limin(gate)
commited on
Update README.md
Browse files
README.md
CHANGED
|
@@ -17,6 +17,115 @@ Liph.43 is a merge of the following models using [mergekit](https://github.com/c
|
|
| 17 |
## 🧩 Configuration
|
| 18 |
|
| 19 |
```yaml
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
slices:
|
| 21 |
- sources:
|
| 22 |
- model: liminerity/merge5
|
|
|
|
| 17 |
## 🧩 Configuration
|
| 18 |
|
| 19 |
```yaml
|
| 20 |
+
|
| 21 |
+
slices:
|
| 22 |
+
- sources:
|
| 23 |
+
- model: liminerity/phigment6-slerp
|
| 24 |
+
layer_range: [0, 32]
|
| 25 |
+
- model: liminerity/phigment6-slerp
|
| 26 |
+
layer_range: [0, 32]
|
| 27 |
+
merge_method: slerp
|
| 28 |
+
base_model: liminerity/phigment6-slerp
|
| 29 |
+
parameters:
|
| 30 |
+
t:
|
| 31 |
+
- filter: self_attn
|
| 32 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 33 |
+
- filter: mlp
|
| 34 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 35 |
+
- value: 0.5
|
| 36 |
+
dtype: bfloat16
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
slices:
|
| 40 |
+
- sources:
|
| 41 |
+
- model: rhysjones/phi-2-orange-v2
|
| 42 |
+
layer_range: [0, 32]
|
| 43 |
+
- model: liminerity/merge
|
| 44 |
+
layer_range: [0, 32]
|
| 45 |
+
merge_method: slerp
|
| 46 |
+
base_model: rhysjones/phi-2-orange-v2
|
| 47 |
+
parameters:
|
| 48 |
+
t:
|
| 49 |
+
- filter: self_attn
|
| 50 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 51 |
+
- filter: mlp
|
| 52 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 53 |
+
- value: 0.5
|
| 54 |
+
dtype: bfloat16
|
| 55 |
+
|
| 56 |
+
|
| 57 |
+
slices:
|
| 58 |
+
- sources:
|
| 59 |
+
- model: liminerity/merge1
|
| 60 |
+
layer_range: [0, 32]
|
| 61 |
+
- model: liminerity/phigment6-slerp
|
| 62 |
+
layer_range: [0, 32]
|
| 63 |
+
merge_method: slerp
|
| 64 |
+
base_model: liminerity/merge1
|
| 65 |
+
parameters:
|
| 66 |
+
t:
|
| 67 |
+
- filter: self_attn
|
| 68 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 69 |
+
- filter: mlp
|
| 70 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 71 |
+
- value: 0.5
|
| 72 |
+
dtype: bfloat16
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
slices:
|
| 76 |
+
- sources:
|
| 77 |
+
- model: liminerity/Liph.42
|
| 78 |
+
layer_range: [0, 32]
|
| 79 |
+
- model: liminerity/merge2
|
| 80 |
+
layer_range: [0, 32]
|
| 81 |
+
merge_method: slerp
|
| 82 |
+
base_model: liminerity/Liph.42
|
| 83 |
+
parameters:
|
| 84 |
+
t:
|
| 85 |
+
- filter: self_attn
|
| 86 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 87 |
+
- filter: mlp
|
| 88 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 89 |
+
- value: 0.5
|
| 90 |
+
dtype: bfloat16
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
slices:
|
| 94 |
+
- sources:
|
| 95 |
+
- model: liminerity/merge3
|
| 96 |
+
layer_range: [0, 32]
|
| 97 |
+
- model: rhysjones/phi-2-orange-v2
|
| 98 |
+
layer_range: [0, 32]
|
| 99 |
+
merge_method: slerp
|
| 100 |
+
base_model: liminerity/merge3
|
| 101 |
+
parameters:
|
| 102 |
+
t:
|
| 103 |
+
- filter: self_attn
|
| 104 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 105 |
+
- filter: mlp
|
| 106 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 107 |
+
- value: 0.5
|
| 108 |
+
dtype: bfloat16
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
slices:
|
| 112 |
+
- sources:
|
| 113 |
+
- model: liminerity/Phigments12
|
| 114 |
+
layer_range: [0, 32]
|
| 115 |
+
- model: liminerity/merge4
|
| 116 |
+
layer_range: [0, 32]
|
| 117 |
+
merge_method: slerp
|
| 118 |
+
base_model: liminerity/Phigments12
|
| 119 |
+
parameters:
|
| 120 |
+
t:
|
| 121 |
+
- filter: self_attn
|
| 122 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 123 |
+
- filter: mlp
|
| 124 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 125 |
+
- value: 0.5
|
| 126 |
+
dtype: bfloat16
|
| 127 |
+
|
| 128 |
+
|
| 129 |
slices:
|
| 130 |
- sources:
|
| 131 |
- model: liminerity/merge5
|