zarakiquemparte commited on
Commit
1210b80
1 Parent(s): 2de3bfe

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +64 -0
README.md ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ tags:
4
+ - llama2
5
+ ---
6
+ # Model Card: ZaraRP L2 7b
7
+ This model uses [Nous Hermes Llama2 7b](https://huggingface.co/NousResearch/Nous-Hermes-llama-2-7b) (53%) as a base with [Stable Beluga 7b](https://huggingface.co/stabilityai/StableBeluga-7B) (47%) and the result of this merge was merged with [LimaRP LLama2 7B Lora version of the day 07/23/2023](https://huggingface.co/lemonilia/limarp-llama2) and [PIPPA ShareGPT Subset Variation Two Lora 7b](https://huggingface.co/zarakiquemparte/PIPPA-ShareGPT-Subset-Lora-VT-7b).
8
+
9
+ This merge of models(hermes and stable beluga) was done with this [script](https://github.com/zarakiquemparte/zaraki-tools/blob/main/merge-cli.py)
10
+
11
+ This merge of Lora with Model was done with this [script](https://github.com/zarakiquemparte/zaraki-tools/blob/main/apply-lora.py)
12
+
13
+ Merge illustration:
14
+
15
+ ![illustration](zararp-merge-illustration.png)
16
+
17
+ ## Usage:
18
+
19
+ Since this is a merge between Nous Hermes, Stable Beluga, LimaRP, and PIPPA ShareGPT, the following instruction formats should work:
20
+
21
+ Alpaca 2:
22
+
23
+ ```
24
+ ### Instruction:
25
+ <prompt>
26
+
27
+ ### Response:
28
+ <leave a newline blank for model to respond>
29
+ ```
30
+
31
+ Custom:
32
+
33
+ ```
34
+ SYSTEM: Do thing
35
+ USER: {prompt}
36
+ CHARACTER:
37
+ ```
38
+
39
+ Alpaca LimaRP instruction format:
40
+
41
+ ```
42
+ ### Instruction:
43
+ Character's Persona: {bot character description}
44
+
45
+ User's Persona: {user character description}
46
+
47
+ Scenario: {what happens in the story}
48
+
49
+ Play the role of Character. You must engage in a roleplaying chat with User below this line. Do not write dialogues and narration for User. Character should respond with messages of medium length.
50
+
51
+ ### Input:
52
+ User: {utterance}
53
+
54
+ ### Response:
55
+ Character: {utterance}
56
+ ```
57
+
58
+ ## Bias, Risks, and Limitations
59
+
60
+ This model is not intended for supplying factual information or advice in any form
61
+
62
+ ## Training Details
63
+
64
+ This model is merged and can be reproduced using the tools mentioned above. Please refer to all provided links for extra model-specific details.