Update README.md
Browse files
README.md
CHANGED
|
@@ -31,6 +31,25 @@ ECE-TW3-JRGL-V1 is a merge of the following models using **[mergekit](https://g
|
|
| 31 |
* [ShinojiResearch/Senku-70B-Full](https://huggingface.co/ShinojiResearch/Senku-70B-Full)
|
| 32 |
* [152334H/miqu-1-70b-sf](https://huggingface.co/152334H/miqu-1-70b-sf)
|
| 33 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
## Results
|
| 35 |
|
| 36 |
- ECE-TW3-JRGL-v1 scores 83.07 on [EQ-Bench V2](https://eqbench.com/index.html)
|
|
|
|
| 31 |
* [ShinojiResearch/Senku-70B-Full](https://huggingface.co/ShinojiResearch/Senku-70B-Full)
|
| 32 |
* [152334H/miqu-1-70b-sf](https://huggingface.co/152334H/miqu-1-70b-sf)
|
| 33 |
|
| 34 |
+
```yaml
|
| 35 |
+
slices:
|
| 36 |
+
- sources:
|
| 37 |
+
- model: ShinojiResearch/Senku-70B-Full
|
| 38 |
+
layer_range: [0, 80]
|
| 39 |
+
- model: 152334H/miqu-1-70b-sf
|
| 40 |
+
layer_range: [0, 80]
|
| 41 |
+
merge_method: slerp
|
| 42 |
+
base_model: 152334H/miqu-1-70b-sf
|
| 43 |
+
parameters:
|
| 44 |
+
t:
|
| 45 |
+
- filter: self_attn
|
| 46 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
| 47 |
+
- filter: mlp
|
| 48 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
| 49 |
+
- value: 0.5
|
| 50 |
+
dtype: float16
|
| 51 |
+
```
|
| 52 |
+
|
| 53 |
## Results
|
| 54 |
|
| 55 |
- ECE-TW3-JRGL-v1 scores 83.07 on [EQ-Bench V2](https://eqbench.com/index.html)
|