K and ShuttleAI for their amazing models.
|
16 |
+
|
17 |
+
## Instruct
|
18 |
+
|
19 |
+
ChatML but Mistral Instruct should work too (theoretically).
|
20 |
+
|
21 |
+
```
|
22 |
+
<|im_start|>system
|
23 |
+
{system}<|im_end|>
|
24 |
+
<|im_start|>user
|
25 |
+
{message}<|im_end|>
|
26 |
+
<|im_start|>assistant
|
27 |
+
{response}<|im_end|>
|
28 |
+
```
|
29 |
+
|
30 |
+
## Parameters
|
31 |
+
|
32 |
+
I recommend running Temperature 1.0-1.2 with 0.1 Top A or 0.01-0.1 Min P, and with 0.8/1.75/2/0 DRY. Also works with lower temperatures below 1.0.
|
33 |
+
|
34 |
+
### Settings
|
35 |
+
|
36 |
+
You can use my exact settings from here: https://huggingface.co/MarinaraSpaghetti/SillyTavern-Settings/tree/main
|
37 |
+
|
38 |
+
## GGUF
|
39 |
+
|
40 |
+
https://huggingface.co/MarinaraSpaghetti/NemoRemix-12B-GGUF
|
41 |
+
|
42 |
+
# NemoRemix-v4.0-12B
|
43 |
+
|
44 |
+
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|
45 |
+
|
46 |
+
## Merge Details
|
47 |
+
|
48 |
+
### Merge Method
|
49 |
+
|
50 |
+
This model was merged using the della_linear merge method using F:\mergekit\mistralaiMistral-Nemo-Base-2407 as a base.
|
51 |
+
|
52 |
+
### Models Merged
|
53 |
+
|
54 |
+
The following models were included in the merge:
|
55 |
+
* F:\mergekit\mistralaiMistral-Nemo-Instruct-2407
|
56 |
+
* F:\mergekit\Gryphe_Pantheon-RP-1.5-12b-Nemo
|
57 |
+
* F:\mergekit\shuttleai_shuttle-2.5-mini
|
58 |
+
* F:\mergekit\Sao10K_MN-12B-Lyra-v1
|
59 |
+
* F:\mergekit\anthracite-org_magnum-12b-v2
|
60 |
+
|
61 |
+
### Configuration
|
62 |
+
|
63 |
+
The following YAML configuration was used to produce this model:
|
64 |
+
|
65 |
+
```yaml
|
66 |
+
models:
|
67 |
+
- model: F:\mergekit\Gryphe_Pantheon-RP-1.5-12b-Nemo
|
68 |
+
parameters:
|
69 |
+
weight: 0.1
|
70 |
+
density: 0.3
|
71 |
+
- model: F:\mergekit\mistralaiMistral-Nemo-Instruct-2407
|
72 |
+
parameters:
|
73 |
+
weight: 0.12
|
74 |
+
density: 0.4
|
75 |
+
- model: F:\mergekit\Sao10K_MN-12B-Lyra-v1
|
76 |
+
parameters:
|
77 |
+
weight: 0.2
|
78 |
+
density: 0.5
|
79 |
+
- model: F:\mergekit\shuttleai_shuttle-2.5-mini
|
80 |
+
parameters:
|
81 |
+
weight: 0.25
|
82 |
+
density: 0.6
|
83 |
+
- model: F:\mergekit\anthracite-org_magnum-12b-v2
|
84 |
+
parameters:
|
85 |
+
weight: 0.33
|
86 |
+
density: 0.8
|
87 |
+
merge_method: della_linear
|
88 |
+
base_model: F:\mergekit\mistralaiMistral-Nemo-Base-2407
|
89 |
+
parameters:
|
90 |
+
epsilon: 0.05
|
91 |
+
lambda: 1
|
92 |
+
dtype: bfloat16
|
93 |
+
```
|
94 |
+
|
95 |
+
# Ko-fi
|
96 |
+
## Enjoying what I do? Consider donating here, thank you!
|
97 |
+
|
98 |
+
https://ko-fi.com/spicy_marinara
|