Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
Paper
•
2203.05482
•
Published
•
7
For experimental purposes only. Probably not that good.
This is a merge of pre-trained language models created using mergekit.
This model was merged using the linear merge method.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: monology/mixtral-expert0
- model: monology/mixtral-expert1
- model: monology/mixtral-expert2
- model: monology/mixtral-expert3
- model: monology/mixtral-expert4
- model: monology/mixtral-expert5
- model: monology/mixtral-expert6
- model: monology/mixtral-expert7
parameters:
weight: 1.0
merge_method: linear
dtype: float16