mav23 commited on
Commit
19dc423
1 Parent(s): bcaa56d

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. .gitattributes +1 -0
  2. README.md +159 -0
  3. ms-schisandra-22b-v0.1.Q4_0.gguf +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ ms-schisandra-22b-v0.1.Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,159 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - unsloth/Mistral-Small-Instruct-2409
4
+ - TheDrummer/Cydonia-22B-v1.2
5
+ - Gryphe/Pantheon-RP-Pure-1.6.2-22b-Small
6
+ - anthracite-org/magnum-v4-22b
7
+ - ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1
8
+ - spow12/ChatWaifu_v2.0_22B
9
+ - rAIfle/Acolyte-22B
10
+ - Envoid/Mistral-Small-NovusKyver
11
+ - InferenceIllusionist/SorcererLM-22B
12
+ library_name: transformers
13
+ tags:
14
+ - mergekit
15
+ - merge
16
+ license: other
17
+ language:
18
+ - en
19
+ ---
20
+ ***
21
+ ## Schisandra
22
+
23
+ Many thanks to the authors of the models used!
24
+
25
+ [RPMax v1.1](https://huggingface.co/ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1) | [Pantheon-RP](https://huggingface.co/Gryphe/Pantheon-RP-Pure-1.6.2-22b-Small) | [Cydonia v1.2](https://huggingface.co/TheDrummer/Cydonia-22B-v1.2) | [Magnum V4](https://huggingface.co/anthracite-org/magnum-v4-22b) | [ChatWaifu v2.0](https://huggingface.co/spow12/ChatWaifu_v2.0_22B) | [SorcererLM](https://huggingface.co/InferenceIllusionist/SorcererLM-22B) | [Acolyte](https://huggingface.co/rAIfle/Acolyte-22B) | [NovusKyver](https://huggingface.co/Envoid/Mistral-Small-NovusKyver)
26
+ ***
27
+
28
+ The new version writes better and doesn't mispronounce names anymore!
29
+
30
+ https://huggingface.co/Nohobby/MS-Schisandra-22B-v0.2
31
+
32
+ ***
33
+
34
+ ### Overview
35
+
36
+ Main uses: RP, Storywriting
37
+
38
+ Merge of 8 Mistral Small finetunes in total, which were then merged back into the original model to make it less stupid. Worked somehow? Definitely smarter than my previous MS merge and maybe some finetunes. Seems to really adhere to the writing style of the previous output, so you'll need either a good character card or an existing chat for a better replies.
39
+
40
+ ***
41
+
42
+ ### Quants
43
+
44
+ [Static](https://huggingface.co/mradermacher/MS-Schisandra-22B-vB-GGUF)
45
+
46
+ [Imatrix](https://huggingface.co/mradermacher/MS-Schisandra-22B-vB-i1-GGUF)
47
+
48
+ ***
49
+
50
+ ### Settings
51
+
52
+ Prompt format: Mistral-V3 Tekken
53
+
54
+ Samplers: [These](https://qu.ax/OusTx.json) or [These](https://huggingface.co/ToastyPigeon/ST-Presets-Mistral-Small/resolve/main/ST-sampling-preset-Mistral-Small.json?download=true)
55
+
56
+ ***
57
+
58
+ ## Merge Details
59
+ ### Merging steps
60
+
61
+ ## QCmix
62
+
63
+ ```yaml
64
+ base_model: InferenceIllusionist/SorcererLM-22B
65
+ parameters:
66
+ int8_mask: true
67
+ rescale: true
68
+ normalize: false
69
+ dtype: bfloat16
70
+ tokenizer_source: base
71
+ merge_method: della
72
+ models:
73
+ - model: Envoid/Mistral-Small-NovusKyver
74
+ parameters:
75
+ density: [0.35, 0.65, 0.5, 0.65, 0.35]
76
+ epsilon: [0.1, 0.1, 0.25, 0.1, 0.1]
77
+ lambda: 0.85
78
+ weight: [-0.01891, 0.01554, -0.01325, 0.01791, -0.01458]
79
+ - model: rAIfle/Acolyte-22B
80
+ parameters:
81
+ density: [0.6, 0.4, 0.5, 0.4, 0.6]
82
+ epsilon: [0.15, 0.15, 0.25, 0.15, 0.15]
83
+ lambda: 0.85
84
+ weight: [0.01768, -0.01675, 0.01285, -0.01696, 0.01421]
85
+ ```
86
+
87
+ ## Schisandra-vA
88
+
89
+ ```yaml
90
+ merge_method: della_linear
91
+ dtype: bfloat16
92
+ parameters:
93
+ normalize: true
94
+ int8_mask: true
95
+ tokenizer_source: union
96
+ base_model: TheDrummer/Cydonia-22B-v1.2
97
+ models:
98
+ - model: ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1
99
+ parameters:
100
+ density: 0.55
101
+ weight: 1
102
+ - model: Gryphe/Pantheon-RP-Pure-1.6.2-22b-Small
103
+ parameters:
104
+ density: 0.55
105
+ weight: 1
106
+ - model: spow12/ChatWaifu_v2.0_22B
107
+ parameters:
108
+ density: 0.55
109
+ weight: 1
110
+ - model: anthracite-org/magnum-v4-22b
111
+ parameters:
112
+ density: 0.55
113
+ weight: 1
114
+ - model: QCmix
115
+ parameters:
116
+ density: 0.55
117
+ weight: 1
118
+ ```
119
+
120
+ ## Schisandra
121
+
122
+ ```yaml
123
+ dtype: bfloat16
124
+ tokenizer_source: base
125
+ merge_method: della_linear
126
+ parameters:
127
+ density: 0.5
128
+ base_model: Schisandra
129
+ models:
130
+ - model: unsloth/Mistral-Small-Instruct-2409
131
+ parameters:
132
+ weight:
133
+ - filter: v_proj
134
+ value: [0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0]
135
+ - filter: o_proj
136
+ value: [1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1]
137
+ - filter: up_proj
138
+ value: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
139
+ - filter: gate_proj
140
+ value: [0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0]
141
+ - filter: down_proj
142
+ value: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
143
+ - value: 0
144
+ - model: Schisandra
145
+ parameters:
146
+ weight:
147
+ - filter: v_proj
148
+ value: [1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1]
149
+ - filter: o_proj
150
+ value: [0, 1, 0, 1, 1, 1, 1, 1, 0, 0, 0]
151
+ - filter: up_proj
152
+ value: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
153
+ - filter: gate_proj
154
+ value: [1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1]
155
+ - filter: down_proj
156
+ value: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
157
+ - value: 1
158
+
159
+ ```
ms-schisandra-22b-v0.1.Q4_0.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d864585d6e3178995b26a5a6337f9b8e8d1e774b7edd76c131b129ddc60ebc08
3
+ size 12569164512