File size: 1,432 Bytes
b195331
 
 
 
 
ae6673e
b195331
 
 
 
 
a3b1fcc
 
b195331
 
 
ae6673e
b195331
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
---
license: apache-2.0
tags:
- merge
- mergekit
- bardsai/jaskier-7b-dpo-v5.6
- mlabonne/NeuralDaredevil-7B
- Gille/StrangeMerges_21-7B-slerp
- CultriX/NeuralTrix-7B-dpo
---

<img src="pikus-pikantny.png" alt="Pikus pikantny logo" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>

# pikus-pikantny-7B-dare

pikus-pikantny-7B-dare is a DARE merge of the following models using [mergekit](https://github.com/cg123/mergekit):
* [bardsai/jaskier-7b-dpo-v5.6](https://huggingface.co/bardsai/jaskier-7b-dpo-v5.6)
* [mlabonne/NeuralDaredevil-7B](https://huggingface.co/mlabonne/NeuralDaredevil-7B)
* [Gille/StrangeMerges_21-7B-slerp](https://huggingface.co/Gille/StrangeMerges_21-7B-slerp)
* [CultriX/NeuralTrix-7B-dpo](https://huggingface.co/CultriX/NeuralTrix-7B-dpo)

See the paper [Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch](https://arxiv.org/abs/2311.03099) for more on the method.

## 🧩 Configuration

```yaml
models:
  - model: bardsai/jaskier-7b-dpo-v5.6
  - model: mlabonne/NeuralDaredevil-7B
    parameters:
      density: 0.53
      weight: 0.3
  - model: Gille/StrangeMerges_21-7B-slerp
    parameters:
      density: 0.53
      weight: 0.4
  - model: CultriX/NeuralTrix-7B-dpo
    parameters:
      density: 0.53
      weight: 0.3
merge_method: dare_ties
base_model: bardsai/jaskier-7b-dpo-v5.6
parameters:
  int8_mask: true
dtype: bfloat16
```