chargoddard
commited on
Commit
•
7a08a96
1
Parent(s):
74a1495
Update README.md
Browse files
README.md
CHANGED
@@ -4,4 +4,24 @@ tags:
|
|
4 |
- llama2
|
5 |
---
|
6 |
|
7 |
-
Experimental multi-lingual model using a new merge technique.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
- llama2
|
5 |
---
|
6 |
|
7 |
+
Experimental multi-lingual model using a new merge technique.
|
8 |
+
|
9 |
+
Mergekit configuration (experimental branch):
|
10 |
+
```yaml
|
11 |
+
models:
|
12 |
+
- model: clibrain/Llama-2-13b-ft-instruct-es
|
13 |
+
- model: LeoLM/leo-hessianai-13b
|
14 |
+
- model: daekeun-ml/Llama-2-ko-DPO-13B
|
15 |
+
- model: pleisto/yuren-13b-chatml
|
16 |
+
- model: bofenghuang/vigogne-2-13b-instruct
|
17 |
+
- model: OpenBuddy/openbuddy-llama2-13b-v8.1-fp16
|
18 |
+
merge_method: dare_ties
|
19 |
+
base_model: TheBloke/Llama-2-13B-fp16
|
20 |
+
dtype: float16
|
21 |
+
parameters:
|
22 |
+
density: 0.3
|
23 |
+
weight: 1.0
|
24 |
+
normalize: true
|
25 |
+
int8_mask: true
|
26 |
+
tokenizer_source: base
|
27 |
+
```
|