Undi95 commited on
Commit
a554123
1 Parent(s): ece757a

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +188 -0
README.md ADDED
@@ -0,0 +1,188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-nc-4.0
3
+ base_model:
4
+ - vicgalle/Roleplay-Llama-3-8B
5
+ - Undi95/Llama-3-Unholy-8B-e4
6
+ - Undi95/Llama-3-LewdPlay-8B
7
+ library_name: transformers
8
+ tags:
9
+ - mergekit
10
+ - merge
11
+
12
+ ---
13
+ # LewdPlay-8B
14
+
15
+ This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
16
+
17
+ The new EVOLVE merge method was used (on MMLU specifically), see below for more information!
18
+
19
+ Unholy was used for uncensoring, Roleplay Llama 3 for the DPO train he got on top, and LewdPlay for the... lewd side.
20
+
21
+ ## Prompt template: Llama3
22
+
23
+ ```
24
+ <|begin_of_text|><|start_header_id|>system<|end_header_id|>
25
+
26
+ {system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
27
+
28
+ {input}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
29
+
30
+ {output}<|eot_id|>
31
+ ```
32
+
33
+ ## Merge Details
34
+ ### Merge Method
35
+
36
+ This model was merged using the [DARE](https://arxiv.org/abs/2311.03099) [TIES](https://arxiv.org/abs/2306.01708) merge method using ./mergekit/input_models/Roleplay-Llama-3-8B_213413727 as a base.
37
+
38
+ ### Models Merged
39
+
40
+ The following models were included in the merge:
41
+ * ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
42
+ * ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
43
+
44
+ ### Configuration
45
+
46
+ The following YAML configuration was used to produce this model:
47
+
48
+ ```yaml
49
+ base_model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
50
+ dtype: bfloat16
51
+ merge_method: dare_ties
52
+ parameters:
53
+ int8_mask: 1.0
54
+ normalize: 0.0
55
+ slices:
56
+ - sources:
57
+ - layer_range: [0, 4]
58
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
59
+ parameters:
60
+ density: 1.0
61
+ weight: 0.6861808716092435
62
+ - layer_range: [0, 4]
63
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
64
+ parameters:
65
+ density: 0.6628290134113985
66
+ weight: 0.5815923052193855
67
+ - layer_range: [0, 4]
68
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
69
+ parameters:
70
+ density: 1.0
71
+ weight: 0.5113886163963061
72
+ - sources:
73
+ - layer_range: [4, 8]
74
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
75
+ parameters:
76
+ density: 0.892655547455918
77
+ weight: 0.038732602391021484
78
+ - layer_range: [4, 8]
79
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
80
+ parameters:
81
+ density: 1.0
82
+ weight: 0.1982145486303527
83
+ - layer_range: [4, 8]
84
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
85
+ parameters:
86
+ density: 1.0
87
+ weight: 0.6843011350690802
88
+ - sources:
89
+ - layer_range: [8, 12]
90
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
91
+ parameters:
92
+ density: 0.7817511027396784
93
+ weight: 0.13053333213489704
94
+ - layer_range: [8, 12]
95
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
96
+ parameters:
97
+ density: 0.6963703515864826
98
+ weight: 0.20525481492667985
99
+ - layer_range: [8, 12]
100
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
101
+ parameters:
102
+ density: 0.6983086326765777
103
+ weight: 0.5843953969574106
104
+ - sources:
105
+ - layer_range: [12, 16]
106
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
107
+ parameters:
108
+ density: 0.9632895768462915
109
+ weight: 0.2101146706607748
110
+ - layer_range: [12, 16]
111
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
112
+ parameters:
113
+ density: 0.597557434542081
114
+ weight: 0.6728172621848589
115
+ - layer_range: [12, 16]
116
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
117
+ parameters:
118
+ density: 0.756263557607837
119
+ weight: 0.2581423726361908
120
+ - sources:
121
+ - layer_range: [16, 20]
122
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
123
+ parameters:
124
+ density: 1.0
125
+ weight: 0.2116035543552448
126
+ - layer_range: [16, 20]
127
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
128
+ parameters:
129
+ density: 1.0
130
+ weight: 0.22654226422958418
131
+ - layer_range: [16, 20]
132
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
133
+ parameters:
134
+ density: 0.8925914810507647
135
+ weight: 0.42243766315440867
136
+ - sources:
137
+ - layer_range: [20, 24]
138
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
139
+ parameters:
140
+ density: 0.7697608089825734
141
+ weight: 0.1535118632140203
142
+ - layer_range: [20, 24]
143
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
144
+ parameters:
145
+ density: 0.9886758076773643
146
+ weight: 0.3305040603868546
147
+ - layer_range: [20, 24]
148
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
149
+ parameters:
150
+ density: 1.0
151
+ weight: 0.40670083428654535
152
+ - sources:
153
+ - layer_range: [24, 28]
154
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
155
+ parameters:
156
+ density: 1.0
157
+ weight: 0.4542810478500622
158
+ - layer_range: [24, 28]
159
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
160
+ parameters:
161
+ density: 0.8330662483310117
162
+ weight: 0.2587495367324508
163
+ - layer_range: [24, 28]
164
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
165
+ parameters:
166
+ density: 0.9845313983551542
167
+ weight: 0.40378452705975915
168
+ - sources:
169
+ - layer_range: [28, 32]
170
+ model: ./mergekit/input_models/Llama-3-LewdPlay-8B-e3_2981937066
171
+ parameters:
172
+ density: 1.0
173
+ weight: 0.2951962192288415
174
+ - layer_range: [28, 32]
175
+ model: ./mergekit/input_models/Llama-3-Unholy-8B-e4_1440388923
176
+ parameters:
177
+ density: 0.960315594933433
178
+ weight: 0.13142971773782525
179
+ - layer_range: [28, 32]
180
+ model: ./mergekit/input_models/Roleplay-Llama-3-8B_213413727
181
+ parameters:
182
+ density: 1.0
183
+ weight: 0.30838472094518804
184
+ ```
185
+
186
+ ## Support
187
+
188
+ If you want to support me, you can [here](https://ko-fi.com/undiai).