Update README.md
Browse files
README.md
CHANGED
@@ -18,37 +18,11 @@ The new MODEL STOCK merge method was used, see below for more information!
|
|
18 |
|
19 |
Feedback on this model is greatly appreciated! I hope this new merge method will be able to fill some hole Miqu have.
|
20 |
|
21 |
-
##
|
22 |
-
### Merge Method
|
23 |
-
|
24 |
-
This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [152334H/miqu-1-70b-sf](https://huggingface.co/152334H/miqu-1-70b-sf) as a base.
|
25 |
-
|
26 |
-
### Models Merged
|
27 |
-
|
28 |
-
The following models were included in the merge:
|
29 |
-
* [migtissera/Tess-70B-v1.6](https://huggingface.co/migtissera/Tess-70B-v1.6)
|
30 |
-
* [NeverSleep/MiquMaid-v2-70B](https://huggingface.co/NeverSleep/MiquMaid-v2-70B)
|
31 |
-
* [sophosympatheia/Midnight-Miqu-70B-v1.0](https://huggingface.co/sophosympatheia/Midnight-Miqu-70B-v1.0)
|
32 |
-
|
33 |
-
### Configuration
|
34 |
-
|
35 |
-
The following YAML configuration was used to produce this model:
|
36 |
-
|
37 |
-
```yaml
|
38 |
-
models:
|
39 |
-
- model: NeverSleep/MiquMaid-v2-70B
|
40 |
-
- model: sophosympatheia/Midnight-Miqu-70B-v1.0
|
41 |
-
- model: migtissera/Tess-70B-v1.6
|
42 |
-
- model: 152334H/miqu-1-70b-sf
|
43 |
-
merge_method: model_stock
|
44 |
-
base_model: 152334H/miqu-1-70b-sf
|
45 |
-
dtype: bfloat16
|
46 |
-
```
|
47 |
-
### Prompt format
|
48 |
|
49 |
Since it was made with model using different prompt format, the following should work.
|
50 |
|
51 |
-
|
52 |
|
53 |
```
|
54 |
### Instruction:
|
@@ -61,13 +35,13 @@ Since it was made with model using different prompt format, the following should
|
|
61 |
{output}
|
62 |
```
|
63 |
|
64 |
-
|
65 |
|
66 |
```
|
67 |
[INST] {prompt} [/INST]
|
68 |
```
|
69 |
|
70 |
-
|
71 |
|
72 |
```
|
73 |
SYSTEM: <ANY SYSTEM CONTEXT>
|
@@ -75,6 +49,33 @@ USER:
|
|
75 |
ASSISTANT:
|
76 |
```
|
77 |
|
78 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
79 |
|
80 |
If you want to support me, you can [here](https://ko-fi.com/undiai).
|
|
|
18 |
|
19 |
Feedback on this model is greatly appreciated! I hope this new merge method will be able to fill some hole Miqu have.
|
20 |
|
21 |
+
## Prompt format
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
Since it was made with model using different prompt format, the following should work.
|
24 |
|
25 |
+
### Alpaca
|
26 |
|
27 |
```
|
28 |
### Instruction:
|
|
|
35 |
{output}
|
36 |
```
|
37 |
|
38 |
+
### Mistral
|
39 |
|
40 |
```
|
41 |
[INST] {prompt} [/INST]
|
42 |
```
|
43 |
|
44 |
+
### Vicuna
|
45 |
|
46 |
```
|
47 |
SYSTEM: <ANY SYSTEM CONTEXT>
|
|
|
49 |
ASSISTANT:
|
50 |
```
|
51 |
|
52 |
+
## Merge Details
|
53 |
+
### Merge Method
|
54 |
+
|
55 |
+
This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [152334H/miqu-1-70b-sf](https://huggingface.co/152334H/miqu-1-70b-sf) as a base.
|
56 |
+
|
57 |
+
### Models Merged
|
58 |
+
|
59 |
+
The following models were included in the merge:
|
60 |
+
* [migtissera/Tess-70B-v1.6](https://huggingface.co/migtissera/Tess-70B-v1.6)
|
61 |
+
* [NeverSleep/MiquMaid-v2-70B](https://huggingface.co/NeverSleep/MiquMaid-v2-70B)
|
62 |
+
* [sophosympatheia/Midnight-Miqu-70B-v1.0](https://huggingface.co/sophosympatheia/Midnight-Miqu-70B-v1.0)
|
63 |
+
|
64 |
+
### Configuration
|
65 |
+
|
66 |
+
The following YAML configuration was used to produce this model:
|
67 |
+
|
68 |
+
```yaml
|
69 |
+
models:
|
70 |
+
- model: NeverSleep/MiquMaid-v2-70B
|
71 |
+
- model: sophosympatheia/Midnight-Miqu-70B-v1.0
|
72 |
+
- model: migtissera/Tess-70B-v1.6
|
73 |
+
- model: 152334H/miqu-1-70b-sf
|
74 |
+
merge_method: model_stock
|
75 |
+
base_model: 152334H/miqu-1-70b-sf
|
76 |
+
dtype: bfloat16
|
77 |
+
```
|
78 |
+
|
79 |
+
## Support
|
80 |
|
81 |
If you want to support me, you can [here](https://ko-fi.com/undiai).
|