Triangle104
commited on
Commit
•
c6775cb
1
Parent(s):
f509c17
Update README.md
Browse files
README.md
CHANGED
@@ -15,6 +15,112 @@ base_model: Nohobby/MS-Schisandra-22B-v0.3
|
|
15 |
This model was converted to GGUF format from [`Nohobby/MS-Schisandra-22B-v0.3`](https://huggingface.co/Nohobby/MS-Schisandra-22B-v0.3) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
16 |
Refer to the [original model card](https://huggingface.co/Nohobby/MS-Schisandra-22B-v0.3) for more details on the model.
|
17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
## Use with llama.cpp
|
19 |
Install llama.cpp through brew (works on Mac and Linux)
|
20 |
|
|
|
15 |
This model was converted to GGUF format from [`Nohobby/MS-Schisandra-22B-v0.3`](https://huggingface.co/Nohobby/MS-Schisandra-22B-v0.3) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
16 |
Refer to the [original model card](https://huggingface.co/Nohobby/MS-Schisandra-22B-v0.3) for more details on the model.
|
17 |
|
18 |
+
Merge Details
|
19 |
+
-
|
20 |
+
Merging steps
|
21 |
+
|
22 |
+
Karasik-v0.3
|
23 |
+
|
24 |
+
models:
|
25 |
+
- model: Mistral-Small-22B-ArliAI-RPMax-v1.1
|
26 |
+
parameters:
|
27 |
+
weight: [0.2, 0.3, 0.2, 0.3, 0.2]
|
28 |
+
density: [0.45, 0.55, 0.45, 0.55, 0.45]
|
29 |
+
- model: Mistral-Small-NovusKyver
|
30 |
+
parameters:
|
31 |
+
weight: [0.01768, -0.01675, 0.01285, -0.01696, 0.01421]
|
32 |
+
density: [0.6, 0.4, 0.5, 0.4, 0.6]
|
33 |
+
- model: MiS-Firefly-v0.2-22B
|
34 |
+
parameters:
|
35 |
+
weight: [0.208, 0.139, 0.139, 0.139, 0.208]
|
36 |
+
density: [0.7]
|
37 |
+
- model: magnum-v4-22b
|
38 |
+
parameters:
|
39 |
+
weight: [0.33]
|
40 |
+
density: [0.45, 0.55, 0.45, 0.55, 0.45]
|
41 |
+
merge_method: della_linear
|
42 |
+
base_model: Mistral-Small-22B-ArliAI-RPMax-v1.1
|
43 |
+
parameters:
|
44 |
+
epsilon: 0.05
|
45 |
+
lambda: 1.05
|
46 |
+
int8_mask: true
|
47 |
+
rescale: true
|
48 |
+
normalize: false
|
49 |
+
dtype: bfloat16
|
50 |
+
tokenizer_source: base
|
51 |
+
|
52 |
+
SchisandraVA3
|
53 |
+
|
54 |
+
(Config taken from here)
|
55 |
+
|
56 |
+
merge_method: della_linear
|
57 |
+
dtype: bfloat16
|
58 |
+
parameters:
|
59 |
+
normalize: true
|
60 |
+
int8_mask: true
|
61 |
+
tokenizer_source: base
|
62 |
+
base_model: Cydonia-22B-v1.3
|
63 |
+
models:
|
64 |
+
- model: Karasik03
|
65 |
+
parameters:
|
66 |
+
density: 0.55
|
67 |
+
weight: 1
|
68 |
+
- model: Pantheon-RP-Pure-1.6.2-22b-Small
|
69 |
+
parameters:
|
70 |
+
density: 0.55
|
71 |
+
weight: 1
|
72 |
+
- model: ChatWaifu_v2.0_22B
|
73 |
+
parameters:
|
74 |
+
density: 0.55
|
75 |
+
weight: 1
|
76 |
+
- model: MS-Meadowlark-Alt-22B
|
77 |
+
parameters:
|
78 |
+
density: 0.55
|
79 |
+
weight: 1
|
80 |
+
- model: SorcererLM-22B
|
81 |
+
parameters:
|
82 |
+
density: 0.55
|
83 |
+
weight: 1
|
84 |
+
|
85 |
+
Schisandra-v0.3
|
86 |
+
|
87 |
+
dtype: bfloat16
|
88 |
+
tokenizer_source: base
|
89 |
+
merge_method: della_linear
|
90 |
+
parameters:
|
91 |
+
density: 0.5
|
92 |
+
base_model: SchisandraVA3
|
93 |
+
models:
|
94 |
+
- model: unsloth/Mistral-Small-Instruct-2409
|
95 |
+
parameters:
|
96 |
+
weight:
|
97 |
+
- filter: v_proj
|
98 |
+
value: [0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0]
|
99 |
+
- filter: o_proj
|
100 |
+
value: [1, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1]
|
101 |
+
- filter: up_proj
|
102 |
+
value: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
103 |
+
- filter: gate_proj
|
104 |
+
value: [0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0]
|
105 |
+
- filter: down_proj
|
106 |
+
value: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
107 |
+
- value: 0
|
108 |
+
- model: SchisandraVA3
|
109 |
+
parameters:
|
110 |
+
weight:
|
111 |
+
- filter: v_proj
|
112 |
+
value: [1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1]
|
113 |
+
- filter: o_proj
|
114 |
+
value: [0, 1, 0, 1, 1, 1, 1, 1, 0, 0, 0]
|
115 |
+
- filter: up_proj
|
116 |
+
value: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
117 |
+
- filter: gate_proj
|
118 |
+
value: [1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1]
|
119 |
+
- filter: down_proj
|
120 |
+
value: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
121 |
+
- value: 1
|
122 |
+
|
123 |
+
---
|
124 |
## Use with llama.cpp
|
125 |
Install llama.cpp through brew (works on Mac and Linux)
|
126 |
|