Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
Paper
•
2311.03099
•
Published
•
30
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DARE TIES merge method using bunnycore/Gemma2-9B-TitanFusion as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: bunnycore/Gemma2-9B-TitanFusion
parameters:
weight: 0.5
density: 0.5
- model: flammenai/Mahou-1.3-gemma2-9B
parameters:
weight: 0.5
density: 0.5
- model: anthracite-org/magnum-v3-9b-customgemma2
parameters:
weight: 0.5
density: 0.5
merge_method: dare_ties
base_model: bunnycore/Gemma2-9B-TitanFusion
dtype: bfloat16