Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pipeline_tag: text-generation
|
| 3 |
+
tags:
|
| 4 |
+
- mistral
|
| 5 |
+
- merge
|
| 6 |
+
license: cc-by-4.0
|
| 7 |
+
---
|
| 8 |
+
# Model Card for Moko-DARE
|
| 9 |
+
<!-- Provide a quick summary of what the model is/does. -->
|
| 10 |
+
Part of a merge method experiment.
|
| 11 |
+
|
| 12 |
+
.yaml file for mergekit
|
| 13 |
+
```.yaml:
|
| 14 |
+
models:
|
| 15 |
+
- model: Open-Orca/Mistral-7B-OpenOrca
|
| 16 |
+
parameters:
|
| 17 |
+
density: [1, 0.7, 0.1] # density gradient
|
| 18 |
+
weight: 1.0
|
| 19 |
+
- model: akjindal53244/Mistral-7B-v0.1-Open-Platypus
|
| 20 |
+
parameters:
|
| 21 |
+
density: 0.5
|
| 22 |
+
weight: [0, 0.3, 0.7, 1] # weight gradient
|
| 23 |
+
- model: WizardLM/WizardMath-7B-V1.1
|
| 24 |
+
parameters:
|
| 25 |
+
density: 0.33
|
| 26 |
+
weight:
|
| 27 |
+
- filter: mlp
|
| 28 |
+
value: 0.5
|
| 29 |
+
- value: 0
|
| 30 |
+
merge_method: dare_ties
|
| 31 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 32 |
+
parameters:
|
| 33 |
+
normalize: true
|
| 34 |
+
int8_mask: true
|
| 35 |
+
dtype: float16
|
| 36 |
+
```
|