35 lines
692 B
Markdown
35 lines
692 B
Markdown
---
|
|
license: mit
|
|
language:
|
|
- en
|
|
pipeline_tag: text-generation
|
|
inference: false
|
|
tags:
|
|
- dare
|
|
- super mario merge
|
|
- pytorch
|
|
- mistral
|
|
- merge
|
|
---
|
|
|
|
|
|
# mistral 7b mega merge
|
|
|
|
This model was merged using [https://github.com/martyn/safetensors-merge-supermario](https://github.com/martyn/safetensors-merge-supermario) with hyperparams `p=0.12` and `lambda=2.1`.
|
|
|
|
The first entry is the base model:
|
|
|
|
```
|
|
mistralai/Mistral-7B-Instruct-v0.2
|
|
uukuguy/speechless-code-mistral-7b-v1.0
|
|
AIDC-ai-business/Marcoroni-7B-v3
|
|
Weyaxi/Seraph-7B
|
|
rwitz/dec10
|
|
Intel/neural-chat-7b-v3-3
|
|
rwitz/go-bruins-v2
|
|
```
|
|
|
|
To merge your own model:
|
|
```
|
|
python hf_merge.py to_merge_7b.txt mistral_7b_0.2_merge -p 0.12 -lambda 2.1
|
|
``` |