Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
Paper
• 2203.05482 • Published
• 7
Previous merge was experimental, and I wasn't happy about it. So I did another one, that seems better in creative writing and role-play.
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: Glowing-Forest-1.5
parameters:
weight: 0.6
- model: Rei-12B
parameters:
weight: 0.4
merge_method: linear
dtype: float16