HiroseKoichi commited on
Commit
4ad4542
1 Parent(s): 7c2a3b1

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +118 -0
README.md ADDED
@@ -0,0 +1,118 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: llama3
3
+ library_name: transformers
4
+ tags:
5
+ - nsfw
6
+ - not-for-all-audiences
7
+ - llama-3
8
+ - text-generation-inference
9
+ - moe
10
+ - mergekit
11
+ - merge
12
+ ---
13
+
14
+ # Llama-Salad-8x8B
15
+
16
+ # Details
17
+ - **License**: [llama3](https://llama.meta.com/llama3/license/)
18
+ - **Instruct Format**: [llama-3](https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3/)
19
+ - **Context Size**: 8K
20
+
21
+ ## Models Used
22
+ - [Meta-Llama-3-8B-Instruct](https://huggingface.co/NousResearch/Meta-Llama-3-8B-Instruct)
23
+ - [Llama-3-8B-Synthia-v3.5](https://huggingface.co/migtissera/Llama-3-8B-Synthia-v3.5)
24
+ - [Llama-3-Soliloquy-8B-v2](https://huggingface.co/openlynn/Llama-3-Soliloquy-8B-v2)
25
+ - [opus-v1.2-llama-3-8b-instruct-run3.5-epoch2.5](https://huggingface.co/dreamgen-preview/opus-v1.2-llama-3-8b-instruct-run3.5-epoch2.5)
26
+ - [Einstein-v6.1-Llama3-8B](https://huggingface.co/Weyaxi/Einstein-v6.1-Llama3-8B)
27
+ - [suzume-llama-3-8B-multilingual](https://huggingface.co/lightblue/suzume-llama-3-8B-multilingual)
28
+ - [Llama-3-8B-UltraMedical](https://huggingface.co/TsinghuaC3I/Llama-3-8B-UltraMedical)
29
+ - [Llama-3-8B-Instruct-Coder](https://huggingface.co/rombodawg/Llama-3-8B-Instruct-Coder)
30
+
31
+ ## Merge Config
32
+ ```yaml
33
+ base_model: migtissera/Llama-3-8B-Synthia-v3.5
34
+ gate_mode: hidden
35
+ dtype: bfloat16
36
+ experts_per_token: 2
37
+ experts:
38
+ - source_model: rombodawg/Llama-3-8B-Instruct-Coder
39
+ positive_prompts:
40
+ - "programming language"
41
+ - "JavaScript"
42
+ - "Python programming language"
43
+ - "Rust programming language"
44
+ - "C++ programming language"
45
+ - "GO programming language"
46
+ - "Ruby programming language"
47
+ - "Haskell programming language"
48
+ - "SQL query language"
49
+ - "CSS markup styling language"
50
+ - "code"
51
+ - source_model: openlynn/Llama-3-Soliloquy-8B-v2
52
+ positive_prompts:
53
+ - "characters"
54
+ - "scene"
55
+ - "roleplay"
56
+ - "erotic roleplay"
57
+ - "sexual fetish"
58
+ - "NSFW"
59
+ negative_prompts:
60
+ - "biology"
61
+ - source_model: dreamgen-preview/opus-v1.2-llama-3-8b-instruct-run3.5-epoch2.5
62
+ positive_prompts:
63
+ - "creative writing"
64
+ - "storytelling"
65
+ - "narration"
66
+ - "narrative setting"
67
+ - "narrative plot"
68
+ - "narrative exposition"
69
+ - "narrative theme"
70
+ - "narrative climax"
71
+ - source_model: Weyaxi/Einstein-v6.1-Llama3-8B
72
+ positive_prompts:
73
+ - "science"
74
+ - "physics"
75
+ - "chemistry"
76
+ - "biology"
77
+ - "math"
78
+ - "step-by-step"
79
+ - "logical reasoning"
80
+ negative_prompts:
81
+ - "programming language"
82
+ - source_model: migtissera/Llama-3-8B-Synthia-v3.5
83
+ positive_prompts:
84
+ - "summarize"
85
+ - "paraphrase"
86
+ - "list"
87
+ - "explain"
88
+ - "define"
89
+ - "analyze"
90
+ - "rephrase"
91
+ - "elaborate"
92
+ - source_model: lightblue/suzume-llama-3-8B-multilingual
93
+ positive_prompts:
94
+ - "multilingual"
95
+ - "language translation"
96
+ - "日本語"
97
+ - "汉语"
98
+ - "Deutsch"
99
+ - "Français"
100
+ - "русский язык"
101
+ negative_prompts:
102
+ - "programming language"
103
+ - "English"
104
+ - source_model: TsinghuaC3I/Llama-3-8B-UltraMedical
105
+ positive_prompts:
106
+ - "anatomy"
107
+ - "medical diagnosis"
108
+ - "symptom"
109
+ - "healthcare"
110
+ - "medicine"
111
+ - "medication"
112
+ negative_prompts:
113
+ - "sexual fetish"
114
+ - source_model: NousResearch/Meta-Llama-3-8B-Instruct
115
+ positive_prompts:
116
+ - "chat"
117
+ - "conversation"
118
+ ```